This file is raw output from pdftotext and may not be ideal for distribution. If you are a maintainer for Hackipedia, please sit down when you have time and clean this text version up. Source PDF: /mnt/main/jmc-storage/docs/DVB/DVB A120 DVB-SH Implementation Guidelines Issue 2 (June 2010).pdf Like all conversions the text below should be fully readable as UTF-8 unicode text. --------------------------------------------------------------- ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! Digital Video Broadcasting (DVB); DVB-SH Implementation Guidelines Issue 2 DVB Document A120 June 2010 3 ETSI TS 102 584 V1.1.2 (2010-XY) Contents Intellectual Property Rights ..............................................................................................................................14 Foreword ..........................................................................................................................................................14 Introduction ......................................................................................................................................................14 1 Scope ......................................................................................................................................................16 2 References ..............................................................................................................................................16 2.1 Normative references....................................................................................................................................... 17 2.2 Informative references ..................................................................................................................................... 18 3 Definitions and abbreviations.................................................................................................................21 3.1 Definitions ....................................................................................................................................................... 21 3.2 Abbreviations................................................................................................................................................... 22 4 System outline........................................................................................................................................26 4.1 DVB-SH system capability outline.................................................................................................................. 26 4.1.1 System overview ........................................................................................................................................ 27 4.1.2 Frequency planning aspects ....................................................................................................................... 28 4.1.3 Other considerations for spectrum efficiency .......................................................................................................... 31 4.1.4 Frequency bands and their regulatory aspects............................................................................................ 32 4.1.4.1 Frequency Allocations and coordination procedures............................................................................ 32 4.1.4.1.1 L band ............................................................................................................................................. 33 4.1.4.1.2 2 GHz S band (1 980 MHz to 2 010 MHz/2 170 MHz to 2 200 MHz)........................................... 34 4.1.4.1.3 Sound Broadcasting Satellite (SDARS) S band (2 310 MHz to 2 360 MHz) ................................. 35 4.1.4.1.4 2,5 GHz S band (2 500 MHz to 2 690 MHz) .................................................................................. 35 4.1.4.2 Regulatory aspects................................................................................................................................ 35 4.1.4.2.1 Europe............................................................................................................................................. 35 4.1.4.2.2 North America (all bands) .............................................................................................................. 37 4.2 Main issues addressed by DVB-SH................................................................................................................. 39 4.2.1 Differences in propagation characteristics between satellite and terrestrial channels ................................ 39 4.2.2 LMS channels (in the L and S bands) ........................................................................................................ 39 4.2.2.1 Empirical Models ................................................................................................................................. 40 4.2.2.2 Statistical models (for narrow-band signals) ........................................................................................ 40 4.2.2.2.1 Single-state models ......................................................................................................................... 40 4.2.2.2.2 Two-state (Lutz) model................................................................................................................... 41 4.2.2.2.3 Three-state (Fontan) Model ............................................................................................................ 41 4.2.2.2.4 Quasi-static Channel Model............................................................................................................ 42 4.2.3 Terrestrial Channels ................................................................................................................................... 43 4.2.4 Satellite specific issues............................................................................................................................... 44 4.2.4.1 Satellite Payload architectures.............................................................................................................. 44 4.2.4.2 Nonlinear Payload Distortion Effects................................................................................................... 44 4.2.4.3 Phase Noise Aspects............................................................................................................................. 45 4.2.4.4 Satellite induced Doppler Shift............................................................................................................. 45 4.2.5 DVB-SH Impairments Countermeasures ................................................................................................... 47 4.2.5.1 LMS Mobile Channel Impact Mitigation ............................................................................................. 47 4.2.5.1.1 Exploitation of Time and Space Diversity ...................................................................................... 47 4.2.5.1.2 Robust Demodulator Operations..................................................................................................... 47 4.2.5.2 Nonlinear Channel Impairments Countermeasures .............................................................................. 47 4.2.6 Receiver characteristics.............................................................................................................................. 48 4.2.6.1 Vehicular reception constraints ............................................................................................................ 49 4.2.6.2 Handheld reception constraints (pedestrian)......................................................................................... 49 4.2.7 Terminal High-level architecture ............................................................................................................... 50 4.2.8 Compatibility with Existing DVB-H System Features............................................................................... 50 4.2.8.1 Battery-powered receivers.................................................................................................................... 50 4.2.8.1.1 Time slicing .................................................................................................................................... 50 4.2.8.1.2 Low-power Microelectronics.......................................................................................................... 50 4.2.8.1.3 Antenna Diversity (for S-band)....................................................................................................... 50 ETSI 4 ETSI TS 102 584 V1.1.2 (2010-XY) 4.2.8.2 Service Switching time (Zapping time) ................................................................................................ 51 4.2.8.3 Support for Variable Bit Rate (VBR) and Statistical Multiplexing (Statmux) ..................................... 51 4.2.8.4 Multiple QoS support ........................................................................................................................... 51 4.2.8.4.1 Introduction..................................................................................................................................... 51 4.2.8.4.2 At physical layer ............................................................................................................................. 52 4.2.8.4.3 At link layer .................................................................................................................................... 52 4.2.8.4.4 At application layer......................................................................................................................... 52 4.2.8.4.5 Recommendations........................................................................................................................... 52 4.2.8.5 Service announcements in hybrid networks ......................................................................................... 53 4.2.8.5.1 Introduction..................................................................................................................................... 53 4.2.8.6 Handover issues.................................................................................................................................... 54 4.2.9 Support of Other Frequency Bands ............................................................................................................ 54 5 System Configurations and possible Deployment..................................................................................55 5.1 Definitions and concepts.................................................................................................................................. 55 5.2 Configurations options definitions................................................................................................................... 55 5.2.1 Radio configuration.................................................................................................................................... 55 5.2.2 Frequency configuration ............................................................................................................................ 56 5.2.3 Topological configuration .......................................................................................................................... 56 5.2.4 Local content insertion configuration......................................................................................................... 56 5.2.5 CGC transmitters configuration ................................................................................................................. 58 5.2.6 Receiver configuration ............................................................................................................................... 58 5.2.7 Configuration naming ................................................................................................................................ 58 5.3 Examples of system deployment ..................................................................................................................... 59 5.3.1 SH-A deployment scenario ........................................................................................................................ 59 5.3.2 SH-B deployment scenario......................................................................................................................... 60 5.3.3 Less likely transitions................................................................................................................................. 60 5.3.4 Forbidden configurations ........................................................................................................................... 61 5.4 Other “advanced” hierarchical modulation use cases ...................................................................................... 61 5.4.1 Introduction................................................................................................................................................ 61 5.4.2 Use case 1: quality increase ....................................................................................................................... 61 5.4.3 Use case 2: capacity increase ..................................................................................................................... 62 5.4.4 Use case 3: coverage extension.................................................................................................................. 62 5.4.5 Use case 4: zapping time reduction............................................................................................................ 63 6 Elements at Link and Service layers ......................................................................................................63 6.1 Introduction...................................................................................................................................................... 63 6.2 MPE-IFEC ....................................................................................................................................................... 65 6.2.1 Framework description............................................................................................................................... 65 6.2.2 Usage in the context of the DVB-SH ......................................................................................................... 70 6.2.3 A practical example.................................................................................................................................... 73 6.2.3.1 Introduction .......................................................................................................................................... 73 6.2.3.2 Time_slice_fec_identifier..................................................................................................................... 73 6.2.3.3 MPE-IFEC parameter derivation.......................................................................................................... 74 6.2.3.4 ADST mapping function ...................................................................................................................... 75 6.2.3.5 ADT mapping....................................................................................................................................... 75 6.2.3.6 ADST view........................................................................................................................................... 75 6.2.3.7 ADT view ............................................................................................................................................. 76 6.2.3.8 FDT generation and code rate computation.......................................................................................... 78 6.2.3.9 IFEC burst generation........................................................................................................................... 80 6.2.3.10 MPE-IFEC section header .................................................................................................................... 82 6.2.3.11 Burst sending arrangement ................................................................................................................... 84 6.2.4 Parameters selection................................................................................................................................... 87 6.2.4.1 Introduction .......................................................................................................................................... 87 6.2.4.2 Recommendations on D........................................................................................................................ 88 6.2.4.3 Recommendation on B and S ............................................................................................................... 89 6.2.4.4 Selection of M and code rate ................................................................................................................ 89 6.2.4.5 Parameter overall selection logic.......................................................................................................... 91 6.2.4.6 Recommendations on signalling parameters ........................................................................................ 92 6.2.5 Simulated performance .............................................................................................................................. 92 6.2.5.1 ESR5(20) performance ......................................................................................................................... 92 6.2.5.1.1 LMS-ITS 50 kmph.......................................................................................................................... 92 ETSI 5 ETSI TS 102 584 V1.1.2 (2010-XY) 6.2.5.1.2 LMS-SU.......................................................................................................................................... 93 6.2.5.1.3 Recommended class 1 configuration .............................................................................................. 93 6.2.5.1.4 TU6 for hybrid frequency ............................................................................................................... 93 6.2.5.1.5 TU6 for non-hybrid frequency........................................................................................................ 94 6.2.5.2 Zapping time performance.................................................................................................................... 94 6.2.5.2.1 Introduction..................................................................................................................................... 94 6.2.5.2.2 Definitions ...................................................................................................................................... 94 6.2.5.2.3 Principles for zapping time analysis in class 1................................................................................ 95 6.2.5.2.4 Zapping performance with measured quality during the zapping period........................................ 98 6.2.5.2.4.1 Zapping performance in class 1................................................................................................. 98 6.2.5.2.4.1 Zapping performance in class 2............................................................................................... 102 6.2.6 Memory requirements .............................................................................................................................. 107 6.2.6.1 Introduction ........................................................................................................................................ 107 6.2.6.2 Memory sizing function...................................................................................................................... 107 6.2.6.3 Implementation aspects ...................................................................................................................... 108 6.2.6.3.1 Introduction................................................................................................................................... 108 6.2.6.3.2 Approach based on peak traffic .................................................................................................... 108 6.2.6.3.3 Approach based on average traffic................................................................................................ 108 6.2.6.3.4 Memory evaluation ....................................................................................................................... 109 6.2.7 MPE-IFEC usage scenarios...................................................................................................................... 111 6.2.7.1 Definition............................................................................................................................................ 111 6.2.7.2 Introduction ........................................................................................................................................ 111 6.2.7.2.1 Single MPE-IFEC source.............................................................................................................. 111 6.2.7.2.2 Multiple MPE-IFEC source .......................................................................................................... 112 6.2.7.2.3 Layered IFEC-encoding in SVC context ...................................................................................... 113 6.3 Time-Slicing .................................................................................................................................................. 116 6.3.1 Signalling ................................................................................................................................................. 116 6.3.2 Zapping time impact ................................................................................................................................ 117 6.3.3 Power saving impact ................................................................................................................................ 117 6.3.3.1 "Terrestrial" physical interleaver ........................................................................................................ 118 6.3.3.2 "Long" physical interleaver ................................................................................................................ 118 6.3.3.3 Summary ............................................................................................................................................ 119 6.3.4 VBR/statmux impact................................................................................................................................ 119 6.3.4.1 Interest of statistical multiplexing with AVC and SVC...................................................................... 119 6.3.4.2 Terrestrial (short) physical interleaver................................................................................................ 120 6.3.4.3 Uniform long interleaver .................................................................................................................... 121 6.3.5 Conclusion ............................................................................................................................................... 121 6.4 Mobility ......................................................................................................................................................... 122 7 Physical Layer elements.......................................................................................................................122 7.1 Overview to the physical layer elements ....................................................................................................... 122 7.2 Turbo code, time interleaver and SH framing................................................................................................ 124 7.2.1 Introduction.............................................................................................................................................. 124 7.2.2 Turbo code ............................................................................................................................................... 124 7.2.2.1 Introduction ........................................................................................................................................ 124 7.2.2.2 Overview of the key elements ............................................................................................................ 125 7.2.2.2.1 Top level description..................................................................................................................... 125 7.2.2.2.2 Parameters and numbers ............................................................................................................... 125 7.2.2.2.3 Examples for puncturing the [X Y0 Y1 X' Y0' Y1'] vector ..................................................... 125 7.2.2.2.4 Bit-wise interleaver....................................................................................................................... 126 7.2.2.3 Combining at the input of the turbo decoder ...................................................................................... 127 7.2.2.3.1 Overview....................................................................................................................................... 127 7.2.2.3.2 Implementation ............................................................................................................................. 128 7.2.2.3.3 Maximum ratio combining and complementary code combining................................................. 129 7.2.2.4 Selection of turbo code rate together with link layer protection......................................................... 130 7.2.2.5 Processing at PHY to support Erasure decoding in UL...................................................................... 131 7.2.2.5.1 Overview on the proposed algorithm............................................................................................ 132 7.2.2.5.2 MPEG-TS packet format .............................................................................................................. 132 7.2.2.5.3 Generation of MPEG-TS null-Packet ........................................................................................... 132 7.2.2.5.4 Signalling of wrong MPEG-TS packet ......................................................................................... 133 7.2.2.6 C/N performance values ..................................................................................................................... 133 7.2.2.6.1 Ideal performance in AWGN channel........................................................................................... 133 ETSI 6 ETSI TS 102 584 V1.1.2 (2010-XY) 7.2.2.6.2 Ideal performance in Rice channel................................................................................................ 134 7.2.2.6.3 Ideal performance in Rayleigh channel......................................................................................... 134 7.2.2.6.4 Ideal performance in burst erasure channel .................................................................................. 134 7.2.3 Time Interleaver ....................................................................................................................................... 135 7.2.3.1 Introduction ........................................................................................................................................ 135 7.2.3.2 Selection of receiver classes ............................................................................................................... 136 7.2.3.2.1 Overview....................................................................................................................................... 136 7.2.3.2.2 Class 1 / Class 2 compatible interleaver profile selection............................................................. 136 7.2.3.3 Interleaver profile description using the waveform parameters.......................................................... 137 7.2.3.3.1 Overview....................................................................................................................................... 137 7.2.3.3.2 Parameters and description ........................................................................................................... 139 7.2.3.3.3 Calculation examples .................................................................................................................... 140 7.2.3.3.4 Calculating the interleaver transit delay........................................................................................ 142 7.2.3.3.4.1 Calculation in terms of integer number of IU.......................................................................... 142 7.2.3.3.4.2 Calculation in terms of time .................................................................................................... 143 7.2.3.3.5 Typical interleaver profiles ........................................................................................................... 146 7.2.3.3.5.1 Short uniform interleaver profile (reference Terr48)............................................................... 146 7.2.3.3.5.2 Long uniform interleaver profile (reference Uni10)................................................................ 147 7.2.3.3.5.3 Long uniform-late interleaver profile (reference ULate10)..................................................... 148 7.2.3.3.5.4 Long early-late interleaver profile........................................................................................... 149 7.2.3.3.5.5 Long optimized-uniform-late interleaver profile..................................................................... 150 7.2.3.4 Interleaver profile alignment between satellite and terrestrial / different carriers in non-SFN configurations ..................................................................................................................................... 152 7.2.3.4.1 Overview....................................................................................................................................... 152 7.2.3.4.2 void ............................................................................................................................................... 154 7.2.3.4.3 Interleaver synchronization on the receiver side........................................................................... 154 7.2.3.5 Fast interleaver synchronization strategies ......................................................................................... 155 7.2.3.5.1 Motivation..................................................................................................................................... 155 7.2.3.5.2 Definition of jitter and delay......................................................................................................... 155 7.2.3.5.3 Early decoding .............................................................................................................................. 155 7.2.3.5.4 Late decoding................................................................................................................................ 156 7.2.3.5.5 Zapping time in case of long physical layer interleaver................................................................ 157 7.2.4 SH framing............................................................................................................................................... 158 7.2.4.1 OFDM ................................................................................................................................................ 159 7.2.4.2 TDM ................................................................................................................................................... 162 7.3 OFDM and TDM elements ............................................................................................................................ 164 7.3.1 OFDM Elements ...................................................................................................................................... 164 7.3.1.1 Overview ............................................................................................................................................ 164 7.3.1.2 Performance in mobile TU6 Channel ................................................................................................. 164 7.3.1.2.1 Simulated capacity performance................................................................................................... 164 7.3.1.2.2 Measured capacity performance ................................................................................................... 165 7.3.1.2.3 Simulated required C/N as a function of the mobile speed........................................................... 169 7.3.1.2.4 Measured required C/N as a function of the mobile speed ........................................................... 169 7.3.1.3 Hierarchical terrestrial modulation performance ................................................................................ 171 7.3.1.3.1 Hierarchical modulation implementation...................................................................................... 171 7.3.1.3.2 Raw performance results............................................................................................................... 172 7.3.1.3.3 Summary:...................................................................................................................................... 174 7.3.2 TDM Overview ........................................................................................................................................ 174 7.3.2.1 TDM Bandwidth Selection................................................................................................................. 175 7.4 Combining techniques ................................................................................................................................... 175 7.4.1 Overview of diversity combining techniques........................................................................................... 175 7.4.2 void .......................................................................................................................................................... 176 7.4.3 void .......................................................................................................................................................... 176 7.4.4 Experimental results for hybrid networks..................................................................................................... 176 7.4.4.1 Introduction ........................................................................................................................................ 176 7.4.4.2 Simulation results ............................................................................................................................... 178 7.4.4.3 Laboratory results............................................................................................................................... 179 7.4.4.4 Conclusions on laboratory measurements .......................................................................................... 183 7.4.4.5 Field trial results ................................................................................................................................. 183 7.4.4.5.1 Results from Field Trials – Italy, January 2010 ............................................................................ 183 7.5 Synchronization ............................................................................................................................................. 186 7.5.0 Introduction to time alignment of satellite / terrestrial signals under consideration of SHIP signaling ... 186 ETSI 7 ETSI TS 102 584 V1.1.2 (2010-XY) 7.5.1 Transmitter configuration in satellite-terrestrial SFN .............................................................................. 188 7.5.1.1 Parameter selection............................................................................................................................. 188 7.5.1.2 Recommendation on network equipment ........................................................................................... 191 7.5.1.2.1 Principles of SFN architecture ...................................................................................................... 191 7.5.1.2.2 DVB-SH particularities................................................................................................................. 191 7.5.1.2.3 Possible implementations.............................................................................................................. 193 7.5.2 Transmitter configuration in non-SFN hybrid networks .......................................................................... 193 7.5.2.1 Overview ............................................................................................................................................ 193 7.5.2.2 DVB-SH particularities ...................................................................................................................... 193 7.5.2.3 Possible Implementations of Interleaver Delay Compensation .......................................................... 194 7.5.2.3.1 Solution 1...................................................................................................................................... 194 7.5.2.3.1 Solution 2...................................................................................................................................... 194 7.5.2.3.1 Solution 3...................................................................................................................................... 195 7.5.3 Receiver synchronization and re-synchronization.................................................................................... 197 7.5.3.1 SH frame synchronization strategy 1 (without a priori information).................................................. 197 7.5.3.2 SH frame synchronization strategy 2 (without a priori information, for OFDM only)....................... 197 7.5.3.2.1 Option 1 ........................................................................................................................................ 197 7.5.3.2.2 Option 2 ........................................................................................................................................ 198 7.5.3.3 SH frame synchronization strategy 3 (with prediction) ...................................................................... 199 7.5.4 HP/LP Synchronization in Hierarchical mode ......................................................................................... 200 7.6 System throughput calculations ..................................................................................................................... 201 7.6.1 Calculation of the number of MPEG TS packets per SH Frame .............................................................. 201 7.6.2 Calculation of the SH Frame duration...................................................................................................... 201 7.6.3 Typical MPEG-TS bit rates for OFDM.................................................................................................... 201 7.6.3.1 Channel Bandwidth 5 MHz, OFDM................................................................................................... 202 7.6.3.2 Channel Bandwidth 1,7 MHz, OFDM................................................................................................ 202 7.6.4 Typical MPEG-TS bit rates for TDM ...................................................................................................... 202 7.6.4.1 Channel Bandwidth 5 MHz, TDM, with 15 % roll-off and QPSK in OFDM .................................... 202 7.6.4.2 Channel Bandwidth 5 MHz, TDM, with 15 % roll-off and 16QAM in OFDM ................................. 203 7.6.4.3 Channel Bandwidth 5 MHz, TDM, with 25 % roll-off and QPSK in OFDM .................................... 203 7.6.4.4 Channel Bandwidth 1,7 MHz, TDM, with 15 % roll-off and QPSK in OFDM ................................. 203 7.7 Coexistence with other Physical layer elements ............................................................................................ 203 7.7.1 Introduction.............................................................................................................................................. 204 7.7.2 Site sharing............................................................................................................................................... 204 7.7.3 Results of simulations .............................................................................................................................. 206 7.7.4 Theoretical impact of UMTS over DVB-SH............................................................................................ 208 7.7.4.1 Assessment of the Interference........................................................................................................... 209 7.7.4.1.1 Methodology................................................................................................................................. 209 7.7.4.2 Exclusion zone with Satellite only reception...................................................................................... 215 7.7.5 Measurement results from On-Field Campaign ....................................................................................... 217 7.7.5.1 Measurements setup ........................................................................................................................... 217 7.7.5.2 On-field Measurement Results ........................................................................................................... 218 7.7.6 Conclusions and Recommendations......................................................................................................... 221 8 Services ................................................................................................................................................222 8.1 General considerations................................................................................................................................... 222 8.2 Service classification ..................................................................................................................................... 223 8.2.1 Service categories..................................................................................................................................... 223 8.3 Service attributes ........................................................................................................................................... 224 8.3.1 Definition ................................................................................................................................................. 224 8.3.2 Application............................................................................................................................................... 224 8.4 Consequences of hybrid architecture on implementation of a DVB-SH service offering ............................. 225 8.4.1 Handover issue ......................................................................................................................................... 225 8.4.2 Network and service discovery mechanisms............................................................................................ 225 8.5 Other considerations ...................................................................................................................................... 226 9 Network configurations........................................................................................................................227 9.1 Considerations on network configurations .................................................................................................... 227 9.1.1 Mobile TV network targeting portable devices........................................................................................ 227 9.1.2 Mobile TV network targeting vehicle ...................................................................................................... 227 9.2 Synchronization of Satellite and CGC for Common content......................................................................... 228 9.2.1 Introduction.............................................................................................................................................. 228 ETSI 8 ETSI TS 102 584 V1.1.2 (2010-XY) 9.2.2 Terrestrial SFN......................................................................................................................................... 228 9.2.2.1 Principle.............................................................................................................................................. 228 9.2.2.2 SFN operation..................................................................................................................................... 228 9.2.2.3 SHIP solution ..................................................................................................................................... 229 9.2.3 Synchronization of Satellite and Terrestrial repeaters for the Common Content..................................... 229 9.2.3.1 SH-B and SH-A/MFN cases............................................................................................................... 229 9.2.3.2 SH-A case........................................................................................................................................... 230 9.2.3.2.1 Principle ........................................................................................................................................ 230 9.2.3.2.2 Application with near GEO........................................................................................................... 231 9.2.3.2.3 Architecture .................................................................................................................................. 234 9.3 Signalling....................................................................................................................................................... 235 9.3.1 SHIP, TPS and SF .................................................................................................................................... 235 9.3.2 MPE and MPE-IFEC................................................................................................................................ 235 9.3.3 PSI/SI case ............................................................................................................................................... 235 9.3.4 ESG case .................................................................................................................................................. 237 9.4 Considerations on the use of repeaters and their feeder links ........................................................................ 237 9.4.1 TR(a) On-channel regenerative repeaters................................................................................................. 237 9.4.2 TR(b) Non-regenerative gap-fillers.......................................................................................................... 239 9.4.2.1 Generalities......................................................................................................................................... 239 9.4.2.2 Filtering issues in non-regenerative gap-fillers .................................................................................. 240 9.4.2.3 Frequency synchronized transposing non-regenerative gap-fillers..................................................... 241 9.4.3 TR(c) Mobile transmitters........................................................................................................................ 241 9.4.3.1 Regenerative TR (c) ........................................................................................................................... 241 9.4.3.2 Non-regenerative TR(c)...................................................................................................................... 242 10 Reference Terminals ............................................................................................................................242 10.1 Top-level design considerations .................................................................................................................... 242 10.1.1 Terminal categories.................................................................................................................................. 242 10.1.2 Mobility aspects ....................................................................................................................................... 243 10.1.2.1 Mobile channels.................................................................................................................................. 243 10.1.2.2 Antenna and diversity considerations................................................................................................. 243 10.1.3 Service aspects ......................................................................................................................................... 244 10.2 Memory requirements for DVB-SH processing ............................................................................................ 244 10.3 DVB-SH reference receiver model ................................................................................................................ 245 10.3.1 Reference model....................................................................................................................................... 245 10.3.2 Receiver for Vehicular terminals ............................................................................................................. 246 10.3.3 Receiver for Terminals with telecom modem .......................................................................................... 246 10.4 Minimum signal input levels for planning..................................................................................................... 247 10.4.1 Noise floor for vehicular receiver ............................................................................................................ 247 10.4.2 Noise floor for handheld receiver............................................................................................................. 247 10.4.3 Minimum C/N requirements .................................................................................................................... 247 10.4.4 Minimum input levels .............................................................................................................................. 248 10.4.4.1 Sensitivity for vehicular receiver........................................................................................................ 248 10.4.4.2 Sensitivity for handheld receiver ........................................................................................................ 248 10.4.5 Antenna considerations ............................................................................................................................ 248 10.4.5.1 Antenna for terminal category 1 ......................................................................................................... 249 10.4.5.2 Antenna for terminal category 2a ....................................................................................................... 249 10.4.5.3 Antenna for terminal category 2b and 3 ............................................................................................. 249 10.4.5.4 Antenna diversity considerations........................................................................................................ 250 10.4.6 Maximum Input Power for Wanted and Unwanted Signals..................................................................... 250 10.4.7 G/T considerations ................................................................................................................................... 250 11 Network planning.................................................................................................................................251 11.1 Introduction to DVB-SH network planning................................................................................................... 251 11.2 DVB-SH reception conditions ....................................................................................................................... 252 11.2.1 Introduction.............................................................................................................................................. 252 11.2.2 DVB SH reception conditions.................................................................................................................. 253 11.3 Coverage definition ....................................................................................................................................... 255 11.3.1 Broadcaster approach ............................................................................................................................... 255 11.3.2 Cellular approach ..................................................................................................................................... 255 11.3.3 Satellite coverage ..................................................................................................................................... 255 11.3.4 Quality of coverage .................................................................................................................................. 255 ETSI 9 ETSI TS 102 584 V1.1.2 (2010-XY) 11.4 Network planning factors............................................................................................................................... 256 11.4.1 Introduction.............................................................................................................................................. 256 11.4.2 Channel dependant factor......................................................................................................................... 257 11.4.3 Technology dependant factors ................................................................................................................. 258 11.4.4 Network architecture or implementation optional factors ........................................................................ 258 11.5 Terrestrial Network Planning based on minimum equivalent field strength (broadcast approach) ............... 259 11.5.1 Introduction.............................................................................................................................................. 259 11.5.2 Quality of coverage .................................................................................................................................. 259 11.5.3 Network planning based on field strength computation methodology ..................................................... 259 11.5.4 Network planning factors ......................................................................................................................... 261 11.5.4.1 Location Correction............................................................................................................................ 261 11.5.4.1.1 Location correction for reception condition A.............................................................................. 261 11.5.4.1.2 Building penetration losses and location correction for reception conditions B1 and B2............. 261 11.5.4.1.3 Location correction for reception conditions C and D .................................................................. 262 11.5.4.2 Network architecture and implementation optional factors................................................................ 262 11.5.5 Field strength computation examples....................................................................................................... 263 11.5.5.1 Field strength computation for reception condition B2 ...................................................................... 263 11.5.5.2 Synthesis............................................................................................................................................. 264 11.5.6 Use of field strength based Radio Network Planning Tool example for DVB-SH Network Design ....... 264 11.6 Terrestrial Network Planning based on minimum received signal level (cellular approach)......................... 265 11.6.1 Cellular network topology........................................................................................................................ 265 11.6.2 Quality of coverage .................................................................................................................................. 266 11.6.3 Network planning tool.............................................................................................................................. 266 11.6.4 Network planning factors ......................................................................................................................... 267 11.6.4.1 Shadowing margins ............................................................................................................................ 267 11.6.4.1.1 Shadowing margins for reception condition A ............................................................................. 267 11.6.4.1.2 In building penetration losses and shadowing margins for reception condition B........................ 267 11.6.4.1.3 Shadowing margins for reception conditions C and D.................................................................. 268 11.6.4.2 Network architecture optional parameters.......................................................................................... 268 11.6.5 Link budgets examples for DVB-SH network planning based on minimum received signal level ......... 268 11.6.5.1 Link budgets for Reception condition B2........................................................................................... 269 11.6.5.2 Overall synthesis ................................................................................................................................ 271 11.6.6 Use of radio network planning tool example for DVB-SH network design............................................. 271 11.7 Satellite Network Planning ............................................................................................................................ 272 11.7.1 Methodology for Satellite Coverage Calculation ..................................................................................... 272 11.7.2 Basic formulas for Satellite link budgets calculation ............................................................................... 273 11.7.2.1 Satellite link margin ........................................................................................................................... 275 11.7.2.2 Satellite fade margin........................................................................................................................... 275 11.7.3 Required margins computation ................................................................................................................ 276 11.7.3.1 Margins required reception condition A............................................................................................. 277 11.7.3.2 Margins required in reception conditions C ....................................................................................... 279 11.7.4 Antenna diversity Gain ............................................................................................................................ 280 11.7.5 Link budget examples .............................................................................................................................. 280 11.7.5.1 Link budget for SH-A......................................................................................................................... 280 11.7.5.2 Link budget for SH-B......................................................................................................................... 286 11.7.5.3 Implementation losses and link budgets ............................................................................................. 287 11.7.5.4 Example of availability calculation for reception condition A ........................................................... 287 11.7.5.5 Example of availability calculation for reception condition C ........................................................... 288 11.7.6 Methodology for Satellite Coverage Calculation for reception condition C ............................................ 289 11.7.7 Example of satellite coverage calculation ................................................................................................ 290 11.8 Hybrid network planning ............................................................................................................................... 294 11.8.1 Introduction.............................................................................................................................................. 294 11.8.2 Hybrid network system model ................................................................................................................. 295 11.8.2.1 Overview ............................................................................................................................................ 295 11.8.2.2 Hybrid channel model and Geometrical Aspects................................................................................ 296 11.8.3 Impact on coverage .................................................................................................................................. 299 11.8.3.1 The Possible zones in the case of SHA Hybrid SFN .......................................................................... 299 11.8.3.2 Hybrid coverage improvement .......................................................................................................... 302 11.8.3.3 Combining gain .................................................................................................................................. 304 11.8.4 Interference limited zones in multibeam hybrid networks ....................................................................... 305 11.8.4.1 Introduction ....................................................................................................................................... 305 11.8.4.2 Reception conditions B1 and B2: Indoor reception............................................................................ 306 ETSI 10 ETSI TS 102 584 V1.1.2 (2010-XY) 11.8.4.3 Reception conditions A: Outdoor pedestrian...................................................................................... 308 11.8.4.3 Conclusions ........................................................................................................................................ 309 A.1 Geostationary Satellite Payload characteristics (S-band).....................................................................309 A.1.1 Payload architectures ..................................................................................................................................... 309 A.1.2 TWTA characteristics and Operating Point Optimization ............................................................................. 310 A.1.3 Performance over nonlinear channels ............................................................................................................ 312 A.1.3.1 Simulations............................................................................................................................................... 312 A.1.3.1.1 Single HPA per beam ......................................................................................................................... 312 A.1.3.1.2 Multiport Amplifier ............................................................................................................................ 313 A.1.3.3 Single HPA per beam – Laboratory Measurements ................................................................................. 314 A.2 Void.....................................................................................................................................................316 A.3 Simulation Conditions.........................................................................................................................316 A.3.1 Simulations with ideal channel estimator in demodulator .................................................................................... 316 A.3.1.1 System parameters ................................................................................................................................... 316 A.3.1.2 Physical layer FEC and De-mapper configuration ................................................................................... 318 A.3.2 Void ...................................................................................................................................................................... 319 A.3.3 Demodulator state machine principle ................................................................................................................... 319 A.3.3.1 States representation of a demodulator .................................................................................................... 319 A.3.3.2 OFDM simulations................................................................................................................................... 321 A.3.3.3 TDM simulations ..................................................................................................................................... 322 A.3.3.4 Analysis of acquisition time for LMS channels ....................................................................................... 322 A.3.4 MPE-IFEC Simulation conditions ........................................................................................................................ 324 A.3.4.1 Definitions and interfaces .................................................................................................................................. 324 A.3.4.2 Decoder architecture ................................................................................................................................ 324 A.3.4.3 Other parameters ...................................................................................................................................... 327 A.4 Void.....................................................................................................................................................327 A.5 Void.....................................................................................................................................................327 A.6 Void.....................................................................................................................................................327 A.7 Propagation channels ...........................................................................................................................327 A.7.1 Terrestrial Channels.............................................................................................................................................. 328 A.7.1.1 Channel Model Definitions................................................................................................................................ 328 A.7.1.1.1 Reference channels: AWGN and Rayleigh..................................................................................................... 328 A.7.1.1.2 Single transmitter: Classical TU6 ................................................................................................................... 328 A.7.1.1. 3 Single transmitter: TU6 plus lognormal shadowing ...................................................................................... 329 A.7.1.1.3 SFN MBRAI model 1 for weak long echoes .................................................................................................. 329 A.7.1.1.4 SFN MBRAI model 2 for strong long echoes................................................................................................. 329 A.7.1.1.5 SFN MBRAI model 3 for strong short echoes................................................................................................ 330 A.7.1.1.6 Portable Indoor and Outdoor channel model .................................................................................................. 330 A.7.2 Satellite Channels ................................................................................................................................................. 331 A.7.2.1 Propagation channels ............................................................................................................................... 331 A.7.2.1.1 AWGN Channel.............................................................................................................................................. 331 A.7.2.1.2 Rician channel Model ..................................................................................................................................... 331 A.7.2.1.3 Land Mobile Satellite Channel Model ............................................................................................................ 332 A.7.2.1.4 Satellite quasi stationary channel........................................................................................................ 335 A.8 Evaluation criteria ...............................................................................................................................336 A.8.1 Introduction.................................................................................................................................................... 336 A.8.2 Laboratory and field trial logging .................................................................................................................. 336 A.8.2.1 Overview.................................................................................................................................................. 336 A.8.2.2 Physical layer logging capabilities ........................................................................................................... 338 A.8.2.3 Link layer logging capabilities ................................................................................................................. 339 A.8.3 Performance metrics ...................................................................................................................................... 340 A.8.3.1 Introduction.............................................................................................................................................. 340 A.8.3.2 Output of the turbo decoder (R3) ............................................................................................................. 341 A.8.3.3 After PID filtering (R2)............................................................................................................................ 341 A.8.3.4 After MPE decapsulation and MPE-IFEC decoding (R1)........................................................................ 341 A.8.3.5 ESR5(20) metrics ..................................................................................................................................... 341 A.8.3.5.1 Generic definition............................................................................................................................... 341 ETSI 11 ETSI TS 102 584 V1.1.2 (2010-XY) A.8.3.5.2 F-ESR5(20) and Mi-ESR5(20) ........................................................................................................... 342 A.8.3.5.3 IP_ESR5(20) ...................................................................................................................................... 342 A.8.3.6 Synthesis .................................................................................................................................................. 342 A.9 Summary of simulated cases ................................................................................................................343 A.10 Configurations description ...................................................................................................................347 A.10.1 Introduction.................................................................................................................................................... 348 A.10.2 OFDM class 1 ................................................................................................................................................ 352 A.10.3 OFDM class 2 ................................................................................................................................................ 356 A.10.3 TDM class 1................................................................................................................................................... 379 A.10.4 TDM class 2................................................................................................................................................... 385 A.11 Reference Demodulator........................................................................................................................406 A.11.1 TDM Case...................................................................................................................................................... 406 A.11.1.1 TDM Reference Demodulator.................................................................................................................. 406 A.11.1.2 TDM Reference Demodulator Simulation Results................................................................................... 407 A.11.2 OFDM Case ................................................................................................................................................... 412 A.11.2.1 OFDM Reference Demodulator (see note)............................................................................................... 412 A.11.2.2 OFDM Reference Demodulator Simulation Results ................................................................................ 416 A.11.2.2.1 Terrestrial Channel ............................................................................................................................. 416 A.11.2.2.2 Satellite Channel................................................................................................................................. 420 A.11.3 TDM Signalling Channel Performance Results............................................................................................. 422 A.12 Simulation results.................................................................................................................................424 A.12.1 Summary of the Results for LMS channel..................................................................................................... 426 A.12.2 Detailed results in LMS environments: Reference Cases and Sensitivity analysis........................................ 427 A.12.2.1 Reference Cases results............................................................................................................................ 427 A.12.2.2 Sensitivity Analysis.................................................................................................................................. 429 A.12.2.3 Vehicular reception in LMS environments .............................................................................................. 430 A.12.2.3.1 Vehicular Reception in LMS-SU and SH-A Waveform (OFDM) ..................................................... 430 A.12.2.3.2 Vehicular Reception in a LMS-SU and SH-B Waveform (TDM)...................................................... 431 A.12.2.3.3 Vehicular Reception in a LMS ITS and DVB SH-A Waveform (OFDM)......................................... 433 A.12.2.3.4 Vehicular Reception in a LMS ITS and DVB SH-B Waveform (TDM)............................................ 434 A.12.2.3.5 Category 2b reception in LMS-SU at 68 dBW satellite EIRP............................................................ 436 A.12.2.4 Uniform Late (UL) and Uniform (U) long interleavers ........................................................................... 438 A.12.2.5 Demodulator state machine performance impact ..................................................................................... 441 A.12.2.6 C/I performance impact............................................................................................................................ 441 A.12.2.6 ESR5 fulfilment at different spectral efficiencies .................................................................................... 443 A.12.3 Reception in TU6 environment...................................................................................................................... 444 A.12.3.1 FER and ESR5 relationship...................................................................................................................... 444 A.12.3.2 FER performance in TU6......................................................................................................................... 445 A.12.3.3 ESR5 Performance in TU......................................................................................................................... 446 A.13 Experimental results.............................................................................................................................449 A13.1 Laboratory results .......................................................................................................................................... 449 A.13.1.1 Terrestrial channels .................................................................................................................................. 449 A13.1.1.1 Configurations .................................................................................................................................... 449 A.13.1.1.2 Synthetic results............................................................................................................................................ 450 A13 1.1.3 Impact of different parameters ....................................................................................................................... 451 A13.1.1.4 Conclusions.................................................................................................................................................... 456 A.13.1.2 Satellite channels measurements results................................................................................................... 456 A13.1.2.1 Measurements with receiver 1 ............................................................................................................ 456 A.13.1.2.1.1 Configurations ........................................................................................................................................... 456 A13.1.2.1.2 Examples of Results...................................................................................................................... 457 A.13.1.2.1.3 Conclusions................................................................................................................................................ 460 A.13.1.2.2 Measurements with receiver 2 ...................................................................................................................... 462 A.13.1.2.2.1 Configurations ........................................................................................................................................... 462 A.13.1.2.2.2 Results 462 A.13.1.2.2.3 Conclusions on receiver 2............................................................................................................. 465 A.13.1.2.3 Conclusions ........................................................................................................................................ 466 A.13.1.3 Hybrid channels ....................................................................................................................................... 467 A13.1.3.1 Configurations .................................................................................................................................... 467 A13.1.3.2 Results ................................................................................................................................................ 469 ETSI 12 ETSI TS 102 584 V1.1.2 (2010-XY) A13.1.3.2.1 Suburban channels, ..................................................................................................................................... 469 A13.1.3.2.2 Rural Channel, ........................................................................................................................................... 474 A13.1.3.2.3 Case of Class 1 plus MPE-IFEC ................................................................................................................. 478 A13.1.3.3 Interpretation and Conclusions ...................................................................................................................... 479 A.13.2 Field trials results........................................................................................................................................... 481 A.13.2.1 Introduction............................................................................................................................................. 481 A.13.2.2 Pau trial ................................................................................................................................................... 482 A.13.2.2.1 Introduction ........................................................................................................................................ 482 A.13.2.2.2 Experimental set up ........................................................................................................................... 482 A.13.2.2.2.1 Base Station Configuration ........................................................................................................... 483 A.13.2.2.2.2 Helicopter Configuration .............................................................................................................. 483 A.13.2.2.2.3 Receiver Configurations ............................................................................................................... 484 A.13.2.2.3 Waveform performance results .......................................................................................................... 485 A.13.2.2.3.1 Waveform configurations ............................................................................................................. 485 A.13.2.2.3.2 Outdoor pedestrian........................................................................................................................ 486 A.13.2.2.3.3 Indoor pedestrian .......................................................................................................................... 487 A.13.2.2.3.4 Waveform performances synthesis ............................................................................................... 488 A.13.2.2.4 SFN measurements............................................................................................................................. 488 A.13.2.2.4.1 Methodology................................................................................................................................. 488 A.13.2.2.4.2 The results..................................................................................................................................... 489 A.13.2.2.5 Indoor diversity measurements........................................................................................................... 490 A.13.2.2.5.1 Methodology................................................................................................................................ 490 A.13.2.2.5.2 Diversity Measurements ............................................................................................................... 491 A.13.2.3 Torino trial .............................................................................................................................................. 491 A.13.2.3.1 Introduction ........................................................................................................................................ 491 A.13.2.3.2 Experimental set up ........................................................................................................................... 492 A.13.2.3.3 Performances of DVB-SH waveform and diversity gain analysis in reception condition C .............. 495 A.13.2.3.3.1 Waveform configurations ............................................................................................................. 495 A.13.2.3.3.2 Diversity gain evaluation with the SNR @ FER 5 method........................................................... 496 A.13.2.3.3.3 Conclusion on diversity gain in outdoor environment (reception condition C) ............................ 497 A.13.2.3.4 Performances of DVB-SH waveform and diversity gain analysis in reception condition D .............. 497 A.13.2.3.4.1 Diversity gain evaluation with the SNR @ FER 5 method.......................................................... 497 A.13.2.3.4.2 Conclusion on Diversity gain in reception condition D ............................................................... 499 A.13.2.3.5 Impact of the interleaving scheme ..................................................................................................... 499 A.13.2.3.5.1 MPE-IFEC interleaving gain in Reception Conditions C and D.................................................. 499 A.13.2.3.5.2 Class 2 interleaving gain in Reception Conditions C and D ........................................................ 499 A.13.2.3.5.3 Conclusion on the use of long interleaving................................................................................... 500 A.13.2.6 Synthesis of results ............................................................................................................................ 501 A.13.2.4 Measurements results of a DVB-SHA satellite signal from Paris to Barcelona....................................... 502 A.13.2.4.1 Receiver platform description ............................................................................................................ 502 A.13.2.4.2 Main measurements results ................................................................................................................ 503 A.13.2.4.2.1 Erroneous seconds on over 1000km of highways........................................................................ 503 A.13.2.4.2.2 ”Frame_lock” loss events distribution analysis ........................................................................... 504 A.13.2.4.2.3 Erroneous seconds versus interleaving depth ............................................................................... 505 A.13.2.4.2.4 Percentage of interruptions versus interleaving depths................................................................. 505 A.13.2.4.3 Conclusions ........................................................................................................................................ 506 A.13.2.5 Pisa trial................................................................................................................................................... 506 A.13.2.5.1 Experimental Setup....................................................................................................................... 506 A.13.2.5.2 Description of the trials conditions ............................................................................................... 509 A.13.2.5.3 Selected configurations................................................................................................................. 511 A.13.2.5.4 Trial results ................................................................................................................................... 512 A.13.2.5.4 Hybrid SFN trials.......................................................................................................................... 523 A.13.2.5.5 Conclusion on Pisa trials............................................................................................................... 523 A.13.2.6 Conclusion on trial results .................................................................................................................. 523 Clause B (informative): Interoperability with cellular telephony networks ...........................................524 B.1 Introduction ..........................................................................................................................................524 B.1.1 General coexistence issues............................................................................................................................. 524 B.1.2 Terminal Architectures .................................................................................................................................. 524 B.2 Frequency bands and power levels.......................................................................................................525 ETSI 13 ETSI TS 102 584 V1.1.2 (2010-XY) B.3 Interference due to uplink signal..........................................................................................................526 B.3.1 Desensitization............................................................................................................................................... 526 B.3.2 Spurious and transmitted PA noise ................................................................................................................ 526 B.4 Cellular Radio Downlink Signal Interference to DVB-SH Receiver...................................................527 B.4.1 Adjacent channel ........................................................................................................................................... 527 B.4.2 ACS requirements.......................................................................................................................................... 528 B.4.2.1 Example of ACS value measurement................................................................................................................. 528 B.4.2.2 Recommendations.............................................................................................................................................. 529 B.4.3 Co channel Interference ........................................................................................................................................ 529 Clause C (informative): Spectrum efficiency and system throughput.....................................................530 C.1 Spectrum efficiency analysis................................................................................................................530 C.1.1 Spectral Efficiency Calculations.................................................................................................................... 531 C.1.2 Numerical examples ...................................................................................................................................... 532 C.2.3 Conclusions on System Spectrum Efficiency ................................................................................................ 533 C.2 Throughput calculations.......................................................................................................................533 C.2.1 Reference DVB-SH parameters..................................................................................................................... 533 C.2.1.1 OFDM Frame ........................................................................................................................................... 534 C.2.1.2 TDM Frame.............................................................................................................................................. 534 C.2.2 Calculation of the SH Frame duration ........................................................................................................... 535 C.2.3 Calculation of the number of MPEG TS packets per SH Frame.................................................................... 535 C.2.4 Calculation of the capacity............................................................................................................................. 536 History ............................................................................................................................................................538 ETSI 14 ETSI TS 102 584 V1.1.2 (2010-XY) Intellectual Property Rights IPRs essential or potentially essential to the present document may have been declared to ETSI. The information pertaining to these essential IPRs, if any, is publicly available for ETSI members and non-members, and can be found in ETSI SR 000 314: "Intellectual Property Rights (IPRs); Essential, or potentially Essential, IPRs notified to ETSI in respect of ETSI standards", which is available from the ETSI Secretariat. Latest updates are available on the ETSI Web server (http://webapp.etsi.org/IPR/home.asp). Pursuant to the ETSI IPR Policy, no investigation, including IPR searches, has been carried out by ETSI. No guarantee can be given as to the existence of other IPRs not referenced in ETSI SR 000 314 (or the updates on the ETSI Web server) which are, or may be, or may become, essential to the present document. Foreword This Technical Specification (TS) has been produced by Joint Technical Committee (JTC) Broadcast of the European Broadcasting Union (EBU), Comité Européen de Normalisation ELECtrotechnique (CENELEC) and the European Telecommunications Standards Institute (ETSI). The present document is complementary to EN 302 583 [1], which provides the waveform-level specifications for the use of TV Broadcast in DVB Satellite to Handhelds applications. The present document is complementary to TS 102 585 [2], which provides the system-level specifications for the use of TV Broadcast in DVB Satellite to Handhelds applications. NOTE: The EBU/ETSI JTC Broadcast was established in 1990 to co-ordinate the drafting of standards in the specific field of broadcasting and related fields. Since 1995 the JTC Broadcast became a tripartite body by including in the Memorandum of Understanding also CENELEC, which is responsible for the standardization of radio and television receivers. The EBU is a professional association of broadcasting organizations whose work includes the co-ordination of its members' activities in the technical, legal, programme-making and programme-exchange domains. The EBU has active members in about 60 countries in the European broadcasting area; its headquarters is in Geneva. European Broadcasting Union CH-1218 GRAND SACONNEX (Geneva) Switzerland Tel: +41 22 717 21 11 Fax:: +41 22 717 24 81 Founded in September 1993, the DVB Project is a market-led consortium of public and private sector organizations in the television industry. Its aim is to establish the framework for the introduction of MPEG-2 based digital television services. Now comprising over 200 organizations from more than 25 countries around the world, DVB fosters market-led systems, which meet the real needs, and economic circumstances, of the consumer electronics and the broadcast industry. Introduction The present document gives the first guidelines for setting up networks and services using the Digital Video Broadcasting - Satellite to Handheld (DVB-SH) specifications. Updates to the present document will be produced when more results become available. Document summary Clause 4 gives an outline of the DVB-SH system, the main hybrid satellite/terrestrial architectures, the distinction between SH-A and SH-B as well as between SFN and non-SFN, the possible approaches for frequency planning, the frequency bands and their regulatory aspects, the characteristics of satellite and terrestrial propagation channels, the receivers constraints, and other system considerations. ETSI 15 ETSI TS 102 584 V1.1.2 (2010-XY) Clause 5 gives a discussion of possible DVB-SH system configurations and likely scenarios of their deployment sequence. Clause 6 introduces the elements of the DVB-SH specifications at the link layer and above. These are, at the link layer, time-slicing and LL-FEC and, at above this layer, the supports of VBR/Statmux and of Mobility. Compatibility with DVB-H link layer is discussed. Clause 7 clarifies the physical layer elements of the DVB-SH specifications that are new to the DVB families of standards, namely the 3GPP2 turbo codes, the long physical time interleaver and its implication on receivers (Class 1 and Class 2 receivers), the 1K-FFT mode for OFDM and the diversity combining techniques (MFN). Clause 8 discusses services and usage scenarios. This clause presents concepts such as service categories and attributes and highlights the consequences of the hybrid architectures on a personal mobile TV service offering. Clause 9 is devoted to DVB-SH network configuration. The following issues are discussed: Network configurations, Service Information, Handover. Clause 10 provides preliminary information about the DVB-SH reference terminals and reference receivers. Clause 11 is devoted to Network coverage planning. The main purpose is to present reference data and methodologies for transmitter network (both satellite and terrestrial) sizing as a function of coverage. Finally three clausees are included: • Clause A documents the methodology used by TM-SSP and presents the simulation and experimental results that support the recommendations given. • Clause B details the implications for "convergence terminals" in which broadcast (DVB-SH, DVB-H) and telecommunication (GSM/UMTS) technologies co-exist. • Clause C documents the methodology and formulas for spectrum efficiency analysis and throughput calculations. ETSI 16 ETSI TS 102 584 V1.1.2 (2010-XY) 1 Scope The present document provides guidelines for the use and implementation of ETSI Digital Video Broadcasting–Satellite to Handheld (DVB-SH) standard EN 302 583 [1] in the context of providing an efficient way of carrying multimedia services over digital hybrid satellite/terrestrial broadcasting networks to handheld terminals. The document should be read in conjunction with the: • DVB-SH waveform specifications [1]; • DVB-SH system specifications [2]; • DVB IP Datacast specifications [21] and [22]; • DVB Data broadcasting specifications [9]; • DVB Specifications for Service information (SI) [24]. Objective The present document draws attention to the technical questions that need to be answered when setting up DVB-SH services and networks and offers some guidance in finding answers to them. It does not cover in detail, issues linked to the content of the broadcasts such as Coding Formats, Electronic Programme Guides (EPG), Access Control (CA), etc. Target readers The present document is aimed at the Technical Departments of organizations that are considering implementing digital hybrid satellite/terrestrial broadcasting to handheld devices. It assumes that readers are familiar with digital satellite and terrestrial networks. Contributors The present document was prepared by members of the Ad-hoc group TM-SSP from the DVB Project. Members include broadcasters, network operators and professional as well as domestic equipment manufacturers. 2 References References are either specific (identified by date of publication and/or edition number or version number) or non-specific. • For a specific reference, subsequent revisions do not apply. • Non-specific reference may be made only to a complete document or a part thereof and only in the following cases: - if it is accepted that it will be possible to use all future changes of the referenced document for the purposes of the referring document; - for informative references. Referenced documents which are not found to be publicly available in the expected location might be found at http://docbox.etsi.org/Reference. NOTE: While any hyperlinks included in this clause were valid at the time of publication ETSI cannot guarantee their long term validity. ETSI 17 ETSI TS 102 584 V1.1.2 (2010-XY) 2.1 Normative references The following referenced documents are indispensable for the application of the present document. For dated references, only the edition cited applies. For non-specific references, the latest edition of the referenced document (including any amendments) applies. [1] ETSI EN 302 583: "Digital Video Broadcasting (DVB); Framing Structure, channel coding and modulation for Satellite Services to Handheld devices (SH) below 3 GHz". [2] ETSI TS 102 585: "Digital Video Broadcasting (DVB); System Specifications for Satellite services to Handheld devices (SH) below 3 GHz". [3] ETSI EN 302 304: "Digital Video Broadcasting (DVB); Transmission System for Handheld Terminals (DVB-H)". [4] Void. [5] Void. [6] ETSI EN 302 307: "Digital Video Broadcasting (DVB); Second generation framing structure, channel coding and modulation systems for Broadcasting, Interactive Services, News Gathering and other broadband satellite applications". [7] Void. [8] ISO/IEC 13818-1 (2000): "International Standard, Information technology, Generic coding of moving pictures and associated audio information: Systems". [9] ETSI EN 301 192: "Digital Video Broadcasting (DVB); DVB specification for data broadcasting". [10] Void. [11] ITU-R Recommendation M.1225 (1997): "Guidelines for evaluation of Radio Transmission Technologies for IMT-2000". [12] ITU-R Recommendation P.681-4 (1997): "Propagation data required for the design of earth-space land mobile telecommunication systems". [13] ITU-R-Recommendation P.1546-2 (2005): "Method for point-to-area predictions for terrestrial services in the frequency range 30 MHz to 3 000 MHz". [14] EICTA/TAC/MBRAI-02-16 Version 1, 2004: "Mobile and Portable DVB-T Radio Access Interface Specifications". [15] ETSI TS 125 104 (V4.3.0): "Universal Mobile Telecommunications System (UMTS); Base Station (BS) radio transmission and reception (FDD) (3GPP TS 25.104 version 4.3.0 Release 4)". [16] IEEE Trans. on Vehic. Techn., Vol. 50, No. 6, November 2001: "Statistical Modelling of the LMS Channel"; Perez-Fontan, F., Vazquez-Castro M, Enjamio Cabado, C., Pita Garcia, J. and Kubista, E. [17] IEEE Trans. On Broadcasting, Vol. 44, No. 1, March 1998, pp. 40-75: "S-band LMS channel behaviour for different environments, degrees of shadowing and elevation angles"; Perez-Fontan, F., Vazquez-Castro M, S. Buonomo, Poiares-Baptista J.P., Arbesser-Rastburg B. [18] Commission of the European Communities, Directorate General Telecommunications, Information Industries and Innovation, 1989, pp. 135-147 : "Digital land mobile radio communications (final report)"; (under the direction of M. Failly). NOTE: Available at http://www.awe-communications.com/Links/COST.html. [19] Broadcast Mobile Convergence Forum. Studies & White Paper: "Mobile Broadcast Technologies; Link Budgets"; January 2007. NOTE: Available at www.bmcoforum.org. ETSI 18 ETSI TS 102 584 V1.1.2 (2010-XY) [20] ITU Radio Regulations (2008) - Complete texts of te Radio Regulations as adopted by the World Radio Communication Conferences (WARC). NOTE: Available at www.itu.int/publ/R-REG-RR/en. [21] ETSI TS 102 470- 2: "Digital Video Broadcasting (DVB); IP Datacast over DVB-SH: Program Specific Information (PSI)/Service Information (SI)". [22] ETSI TS 102 472: "Digital Video Broadcasting (DVB); IP Datacast over DVB-H: Content Delivery Protocols". [23] ISO/IEC 13818-6: "Information technology -- Generic coding of moving pictures and associated audio information; Part 6: Extensions for DSM-CC". [24] ETSI EN 300 468: "Digital Video Broadcasting (DVB); Specification for Service Information (SI) in DVB systems". [25] ITU-R Recommendation P.676 (2007): "Attenuation by atmospheric gases". [26] IEEE 802.11 (2007): "Standard for Information technology-Telecommunications and information exchange between systems-Local and metropolitan area networks-Specific requirements; Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications". [27] ITU-R Recommendation P.526 (2007): " Propagation by diffraction". [28] ETSI EN 300 744: "Digital Video Broadcasting (DVB); Framing structure, channel coding and modulation for digital terrestrial television". [29] ETSI TS 145 005: " Digital cellular telecommunications system (Phase 2+); Radio transmission and reception (3GPP TS 45.00)". [30] ETSI TS 100 910: "Digital cellular telecommunications system (Phase 2+); Radio Transmission and Reception (3GPP TS 05.05)". [31] ITU-R Recommendation P.618 (2007): "Propagation data and prediction methods required for the design of Earth-space telecommunication systems". [32] ITU-R Recommendation P.531 (2007): "Ionospheric propagation data and prediction methods required for the design of satellite services and systems". [33] ITU-R Recommendation P.680 (1999): "Propagation data required for the design of Earth-space maritime mobile telecommunication systems". [34] ETSI TS 102 591-2: “Digital Video Broadcasting (DVB); IP Datacast: Content Delivery Protocols (CDP) Implementation Guidelines;Part 2: IP Datacast over DVB-SH”. [35] ETSI TS 102 611-2: “Digital Video Broadcasting (DVB); IP Datacast: Implementation Guidelines for Mobility;Part 2: IP Datacast over DVB-SH”. [36] ETSI TS 102 772: “Digital Video Broadcasting (DVB); Specification of Multi-Protocol Encapsulation - inter-burst Forward Error Correction [37] ETSI TS 102 592-2: “Digital Video Broadcasting (DVB); IP Datacast: Electronic Service Guide (ESG) Implementation Guidelines;Part 2: IP Datacast over DVB-SH”. 2.2 Informative references The following referenced documents are not essential to the use of the present document but they assist the user with regard to a particular subject area. For non-specific references, the latest version of the referenced document (including any amendments) applies. [i.1] "Antennas and Propagation for Wireless Communication Systems" Chapter 9, Section 9.5. ETSI 19 ETSI TS 102 584 V1.1.2 (2010-XY) [i.2] EERL Technical Report A2A-98-U-0-021 (APL)EERL-98-12A, December 1998: "Handbook of Propagation Effects for Vehicular and Personal Mobile Satellite; Systems Overview of Experimental and Modeling Results"; J.Goldhirsh and W. J. Vogel. [i.3] IEEE Trans. On Vehicular Technology, Vol.43, No.2, 1991: "The Land Mobile Satellite Channel Communication Channel-Recording, Statistics and Channel Model", E.Lutz, et al. [i.4] IEEE Trans. Veh. Technol., Vol. VT-34, No.3, August 1985: "A Statistical Model for a Land Mobile Satellite Link"; Loo, C. [i.5] IEEE Trans. Veh. Technol., Vol. 43, No. 3, August 1994: "A Statistical Model for Land Mobile Satellite Channels and Its Application to Nongeostationary Orbit Systems"; Corazza, G.E. andVatalaro, F. [i.6] IEEE 47th, Volume 1, Issue, 4-7 May 1997 Page(s):41; 45 vol.1: "A channel model for nongeostationary orbiting satellite system"; Vehicular Technology Conference, 1997; Seung-Hoon Hwang, Ki-Jun Kim, Jae-Young Ahn and Keum-Chan Whang. [i.7] EBU Technical Review; Summer 1998: "The effects of phase noise in COFDM"; J. Stott. [i.8] "Simulation of Communication Systems" ( Plenum Press, 1992); M. C. Jeruchin, P. Balaban, and K. S. Shanumgan. [i.9] DVB Technical Module working document; TM3873; DVB-SH Mobility Scenarios. [i.10] ETSI TR 143 030: "Digital cellular telecommunications system (Phase 2+); Radio network planning aspects; (3GPP TR 43.030 Release 7)". [i.11] Coverage Requirements for UMTS. HiQ Data AB. [i.12] Siemens AG, Communications Mobile Networks Munich, Germany: "Comparison of IEEE 802.16 WiMax Scenarios with Fixed and Mobile Subscribers in Tight Reuse", C.F. Ball, E. Humburg, K. Ivanov, and F. Treml. [i.13] "Radio Network Planning and Process and Methods for WCDMA", Jaana Laiho, and Achim Wacker (Nokia Networks). [i.14] ETSI TR 125 996: "Universal Mobile Telecommunications System (UMTS); Spacial channel model for Multiple Input Multiple Output (MIMO) simulations (3GPP TR 25.996 Release 6)". [i.15] "Predicting and Verifying Cellular Network Coverage", Pete Bernardin, University Of Texas. [i.16] International Journal on Satellite Communication Networks, 2004; 22: "DVB-S2 modem algorithms design and performance over typical satellite channels", E. Casini, R. De Gaudenzi, and A. Ginesi. [i.17] "Upper layer for DVB-SH; Technical solutions and performance"; A. Morello, V. Mignone, G. Vitale (RAI, Centro Ricerche e Innovazione Tecnologica, Turin, Italy), P. Burzigotti (ESA/ESTEC, Noordwijk, The Netherlands). [i.18] OFCOM: "Report on future performance of video codec". NOTE: Available at : http://privatewww.essex.ac.uk/~fleum/FinalOfcomReport.pdf. [i.19] IEEE Trans. on Circuits and Systems for Video Technology, 13(17):704-716, 2003: "H.264/AVC Baseline Profile Decoder Complexity Analysis"; M. Horowitz, A. Joch, F. Kossentini, and A. Hallapuro. [i.20] C. E. Gilchriest: "Signal-to-Noise Monitoring", JPL Space programs Summary, No. 37-27, Vol. IV, pp. 169-176. [i.21] ETSI TR 102 377 (V1.2.1): "Digital Video Broadcasting (DVB); DVB-H Implementation Guidelines". [i.22] ETSI TR 101 190 (V1.2.1): "Digital Video Broadcasting (DVB); Implementation guidelines for DVB terrestrial services; Transmission aspects". ETSI 20 ETSI TS 102 584 V1.1.2 (2010-XY) [i.23] ETSI TR 102 376 (V1.1.1): "Digital Video Broadcasting (DVB) User guidelines for the second generation system for Broadcasting, Interactive Services, News Gathering and other broadband satellite applications (DVB-S2)". [i.24] ETSI TR 101 112 (V3.2.0): "Universal Mobile Telecommunications System (UMTS); Selection procedures for the choice of radio transmission technologies of the UMTS (UMTS 30.03 version 3.2.0)". [i.25] ECC Decision ECC/DEC/(05)05 of 18 March 2005 on harmonised utilisation of spectrum for IMT-2000/UMTS systems operating within the band 2500 – 2690 MHz. NOTE: Available at http://www.ero.dk/documentation/docs/doc98/official/pdf/ECCDEC0505.PDF. [i.26] Commission Decision 2007/98/EC of 14 February 2007 on the harmonised use of radio spectrum in the 2 GHz frequency bands for the implementation of systems providing mobile satellite services. NOTE: Available at http://eur-lex.europa.eu/LexUriServ/site/en/oj/2007/l_043/l_04320070215en00320034.pdf. [i.27] ECC Decision ECC/DEC/(04)09 of 12 November 2004 on the designation of the bands 1518 - 1525 MHz and 1670 - 1675 MHz for the Mobile-Satellite Service. NOTE: Available at http://www.ero.dk/documentation/docs/doc98/official/pdf/ECCDEC0409.PDF. [i.28] ECC Decision ECC/DEC/(03)02 of 17 October 2003 on the designation of the frequency band 1479.5 – 1492 MHz for use by Satellite Digital Audio Broadcasting systems. NOTE: Available at http://www.ero.dk/documentation/docs/doc98/official/pdf/ECCDEC0302.PDF. [i.29] ECC Decision ECC/DEC/(06)09 of 1 December 2006 on the designation of the bands 1980-2010 MHz and 2170-2200 MHz for use by systems in the Mobile-Satellite Service including those supplemented by a Complementary Ground Component (CGC). NOTE: Available at http://www.ero.dk/documentation/docs/doc98/official/Word/ECCDEC0609.DOC. [i.30] ECC Decision ECC/DEC/(06)10 of 1 December 2006 on transitional arrangements for the Fixed Service and tactical radio relay systems in the bands 1980-2010 MHz and 2170-2200 MHz in order to facilitate the harmonised introduction and development of systems in the Mobile Satellite Service including those supplemented by a Complementary Ground Component. NOTE: Available at http://www.erodocdb.dk/docs/doc98/Official/word/ECCDec0610.doc. [i.31] ECC Recommendation (06)05 on the provision of information on the progress of implementation of the mobile satellite systems which are candidates to use the 1980-2010 MHz and 2170-2200 MHz MSS frequency bands. NOTE: Available at http://www.erodocdb.dk/docs/doc98/official/pdf/Rec0605.pdf. [i.32] ETSI EN 301 442: "Satellite Earth Stations and Systems (SES); Harmonized EN for Mobile Earth Stations (MESs), including handheld earth stations, for Satellite Personal Communications Networks (S-PCN) in the 2,0 GHz bands under the Mobile Satellite Service (MSS) covering essential requirements under Article 3.2 of the R&TTE directive". [i33] Decision 2008/626/EC of the European Parliament and of the Council of 30 June2008 on the selection and authorisation of systems providing mobile satellite services (MSS). NOTE: Available at http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2008:172:0015:0024:EN:PDF [i.34] EICTA, MBRAI documents MOBILE AND PORTABLE DVB-T/H RADIO ACCESS – Part 1: Interface specification. NOTE: Available at http://www.eicta.org/web/news/telecharger.php?iddoc=587 [i.33] ESA J-Ortigia ARTES 3 Project NOTE: Available at http://telecom.esa.int/telecom/www/object/index.cfm?fobjectid=29990 ETSI 21 ETSI TS 102 584 V1.1.2 (2010-XY) [i.34] Draft ETSI EN 302 574-1 V <0.1.2> (2009-01) Harmonized standard for Satellite Earth Station for MSS operating in the 1 980 MHz to 2 010 MHz (earth-to-space) and 2 170 MHz to 2 200 MHz (space –to-earth) frequency bands [i.35] 3GPP TS 25.101 v 7.16.0 (2009-05) Technical Specification Group Radio Access Network; User Equipment (UE) radio transmission and reception (FDD) [i.36] 3GPP TS 25.104 v 9.2.0 (2009-12) 3rd Generation Partnership Project; Technical Specification Group Radio Access Network; Base Station (BS) radio transmission and reception (FDD) (Release 9) [i.37] Agilent Spectrum Analysis Basis. Application note 150, chapter 5. [i.38] Performance evaluation of channel change for DVB-SH streaming services, ICC 2010, Marie-Line Alberi Morel, Sylvaine Kerboeuf, Bessem Sayadi, Yann Leprovost, and Frederic Faucheux, Alcatel Lucent Bell Labs France, Route de Villejust 91620 NOZAY, France, E-mail: {marie_line.alberi-morel, sylvaine.kerboeuf}@alcatel-lucent.com, {bessem.sayadi, yann.leprovost, frederic.faucheux}@alcatel-lucent.com 3 Definitions and abbreviations 3.1 Definitions For the purposes of the present document, the following terms and definitions apply: common content (or Service): content available via the SC and also obligatorily transmitted in each cell of the CGC Complementary Ground Component (CGC): terrestrial leg of a hybrid network that uses a portion of the frequency allocated to that network, under a special regulatory regime decided at regional or national levels, to provide a complementary cellular coverage of the regional/national territory NOTE: It can not exist independently of the SC it is associated to. Also called ATC in the US. hybrid frequency: frequency used by a terrestrial transmitter carrying the Common content NOTE: A Non-Hybrid frequency carries only Local content. Inter-Burst FEC (IFEC): is a new FEC at link layer standardized in DVB-SH to combat erasures of several complete Bursts NOTE: It usually operates over protection periods of 10 s or more. It uses the syntax of the conventional MPE-FEC (see EN 301 192 [9]), but extends this technique with spreading and interleaving, to achieve the stated improved protection. late/early decoding: "Late decoding" refers to the technique of buffering the incoming channel data so that delivery to the source decoders occurs only when all data has been received that ensure maximum FEC protection NOTE: On the contrary, "Early decoding" refers to the technique of delivering data to the source decoders before all data has been received, for the purpose of improving zapping time. This introduces jitter, in adverse reception conditions, that has to be taken into account by the presentation layer. Link Layer FEC (LL-FEC): FEC implemented at the Link layer. It usually includes a time interleaver and relies on independent framing and error detection mechanisms to correct long strings of erased bits NOTE: In DVB-SH, link-layer FEC are constructed using the MPE syntax. local content: content not available via the SC and the not guaranteed to be available in every cells of the CGC modulation: process of varying a sine wave to convey a message ETSI 22 ETSI TS 102 584 V1.1.2 (2010-XY) NOTE: Four digital modulations are used in DVB-SH: QPSK, 8PSK(TDM only), 16APSK(TDM only) and 16QAM(OFDM only). OFDM (waveform): waveform composed of a large number of closely packed sine waves each modulated at low speed NOTE: This waveform is optimized for terrestrial propagation, but can also be used with satellite. This waveform allows SFN operation by several transmitters. partially available Transport Stream: ransport Stream that carries at least one DVB service that is not transmitted in all cells where the TS is transmitted NOTE: This is signalled by the service_availability_descriptor in the SDT. receiver: functional module of a terminal from antenna(s) to IP baseband output. TS 102 585 [2] currently specifies two classes of receivers, according to their capability and methods to process time diversity elements in the waveform (see clause 10) Satellite Component (SC): satellite leg of a hybrid network, using a portion of the frequency allocated to satellite services at international level, and which appears as an umbrella cell to the DVB-SH receivers sub-band: contiguous segment of the allocated spectrum usually dedicated to the transmission of a multiplex NOTE: In DVB-SH, sub-bands can have 1,7 MHz; 5 MHz; 6 MHz; 7 MHz and 8 MHz of bandwidth. TDM (waveform): waveform composed of a sine wave modulated at high speed and with a modulation designed to produce a low temporal envelop variation NOTE: This waveform optimized for use with satellite. It does not allow SFN operations. terminal: device comprising several functional modules: receiver, video/audio processing, battery/power, display and optionally (non-DVB-SH) bidirectional radio functions NOTE: For network planning, several terminal categories are distinguished in DVB-SH (see clause 10). time interleaver: physical layer operation introduced for mitigating long fades NOTE: At the transmitter it breaks one turbo code word into several pieces (also called Interleaver Units) and spread these over time (from several hundred ms to more than 10 s). In the receiver, these interleaver units are collected, reordered and forwarded to the turbo decoder. Transport Stream (TS): multiplex based on the MPEG2 packet protocol and complete with its EN 300 468 [24] compliant SI signalling NOTE: A TS has a unique identification through the Transport_Stream_ID and the Original_ network_ID. By extension, two multiplexes having the same unique identification but transmitted in parallel in the SC and the CGC are considered in DVB-SH as forming a unique TS. waveform: shape and form of a transmitted signal. DVB-SH has defined two waveforms: TDM and OFDM 3.2 Abbreviations For the purposes of the present document, the following abbreviations apply: 16APSK 16-state APSK 16QAM please change to 16-QAM 3G Third Generation 3GPP Third Generation Partnership Program 8PSK please change to 8-PSK A3GH Above 3 GHz (frequency band) ACLR Adjacent Channel Leakage Ratio ACP Adjacent Channel Power ACS Adjacent Channel Selectivity ADT Application Data Table AM/PM Amplitude Modulation/ Phase Modulation APSK Amplitude Phase Shift Keying ETSI 23 ETSI TS 102 584 V1.1.2 (2010-XY) ASI Asynchronous Serial Interface ATC Ancillary Terrestrial Component (US) AVC Advanced Video Coding AWGN Additive White Gaussian Noise B3GH Below 3 GHz (frequency band) BER Bit Error Rate BFN Beam Forming Network BS Base Station BpS BitsPerSymbol (= modulation order) BW Bandwidth C/I Carrier-to-Interference ratio C/M Carrier over Multipath ratio C/N Carrier-to-Noise ratio CBMS Convergence of Broadcast and Mobile Systems (DVB specifications) CCSR Call completion success rate CDF Cumulative Distribution Function CEPT Conference Européenne des Postes et Telecommunications CGC Complementary Ground Component CMOS Complementary Metal-Oxide Semiconductor COST COperation europ ne dans le domaine de la recherche Scientifique et Technique CRC16 16-bit CRC CSSR Call set-up success rate CU Capacity Unit (DVB-SH) CW Code-Word, or Continuous Wave, depending on context DA-MLE Data Aided Maximum Likelihood Estimation DANL Displayed Average Noise Level dBic (Gain) in dB relative to Isotropic with Circular polarization dBW Decibel referenced to 1Watt DFL DataField Length DTT Digital Terrestrial Transmission DU Dense Urban DVB-GBS DVB Generic data Broadcasting & Service information protocols DVB-S Digital Video Broadcasting Satellite standard DVB-S2 Digital Video Broadcasting Satellite standard version 2 DVB-SH-IG DVB-SH Implementation Guidelines DVB-T/H This is a short form for « either DVB-T or DVB-H” ECC European Communications Committee EIRP Equivalent Isotropic Radiated Power EMI Electro Magnetic Interference ERP Effective Radiated Power ERS Empirical Road Side ESG Electronic Services Guide ESR5(20) Performance index computed based on ESR5 ESR5 Errored-Second-Ratio at % in 20 s (ITU) FCC Federal Communications Commission (US) FDT FEC Data Table FEC Forward Error Correction FER Frame Error Rate FFT Fast Fourier Transform G/T Receiver figure of merit GBBF Ground Based Beam Forming GBS should be DVB-GBS GEO Geosynchronous or Geostationary (orbit) GI Guard Interval (for OFDM) GMR1 and 2 Geo-Mobile Radio 1 and 2 (ETSI standards) GOP Group Of Pictures GPS Global Positioning System GSE Generic Stream Encapsulation (IETF) GSM Global System for Mobile Communications HEO Highly Elliptical Orbit HM Hierarchical Modulation ETSI 24 ETSI TS 102 584 V1.1.2 (2010-XY) HP High Priority HPA High Power Amplifier Hyb Hybrid IBO Input Back Off iFDT Interburst FEC Data Table IFEC Should be always MPE-IFEC IMT-2000 International Mobile Telecommunications-2000 IMUX Input Multiplexer INT IP Notification Table (DVB-GBS) IPDC IP DataCasting (DVB specifications) ISI Inter Symbol Interference ITS should be LMS-ITS ITU International Telecommunication Union IU (Time) Interleaver Unit (DVB-SH) Ka-band Fequency band (defined by ITU) in the range of 18-30 GHz KIP Key Indicator Parameter Ku-band Frequency band (defined by ITU) in the range of 10-14 GHz LEO Low-Earth Orbit LHCP Left hand circular polarisation LL-FEC Link-Layer Forward Error Correction LLR Log-Likelihood Ratio LMS Land Mobile Satellite (channel) LMS-ITS LMS- Intermediate Tree Shadowing LMS-SU LMS- SubUrban LNA Low Noise Amplifier LNB Low noise block converter LOS Line Of Sight LP Low Priority MAC Medium Access Control (layer) MBRAI Mobile Radio Access Interface (specifications) MCP Multi-Carrier Power MEO Medium Earth Orbit MER Modulation Error Ratio MFER see DVB-H implementation guidelines MFN Multi Frequency Network MIP Megaframe Initialisation Packet MPA Multiport Amplifier MPE Multi Protocol Encapsulation MPE-FEC see DVB-H implementation guidelines MPEG TS please change to MPEG-2 TS MPEG2 TS please change to MPEG-2 TS MPEG2 please change to MPEG-2 MPEG-TS please change to MPEG-2 TS MPE-IFEC MPE Inter-burst Forward Error Correction (DVB-SH) MRP Milestone Review Process MSB Most Significant Bit msec millisecond MSS Mobile Satellite System NF Noise Figure NGSO Non Geostationnary Satellite Orbit NIT Network Identification Table (DVB) nm Nano meter NPR Noise Power Ratio OBO Output Back-Off OFDM Orthogonal Frequency Division Multiple Access OFDM/TDM OFDM or TDM PAT/PMT Program Association Table/ Program Map Table PDA Personal Digital Assistant PDF Probability Density Function PDP Power-Delay Profile PFD Power Flux Density PHY Physical Layer ETSI 25 ETSI TS 102 584 V1.1.2 (2010-XY) PID Program Identifier (MPEG) PIPO Pedestrian Outdoor and Pedestrian Indoor PL SLOTS Physical Layer Slots PL Physical Layer Pps (one) pulse-per-second PSD Power Spectrum Density PSI/SI Program Specific Information / Service Information PVR Personal Video Recording QAM Quadrature Amplitude Modulation QoS Quality of Service QPSK Quadrature Phase Shift Keying RBW Resolution Bandwidth RF Radio Frequency RHCP Right Hand Circular Polarisation RNPT Radio Network Planning Tool RP Repetition Period RS Reed-Solomon (FEC) RSC Recursive Systematic Convolutional Rx Receive Sat Satellite SAT Satellite SC Satellite Component (DVB-SH) SDARS Satellite Digital Audio Radio System (US) SDT Service Description Table (DVB) SFN Single Frequency Network SGC should be CGC SH Satellite to Handheld (specification) SH-A SH- waveform configuration A (OFDM/OFDM) SHA Should be SH-A SH-B SH- waveform configuration B (TDM/OFDM) SHB Should be SH-B SH-F Same as SH-FRAME SH-FRAME A baseband frame defined in EN 302 584 SHIP SH frame Information Packet (DVB-SH) SHWav To be changed to EN 302 584 SI Service Information (DVB) SLA Service Level Agreement SMS Short Messages Service SNIR Signal to Noise + Interference Ratio SNORE Signal to Noise Ratio estimator SNR Signal-to-Noise Ratio SRRC Square-Root Raised-Cosine STS Synchronisation Time Stamp SU Sub-Urban S-UMTS Satellite UMTS SVC Scalable Video Coding TDL Tapped Delay Line TDM Time-Division Multiplex TDMA Time Division Multiple Access TER Terrestrial TMA Tower Mounted Amplifier TPS Transmission Parameter Signalling TR(x) Terrestrial Repeater, type x (DVB-SH) TS PER Transport Stream Packet Error Rate TS Transport Stream TU6 Typical Urban channel with 6 taps TWTA Travelling-Wave Tube Amplifier UEP Unequal Error Protection U Urban UMTS Universal Mobile Telecommunications System VBR Variable Bit Rate (video) VoD Video on Demand ETSI 26 ETSI TS 102 584 V1.1.2 (2010-XY) VSWR Voltage Standing Wave Ratio xDSL x type of Digital Subscriber Line XML eXtensible Markup Language w.r.t. With respect to WER Word Error Rate 4 System outline 4.1 DVB-SH system capability outline DVB-SH system provides an efficient and flexible mean of carrying broadcast services over an hybrid satellite and terrestrial infrastructure operating at frequencies below 3 GHz to a variety of portable, mobile and fixed terminals having compact antennas with very limited or no directivity. Target terminals include handheld defined as light-weight and battery-powered apparatus (e.g. PDAs, mobile phones), vehicle-mounted, nomadic (e.g. laptops, palmtops, etc.) and stationary terminals. The broadcast services encompass streaming services such as television, radio programs as well as download services enabling for example Personal Video Recorder services. The DVB-SH system coverage is obtained by combining a Satellite Component (SC) and, where necessary, a Complementary Ground Component (CGC) to ensure service continuity in areas where the satellite alone can not provide the required QoS. The SC ensures wide area coverage while the CGC provides cellular-type coverage. All types of environment (outdoor, indoor, urban, suburban and rural) can then be served. It should be noted that the area served by a beam of currently planned multibeam satellites is in the order of 600 000 Km2. DVB-SH can provide the following theoretical total capacity per 5 MHz (typical) satellite bandwidth and per beam. Table 4.1: DVB-SH typical and maximum net bit rates (Mbps) in Satellite-only coverage (SAT) and in Terrestrial coverage (TER) Waveform SH-A SH-B Hybrid network frequency configuration SFN MFN MFN Typ Max Typ Max Typ Max Multibeam satellite SAT/beam 2,5 10,0 2,5 10,0 2,7 10,6 system with 3x5 MHz TER/beam 10,0 30,0 7,5 20,0 7,4 20,5 spectrum assigned, 5 MHz allocated to each satellite beam in a 3-color reuse pattern Multibeam satellite SAT/beam 2,5 10,0 2,5 10,0 2,7 10,6 system with 4x5 MHz TER/beam 13,7 40,0 11,2 30,0 11,1 30,43 spectrum assigned, 5 MHz allocated to each satellite beam in a 4-color reuse pattern NOTE 1: Values are indicative as different optimizations allowed by the standard lead to slightly different results. NOTE 2: The TER capacity includes the repetition of the SAT capacity. The total system capacity for a multibeam system is obtained by multiplying each number above by the number of beams. For example a hypothetical 9-beam satellite with 20 MHz bandwidth can offer, in terrestrial coverage typically 100 Mbps total (MFN/SH-B/Typ) and up to 360 Mbps total (SFN/SH-A/Max) when there is no power constraint. The same satellite can offer, in satellite-only coverage typically 24 Mbps total (MFN/SH-B/Typ) and up to 96 Mbps total (MFN/SH-B/Max) when there is no power constraint. Currently planned multibeam satellite systems target a minimum of 6,6 Mbps total per 5 MHz of satellite bandwidth per beam (of which 2,2 Mbps capacity when in satellite-only reception conditions). Higher total capacity can be achieved through more satellite frequency-reuse, more powerful satellites, more CGC transmitters density or more advanced terminal technologies (e.g. antenna diversity). ETSI 27 ETSI TS 102 584 V1.1.2 (2010-XY) 4.1.1 System overview A typical DVB-SH system (see figure 4.1) is based on a hybrid architecture combining a SC and, where necessary, a CGC consisting of terrestrial repeaters fed by a broadcast distribution network of various kinds (satellite [DVB-S/S2], terrestrial [fibre, xDSL, etc.]). Three kinds of terrestrial repeaters are envisaged: • TR(a) are broadcast infrastructure transmitters which complement reception in areas where satellite reception is difficult, especially in urban areas; they may be collocated with mobile cell sites or standalone. Local content insertion at that level is possible, relying on adequate radio frequency planning and/or waveform optimizations. • TR(b) are personal gap-fillers of limited coverage providing local re-transmission, on-frequency and/or with frequency conversion; typical application is indoor coverage provision, locally repeating the satellite signal available outdoor. No local content insertion is foreseen. • TR(c) are mobile broadcast infrastructure transmitters creating a "moving complementary infrastructure" on board moving platforms (cars, trains, bus). Depending on waveform configuration and radio frequency planning, local content insertion may be possible. Direct-to-mobile DVB-SH satellite DVB-SH signal Content Distribution signal Service & Network DVB-SH Head-end broadcast head-end TR(b) personal gap fillers TDM/OFDM OFDM TR(c) transmitters OFDM TR(a) transmitters Broadcast Distribution Network Reception cell Figure 4.1: Overall DVB-SH transmission system reference architecture The DVB-SH transmission system key features are the following: • seamless service continuity between SC and CGC coverage; • support of all reception conditions associated to portable and mobile terminals: indoor/outdoor, urban/suburban/rural, static/mobile conditions. Typical mobility conditions covers pedestrian as well as land vehicular scenarios; • possible implementation of power saving schemes to minimize the power consumption of battery activated terminals in order to maximize autonomy; • local insertion of broadcast services on CGC; • use different kinds of distribution network to feed the CGC repeaters, such as Satellite (DVB-S/S2) and/or terrestrial (Optical fibre, Wireless Local Loop, xDSL, etc.) resources; ETSI 28 ETSI TS 102 584 V1.1.2 (2010-XY) • possible offering of interactive services by inter-working with mobile or wireless systems at service, network head-ends and user terminal levels. Two main different physical layer configurations are supported by the DVB-SH waveform standard [1]: • SH-A exploiting OFDM transmission mode for both SC and CGC. SH-A allows (but does not impose) a Single Frequency Network (SFN) between the SC signal and the CGC signal carrying the same content. If SFN is implemented then FEC, Channel Interleaver, modulation and Guard Interval cannot be optimized separately for the SC and CGC (by definition of SFN). If it is desirable to optimize these parameters separately for the SC and the CGC, a distinct frequency channel for the SC and the CGC can also be used in SH-A, leading to reduced spectrum efficiency and some handover complications for receivers having only one RF front-end. If the receiver has two RF-front-ends, soft combining (see clause 4.2.6) and easier handover (see clause 9), can also be implemented with SH-A. • SH-B exploiting TDM transmission mode for the SC and OFDM transmission mode for the CGC. SH-B requires a distinct frequency band for the SC and the CGC since they transmit signals based on two different physical layers. The impacts on the system frequency plan for this physical layer configuration are discussed in the next clause. Each component of the transmission system can be optimized separately to its respective transmission path. The OFDM waveform is known to exhibit a larger peak-to-average signal envelope fluctuation compared to the TDM waveform. Therefore, SH-A is in general recommended for spectrum limited systems while SH-B is of interest in power limited satellite systems. However, the final choice depends on the detailed architecture of the payload (e.g. single or multiple amplifiers per beam, number of carriers/beam etc..). Differently from SH-A in SFN configuration, SH-A in non-SFN configuration and SH-B also allow an independent optimization of the waveform and LL-FEC parameters for the SC and CGC. 4.1.2 Frequency planning aspects Being a hybrid satellite/terrestrial system, the frequency planning for DVB-SH is typically more complex than for a pure terrestrial system or a pure satellite system. In fact due to the large geographical coverage provided by the satellite and the possible linguistic segmentation of the market, multibeam satellite may be used in certain areas (e.g. Europe). The use of a multibeam satellite with linguistic areas and/or nation-wide broadcasting requires the segmentation of the DVB-SH downlink spectrum. Let first introduce some key definition: Sub-band: elementary unit of frequency resource in which the overall satellite bandwidth is partitioned. Colour scheme: it represents the number of colours Nc adopted for allowing frequency reuse in the system. The colours can be provided by different frequency bands Frequency reuse factor fR: it represents the number of times the same frequency sub-band is reused in the system thanks to antenna beams spatial or cross-polar isolation. If the number of beams, N b , is large enough (e.g. greater than 3), then frequency reuse among satellite beams can be exploited. The multibeam solution brings the following advantages: a) higher satellite antenna gain providing higher user terminal C/N for the same in-orbit RF power. The higher receiver C/N translates in higher spectral efficiency and/or link margin; b) frequency reuse between non-adjacent beams in the SC (independently of the existence of the CGC); c) frequency reuse between the SC and CGC. To facilitate understanding we will assume in the subsequent that the system is using a single polarization and that the bandwidth allocated for the satellite transmission is the same for each beam and denote this unit of resource as a frequency "sub-band". Then the overall satellite downlink spectrum is partitioned into Nc sub-bands, with Nc =1 for a single-beam and typically, Nc =3 for a multibeam as shown in figure 4.2. The same sub-band can be reused only by non adjacent satellite beams with a certain reuse pattern that depends on the antenna design. Typically a three-color frequency reuse scheme is sufficient to overcome interference between adjacent beams. The entire available spectrum (FG in the global beam case) is then split into 3 frequency sub-bands denoted FA, FB and FC. It should be kept in mind that, in the most general case colour scheme may require up to (but no greater than) four colours. ETSI 29 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 4.2:Example of satellite single and multibeam The apparent drawback due to satellite spectrum segmentation is compensated by the fact that the satellite power can be focused over the wanted (and smaller) regions in a multibeam scenario. Furthermore, the global beam scenario would require broadcasting the whole multilingual multiplex to the global footprint, wasting of bandwidth and power in areas where reception of the signal content is not required (or even forbidden). In the more general case also polarization (Left Hand Circular (LHC) and Right Hand Circular (RHC)) may be used to create isolation or frequency reuse in the system. Looking at the following figure one can distinguish four different cases for Nb =6 beams European coverage: a) There are Nc =3 colours obtained dividing the system spectrum in 3 different sub-bands. The antenna polarization is alternated for the three sub-bands with no specific advantage obtained by using two different polarizations. There is an overall frequency reuse fR=2 as each sub-band is reused twice over the coverage region. b) There are Nc =3 colours obtained dividing the system spectrum in 3 different sub-bands each with different polarisation. The antenna polarization is alternated for the three sub-bands with different polarization patters for the first and the second triplet of beams. This polarization reuse scheme allows to increase the co-polar beam isolation compared to case a) by an amount equal to the cross-polar isolation factor. There is an overall frequency reuse fR=2 as each sub-band is reused twice over the coverage region. c) There are Nc =2 colours obtained dividing the system spectrum in 2 different sub-bands each with different polarisations. The antenna polarization is alternated for the two sub-bands with different polarization patters for the first and the second pair of beams. Same inversion of polarisation pattern is adopted for the second and the third couple of beams. This colour scheme allows to increase the frequency reuse scheme compared to cases a) and b) at the expenses of a lower antenna co-polar beam isolation. There is an overall frequency reuse fR=3 as each sub-band is reused three times over the coverage region. d) There are Nc =3 colours obtained dividing the system spectrum in 3 different sub-bands each using the two possible polarisations in each beam. This polarization reuse scheme allows to double the bandiwidth associated to each beam compared to cases a) and b) while obtaining a beam co-polar isolation slightly lower to case a) due to the extra cross-polar co-frequency other beams interference. There is an overall frequency reuse fR=4 as each sub-band is reused four times over the coverage region. ETSI 30 ETSI TS 102 584 V1.1.2 (2010-XY) France UK France Germany IT Germany Iberian Poland Italy Iberian Poland UK a) c) France UK Germany France Germany UK France Germany UK Iberian Poland Italy Poland Iberian Italy Poland Iberian Italy b) d) Figure 4.3: Examples of satellite colour schemes and frequency reuse for a 6 beams ( N b = 6 ) European coverage The resulting spectral efficiency improvement factor is given by: Nb γ ( Nb , N c ) = fR Nc It is observed that for Nb=6 beams there is: • A four-fold spectral efficiency improvement compared to the global beam case for cases a) and b); • A nine-fold spectral efficiency improvement compared to the global beam case for case c); • An eight-fold spectral efficiency improvement compared to the global beam case for case d). In conclusion although more complex from the satellite payload perspective the colour and frequency reuse scheme d) appears a good solution in terms of spectral efficiency and co-polar beam isolation. The system frequency plan and spectrum efficiency are also impacted by the decision to use or not the SFN technique to complement the satellite coverage with the CGC coverage. If SFN is used by the CGC to complement the satellite coverage, the common content is transmitted by the CGC using the same sub-band as the satellite signal. By definition of SFN, no additional local content can be added in this sub- band. However, within each beam, the remaining N c − 1 sub-bands, which cannot be used by satellite due to adjacent beams, may be used by the CGC to carry local content. Alternatively, if SFN is not used, the spectrum must be divided into at least 2 sub-bands: one for the direct-to-user transmission from satellite and the other for the complementary and indirect transmission from the CGC. In this case, this second sub-band may be used to carry additional local content by the CGC since the SFN constraint does not apply. If N c > 2 , the remaining N c − 2 sub-bands may be used by the CGC to carry additional local. When planning additional local content transmission it should be kept in mind that strong interference from satellite into terrestrial may occur in quite large exclusion zones (see for example, in figure 4.2, the overlap of the beam centred on ETSI 31 ETSI TS 102 584 V1.1.2 (2010-XY) France to the south of UK territory). This means that, for sub-bands not reserved for satellite in each beam, the terrestrial transmission parameters and the corresponding terrestrial coverage/capacity cannot be identical within a territory and must be carefully engineered, in coordination with the satellite operator. If Multi Frequency Network (MFN) is used, different physical-layer configurations will be possible in different areas of a territory (e.g. higher spectral efficiencies modulation and FEC and/or guard-times and/or hierarchical modulation). Therefore MFN allows optimization for different transmitter densities, for different terrestrial coverage and different capacity requirements. In SFN instead, the terrestrial signal complementing the satellite signal has to use exactly and uniformly the common configuration with the satellite signal. For the various reasons above DVB-SH does not impose the use of SFN for SH-A. 4.1.3 Other considerations for spectrum efficiency A satellite customer traditionally leases a satellite beam with an associated satellite frequency sub-band. To convert the leased resource into a program offer, he needs a measure of system performance called spectrum efficiency which is given in bits/s/Hz. For example an efficiency of 0,5 bits/s/Hz means that, using 5 MHz satellite bandwidth the customer can build a bouquet of programs with data rates summing up to 2,5 Mbits/s (8 to 10 Mobile TV programs). In DVB-SH, this customer also gets a frequency "bonus" which is the frequency resource available for potential use terrestrially. This by-product is a virtual asset which becomes a real resource when (i) the customer obtains the authorization from his government authority to use this spectrum terrestrially, and (ii) he deploys/leases the appropriate CGC infrastructure. Therefore in DVB-SH, three spectrum efficiencies can be distinguished, all normalized to the leased satellite bandwidth: • The η[COM ] is the spectrum efficiency for the "common" programs, per Hz of satellite bandwidth in a beam. • The η[ LOC ] is the spectrum efficiency for the "local" programs, per Hz of satellite bandwidth in a beam. • The η[TOT ] , or system spectrum efficiency, is the sum of the above efficiency parameters. The η[TOT ] gives an indication of the total programs offer that can be provided with satellite and terrestrial transmissions combined, normalized to the leased satellite bandwidth. Note that in the "exclusion zones", coexistence of terrestrial transmissions with satellite transmissions in adjacent beams can impose lower η[ LOC ] . Also, in MFN between satellite and CGC, the OFDM guard-time can be engineered differently from one local zone to another, resulting in different η[ LOC ] . Therefore η[ LOC ] and consequently η[TOT ] should be used as local performance indexes and not a nation-wide index. To evaluate and/or optimize DVB-SH system spectral efficiency as defined above, first the following must be determined/computed. ETSI 32 ETSI TS 102 584 V1.1.2 (2010-XY) Table 4.2 Parameter/ Explanation Main Dependencies General target Questions values N beams The number of satellite beams Number of countries/linguistic >=3 (for regional regions, satellite antenna sub- coverage) system cost, inter-beam (for large countries, interferences Nbeams = 1) fR Satellite frequency and/or polarization N beams , national/ linguistic 3 reuse over the beams (also called number of "colours") boundaries, satellite power, frequency, polarization allocation and channelization. Are there Local content may be forbidden or its Regulatory context Country dependent regulatory amount restricted. limitations for local content? Is SFN used by For SH-B, the answer is negative by SH-A/SH-B, handover Equally likely CGC to definition. complexity, chipset availability complement For SH-A it is possible to choose satellite? between SFN and MFN Does the satellite In general yes with different levels of System and payload design. Equally likely allow nonlinear operating HPA OBO depending on Market definition operation of selected waveform (SH-A or SH-B) or TWTA? payload architecture (single HPA/beam or shared HPA/beam). ηOFDM [ SAT ] Spectral efficiency of the OFDM Satellite effective EIRP, Terminal 0,5 b/s/Hz waveform when used by satellite, characteristics, required link given minimum terminal margin (QoS.), allowed end-to- characteristics and Quality of Service. end max delay. ηTDM [ SAT ] Spectral efficiency of the TDM Satellite effective EIRP, Terminal 0,5 b/s/Hz waveform when used by satellite, characteristics, required link given minimum terminal margin (QoS.), allowed end-to- characteristics and Quality of Service. end max delay. ηOFDM [TER ] Spectral efficiency of the OFDM CGC network cost, indoor- 0,7 b/s/Hz waveform when used by CGC, given outdoor operations, terminal minimum terminal characteristics, types, required link margin coverage type (indoor/outdoor) and (QoS.) Quality of Service. Total payload Total payload losses due to payload Payload characteristics, physical 1 dB to 1,5 dB for equivalent signal nonlinear effects (output back off, layer settings, required margin single carrier TWTA power signal distortion, intermodulation (QoS), interference scenario operation, degradation noise) 2,5 dB to 3,5 dB for DTOT multi-carrier TWTA operation Several frequency plans are possible depending on the SH-A or SH-B architecture and SFN or MFN configurations. Clause C.1 provides examples and a methodology for spectral efficiency computations and comparisons. 4.1.4 Frequency bands and their regulatory aspects 4.1.4.1 Frequency Allocations and coordination procedures Frequency spectrum is regulated at the international level by a binding treaty called the Radio Regulations. The Radio Regulations deal with two aspects: frequency allocation and regulatory procedures for accessing the spectrum/orbit resources. The binding character of the Radio Regulations implies that national or regional (e.g. European) regulations shall be defined within the framework of this treaty. Frequencies between 1 GHz and 3 GHz are the most suitable, considering the satellite, terminal and mobility constraints. Within this frequency range, the following bands are candidates for the provision of multimedia services based on DVB-SH (the uplink bands are given for completeness and their use are out of the scope of the DVB-SH specifications). ETSI 33 ETSI TS 102 584 V1.1.2 (2010-XY) Table 4.3 Frequency band Others common Frequency range designation names 1 626,5 MHz to 1 660,5 MHz, 1 668 MHz to 1 675 MHz (uplink) MSS GEO 1 518 MHz to 1 559 MHz (downlink) L band L band MSS Big LEO 1 610 MHz to 1 626,5 MHz (up and downlink) L band 1 452 MHz to 1 492 MHz (downlink) S-DAB band 1 980 MHz to 2 010 MHz (uplink) 2 GHz S band MSS S-Band 2 170 MHz to 2 200 MHz (downlink) S-DARS S band 2 320 MHz to 2 345 MHz (downlink) 2 670 MHz to 2 690 MHz (uplink) 2,5 GHz S band 2 500 MHz to 2 520 MHz (downlink) 2 520 MHz to 2 670 MHz (downlink) 4.1.4.1.1 L band Frequency allocations In the field of satellite communications, the term "L band" encompasses three frequency bands: • the bands 1 626,5 MHz to 1 660,5 MHz and 1 668 MHz to 1 675 MHz (uplink)/1 518 MHz to 1 559 MHz (downlink): these bands are allocated almost worldwide to the mobile-satellite service and are mainly used by incumbent MSS operators using geostationary satellites to provide two-ways mobile satellite communications; • the band 1 610 MHz to 1 626,5 MHz (up and downlink): this band is allocated worldwide to the mobile-satellite service and is used by incumbent MSS operators using non-geostationary satellite systems in a low Earth orbit (LEO) to provide the same type of services as in the previous bands. It should be noted that one non-geostationary system uses this band only for its uplinks (its downlink using the band 2 483,5 MHz to 2 500 MHz); • the band 1 452 MHz to 1 492 MHz (downlink): this band is allocated worldwide (with the only exception of the USA) to the broadcasting-satellite service but only the upper 25 MHz (i.e. 1 467 MHz to 1 492 MHz) can actually be used since the lower 15 MHz are not usable until a decision by a future ITU World Radiocommunications Conference. This band is also allocated on a co-primary basis with the broadcasting-satellite service to the terrestrial fixed, mobile and broadcasting services. The use of this band by the broadcasting-satellite and terrestrial broadcasting services is limited to digital audio broadcasting. Access to the spectrum/orbit resources In the bands 1 518 MHz to 1 559 MHz, 1 610 MHz to 1 626,5 MHz, 1 626,5 MHz to 1 660,5 MHz and 1 668 MHz to 1 675 MHz, geostationary satellites and non-geostationary systems are on an equal footing with regards to spectrum access (see provision No. 9.11A of the Radio Regulations [20]). The basic principle governing the coordination under No. 9.11A [20] is "first come, first served". In the band 1 467 MHz to 1 492 MHz (since the band 1 452 MHz to 1467 MHz is de facto unusable), geostationary satellites have regulatory priority with regards to non-geostationary systems (see provision No. 22.2 of the Radio Regulations [20]). Among themselves, geostationary satellites are required to coordinate according to Radio Regulations provision No. 9.7 [20] ("first come, first served" but only between geostationary satellites). Moreover, all systems are required to coordinate with terrestrial services under provision No. 9.11 [20]. It should be noted that the protection of terrestrial services in the United States of America renders operations of a geostationary satellite in visibility of the US territory very challenging if not unfeasible. ETSI 34 ETSI TS 102 584 V1.1.2 (2010-XY) 4.1.4.1.2 2 GHz S band (1 980 MHz to 2 010 MHz/2 170 MHz to 2 200 MHz) Frequency allocations At the international level, the band 2 170 MHz to 2 200 MHz is allocated worldwide to the fixed, mobile and mobile-satellite (space-to-Earth) services on a co-primary basis. It means that, from a regulatory point of view, these three services enjoy the same status. It does not mean that applications provided through these services are necessarily compatible from a technical point of view (i.e. that they can use the same band without interfering each other). On the contrary, numerous studies have shown that terrestrial mobile applications and satellite mobile applications are not compatible if they are operated independently in the same geographical area. Complementary ground components (CGC) are compatible with the provision of satellite applications to mobile terminals since they are operated in a coordinated manner with the satellite system. It should be noted that in the Americas, the adjacent band 2 160 MHz to 2 170 MHz is also allocated, inter alia, to the mobile-satellite service (space-to-Earth). Some countries have chosen to favour terrestrial systems in these bands: in Algeria, Benin, Cape Verde, Egypt, Iran, Mali, Syria and Tunisia, the use of these bands by the mobile-satellite service shall neither cause harmful interference to the fixed and mobile services, nor hamper the development of those services prior to 1 January 2005, nor shall the former service request protection from the latter services. While this has not a direct impact on the overall design and operations of MSS systems intended to cover Europe, it may have an impact on the design of satellite antennas compatible with the protection of terrestrial services in these bands. The fact that a country is not listed above does not imply that it intends to give priority to satellite services but simply that it has chosen to keep satellite and terrestrial services on an equal footing from the point of view of international regulations. National regulators may chose to restrict nationally the use of these bands to a specific service. For more information on the regulatory status of these bands, the national table of frequency allocations shall be consulted. NOTE: National tables of frequency allocations of most of European countries can be accessed in: www.efis.dk. Access to the spectrum/orbit resources at the international level In these bands, geostationary satellites and non-geostationary systems are on an equal footing with regards to spectrum access (see provision No. 9.11A of the Radio Regulations [20]). The basic principle governing the coordination under No. 9.11A [20] is "first come, first served". Moreover, in the band 2 170 MHz to 2200 MHz, a coordination procedure between transmitting space stations and terrestrial services is established through the provision No. 9.14 of the Radio Regulations [20]. The PFD thresholds used to determine whether there is a need to coordinate with terrestrial services in any country are: Table 4.4 For geostationary satellites For non-geostationary satellites For 0°  δ  5° -128 dBW/m²/MHz -123 dBW/m²/MHz For 5°  δ  25° -128 + 0,5( δ - 5) dBW/m²/MHz -123 + 0,5(δ - 5) dBW/m²/MHz For 25°  δ  90° -118 dBW/m²/MHz -113 dBW/m²/MHz NOTE: where δ is the angle of arrival (degrees). It should be noted that these pfd levels are not compatible with MSS operations towards handheld terminals. Therefore this coordination threshold will always be exceeded over the territories of countries within the service area of the MSS system. It will not be a constraint as such, since it is widely recognized that such systems can not share the same geographical area with terrestrial services. Therefore countries willing to encourage the provision of MSS services in these bands over their territory have taken steps to free the bands from any terrestrial systems. However, the same does not apply in countries not belonging to the MSS service area. On their territories, these administrations may wish to continue to deploy terrestrial services as planned and may require protection from MSS transmitting satellites. This request for protection has to be explicit: following the publication of the MSS system special section by the ITU, any administration considering itself as potentially affected has four months to explicitly request to be included in the coordination process. Otherwise, it is deemed to be unaffected. This mechanism substantially reduces the number of administrations actually included in the process of coordination with respect to their terrestrial services. ETSI 35 ETSI TS 102 584 V1.1.2 (2010-XY) 4.1.4.1.3 Sound Broadcasting Satellite (SDARS) S band (2 310 MHz to 2 360 MHz) The band 2 310 MHz to 2 360 MHz is allocated to the sound broadcasting-satellite service in only three countries: India, Mexico and the USA. Only in the latter is the band currently used for such service. More particularly, only 25 MHz are available in the USA for such a service (2 320 MHz to 2 345 MHz). This 25 MHz band was divided into two blocks of 12,5 MHz and auctioned by the FCC in April 1997. At the time of writing these guidelines, the two blocks were fully used by their respective owners for the provision of digital radio via satellite supplemented by ancillary ground repeaters in areas where the satellite signal is too weak to be received with the required quality. These ground repeaters are using part of the frequencies available in the 12,5 MHz blocks. 4.1.4.1.4 2,5 GHz S band (2 500 MHz to 2 690 MHz) The band 2 500 MHz to 2 690 MHz includes various allocations to satellite services. Of particular interest for the introduction of DVB-SH based services are the following: • 2 500 MHz to 2 520 MHz (downlink)/2 670 MHz to 2 690 MHz (uplink): these paired bands are allocated worldwide to the mobile-satellite service (see note 1) and are shared, inter alia, with terrestrial (fixed and mobile) services. As for the bands 1 980 MHz to 2 010 MHz/2 170 MHz to 2 200 MHz (see above), coordination under No. 9.11A [20] (i.e. "first come, first served") applies in these bands and therefore geostationary satellites and non-geostationary systems are on an equal footing with regards to spectrum access. Similarly to the band 2 170 MHz to 2 200 MHz, in the band 2 500 MHz to 2 520 MHz, a coordination procedure between transmitting space stations and terrestrial services is established through the provision No. 9.14 of the Radio Regulations [20]. The pfd thresholds used to determine whether there is a need to coordinate with terrestrial services in any country are -128 MHz/-118 dBW/m²/MHz depending on the elevation angle. NOTE 1: With the exception of Azerbaijan, Bulgaria, Kyrgyzstan and Turkmenistan. • 2 520 MHz to 2 670 MHz: this band is allocated to the broadcasting-satellite service and is also shared, inter-alia, with terrestrial (fixed and mobile) services. It should be noted that the European Conference of Postal and Telecommunications administrations (CEPT) has decided in 2005 that the frequency band 2 500 MHz to 2 690 MHz is designated for terrestrial IMT-2 000/UMTS systems only (see note 2). This band is therefore not suitable for the provisions of satellite services within Europe. NOTE 2: See ECC Decision of 18 March 2005 on the harmonised utilisation of spectrum for IMT-2000/UMTS operating within the band 2 500 MHz to 2 690 MHz (ECC/DEC/(05)05 [i.25]). 4.1.4.2 Regulatory aspects 4.1.4.2.1 Europe 4.1.4.2.1.1 European Union L band and 2,5 GHz S-band: No specific regulatory instruments on the designation of the frequency bands for satellite use have been adopted by the European Union in these bands. 2 GHz S-band: On 14th February 2007, the European Commission adopted a Decision "on the harmonized use of radio spectrum in the 2 GHz frequency bands for the implementation of systems providing mobile satellite services" (see note 1 below) which gives priority to the development of MSS in the 2 GHz bands. Pursuant to the decision, Member States will be obliged to designate and make available as of 1 July 2007 the 2 GHz S-bands for systems providing mobile satellite services, which can include the use of complementary ground components. A European Decision is mandatory for all Member States of the European Union. It gives therefore more regulatory certainty than CEPT Decisions for the use of these bands. NOTE 1: This Decision 2007/98/EC [i.26] was published in the Official Journal of 14th February 2007. On 30 June 2008, the European Parliament and the Council adopted jointly a Decision on the selection and authorisation of systems providing mobile satellite services (MSS) in the 2 GHz band. Pursuant to this decision, ETSI 36 ETSI TS 102 584 V1.1.2 (2010-XY) Member States will be obliged to ensure that the selected applicants have the right to use the specific radio frequency identified in the Commission decision and the right to operate a mobile satellite system. In particular, Member States shall ensure that their competent authorities grant to the selected applicants the authorisations necessary for the provision of the complementary ground components of mobile satellite systems on their territories. In Recital 8 of the Decision, it was recalled that “Complementary ground components are an integral part of a mobile satellite system and are used, typically, to enhance the services offered via the satellite in areas where it may not be possible to retain a continuous line of sight with the satellite due to obstructions in the skyline caused by buildings and terrain.” NOTE 2: This Decision 2008/626/EC [i33] was published in the Official Journal of 2d July 2008. 4.1.4.2.1.2 European Conference of Postal and Telecommunications administrations (CEPT) L band: CEPT has decided in 2004 that the frequency bands 1 518 MHz to 1 525 MHz and 1 670 MHz to 1 675 MHz are designated to the mobile-satellite service from 1 April 2007 (see note 1 above). While no specific CEPT instruments designate the frequency bands 1 525 MHz to 1 559 MHz and 1 610 MHz to 1660,5 MHz to the mobile-satellite service, other CEPT decisions on free circulation and use or on licence exemption of terminals imply that these bands are, de facto, designated to such service. NOTE 1: See ECC Decision of 12 November 2004 on the designation of the bands 1 518 MHz to 1 525 MHz and 1 670 MHz to 1 675 MHz for the Mobile-Satellite Service (ECC/DEC/(04)09 [i.27]). CEPT has decided in 2003 that the frequency band 1 479,5 MHz to 1 492 MHz is designated for use by satellite digital audio broadcasting systems (see note 2). It should be noted that the band 1 467 MHz to 1 479,5 MHz is part of the Maastricht Plan for the introduction of terrestrial T-DAB systems in L band. The use by satellite systems of this lower band in Europe is generally not compatible with T-DAB systems. NOTE 2: See ECC Decision of 17 October 2003 on the designation of the frequency band 1 479,5 MHz to 1 492 MHz for use by Satellite Digital Audio Broadcasting (S-DAB) systems (ECC/DEC/(03)02 [i.28]). No specific regulatory instruments have been adopted by the CEPT in these bands concerning complementary ground components. 2 GHz S band: In December 2006, the European Communications Committee (ECC) adopted two Decisions and one Recommendation designed to foster the regulatory certainty of the 2 GHz bands. • Band designation for MSS systems, including those supplemented by a Complementary Ground Component (CGC) (see note 1): the bands 1 980 MHz to 2 010 MHz and 2 170 MHz to 2 200 MHz are designated for systems of the mobile-satellite service. Since no other ECC Decision exists to designate these bands for other services, it means that these bands are exclusive for MSS systems. The same Decision explicitly mentions the possibility for MSS systems in these bands to incorporate a complementary ground component (CGC). This explicit mention gives more regulatory certainty for deploying such ground-based infrastructure across the whole Europe. NOTE 1: See Decision ECC/DEC/(06)09 [i.29]. • Harmonized regulatory conditions for complementary ground components: within the same Decision, CGC is defined as follows: "CGC is an integral part of a mobile satellite system and consists of ground based stations used at fixed locations to improve the availability of the mobile satellite system in zones where the communications with one or several space stations cannot be ensured with the required quality. CGC uses the same portions of the mobile-satellite service frequency bands (1 980 MHz to 2 010 MHz/2 170 MHz to 2 200 MHz) as authorized for the associated space station(s)" Moreover, some conditions are attached to the use of CGC: - the frequency band to be used by the CGC of a particular satellite system shall be accommodated within the same frequency band used by the satellite component of that satellite system; - the use of CGC shall not increase the spectrum requirement of the satellite component of that particular mobile satellite system; ETSI 37 ETSI TS 102 584 V1.1.2 (2010-XY) - the CGC shall only be deployed in the geographical areas where the mobile earth stations of the associated MSS system are authorized to operate; - the same direction of transmission by CGC and the satellite component shall be used so as to decrease the number and complexity of compatibility issues; - the satellite segment shall be re-established as soon as possible in case of failure of the satellite segment, and no later than 18 months after such a failure, unless justified otherwise on considerations based on reasonableness and/or proportionality. Otherwise, CGC shall cease operation; - compatibility with terrestrial IMT-2 000/UMTS operational systems in adjacent bands should be ensured; - the CGC shall not operate independently from the satellite resource/network management system. • Spectrum reframing of the 2 GHz bands (see note 2): in addition to the Decision designating the 2 GHz bands for MSS, the ECC also agreed on a Decision on "transitional arrangements for the fixed service and tactical radio relay systems in the bands 1 980 MHz to 2 010 MHz and 2 170 MHz to 2 200 MHz in order to facilitate the harmonized introduction and development of systems in the mobile satellite service including those supplemented by a complementary ground component". This Decision is threefold: - finalizing the transitions of existing fixed service systems and tactical radio relay systems operating in or overlapping the 1 980 MHz to 2 010 MHz and 2 170 MHz to 2 200 MHz bands; - not implementing any new FS networks and tactical radio relay systems which are incompatible with MSS operations in the bands 1 980 MHz to 2 010 MHz and 2 170 MHz to 2 200 MHz; - not requiring any longer coordination of MSS systems with respect to fixed service and tactical radio- relay systems (see section 4.1.3.1.2, coordination under No. 9.14 [20]). NOTE 2: See Decision ECC/DEC/(06)10 [i.30]. • Recommendation on Milestone Review Process (see note 3): at its December 2006 meeting, the ECC adopted a Recommendation establishing a Milestone Review Process (MRP) in order to gather precise information about the development of the various satellite projects intending to use the bands. This ECC Recommendation lists the milestones for which MSS systems proponents are invited to submit information in order to monitor the development of projects in these bands. It was initially envisaged to use this MRP as a binding process to choose MSS systems eligible for a frequency authorization in these bands; however the lack of legal power of CEPT to enforce such a process was deterrent to its implementation. The MRP is now only of an informative nature. All declarations are based on good faith and will not be checked against. Any binding decision should be transferred to the European Commission. The milestones consists in evidence of submission of ITU request for co-ordination, satellite manufacturing contract, completion of the Critical Design Review of the MSS system, satellite launch agreement, gateway earth stations procurement, successful satellite mating, successful satellite launches, completion of frequency co-ordination as well as actual provision of satellite service within the territories of CEPT countries. NOTE 3: See Recommendation ECC/Recommendation (06)05 [i.31]. 2,5 GHz S band As mentioned above, CEPT has decided in 2005 that the frequency band 2 500 MHz to 2 690 MHz is designated for terrestrial IMT-2 000/UMTS systems only (see note). This band is therefore not suitable for the provisions of satellite services within Europe. NOTE: See ECC Decision of 18 March 2005 on the harmonised utilisation of spectrum for IMT-2 000/UMTS operating within the band 2 500 MHz to 2 690 MHz (ECC/DEC/(05)05 [i.26]). 4.1.4.2.2 North America (all bands) In the United States of America, the available bandwidth is limited by the Table of Frequency Allocations to 20 MHz (i.e. 2 180 MHz to 2 200 MHz coupled with the uplink band 2 000 MHz to 2 020 MHz). In these bands, ancillary terrestrial components may be operated in conjunction with MSS networks, subject to the Commission's rules for ancillary terrestrial components (ATC) and subject to all applicable conditions and provisions of the MSS authorization. ETSI 38 ETSI TS 102 584 V1.1.2 (2010-XY) The FCC generally imposes to an MSS licensee that wishes to include ATC to meet five requirements (see note 1 above): • geographic coverage: an MSS licensee must provide space-segment service across the entire geographic area where the ATC can be deployed; • coverage continuity: MSS operators must maintain space station coverage over the relevant geographic area, which implies timely replacement of satellites in the event coverage should degrade as a result of satellite failure. In order to implement this condition, a non-geostationary MSS system licensee is required to maintain an in-orbit spare, while a geostationary MSS system licensee is required to maintain a spare satellite on the ground within one year of commencing operations and launch it into orbit during the next commercially reasonable launch window following a satellite failure; • commercial availability: the MSS service via satellite must be commercially available as a perquisite to any offering of the ATC service; • an integrated offering: MSS licensees must offer an integrated service. MSS licensees must make an affirmative showing to the FCC that demonstrates that their ATC service offering is truly integrated with their MSS offering. As an example, MSS licensees that wish to provide ATC services could demonstrate that they use a dual-mode handset to provide the proposed ATC service; and • in-band operation: the ATC operations must remain limited to the precise frequency assignments authorized for the MSS system. NOTE 1: For more information, see Federal Communications Commission - Report and Order and Notice of Proposed Rulemaking 03-15 - In the matter of Flexibility for Delivery of Communications by Mobile Satellite Service Providers in the 2 GHz Band, the L-Band, and the 1,6/2,4 GHz Bands (released 10th February 2003). In Canada, the available bands for mobile-satellite services are 2 165 MHz to 2 200 MHz (space-to-Earth) and 1 990 MHz to 2 025 MHz (Earth-to-space). Industry Canada has issued generic regulations for Ancillary Terrestrial Components (ATC) (see note 2) that can be used in conjunction with MSS systems in the frequency range 1 GHz to 3 GHz. These regulations are based on the following principles: • the ATC mobile service will be an integral part of MSS service offerings. A substantial level of mobile-satellite services will be provided with the ATC service; • the frequencies used for the ATC system will be within the assigned spectrum for a particular MSS network and the ATC service will be limited to the satellite serving areas. The use of the MSS spectrum for ATC operation will be subordinate to the spectrum being available for mobile-satellite service; • the ATC mobile service will be required to cease operation, within a reasonable period, should the mobile-satellite service or network be discontinued; • the ATC operation will be authorized such that it will neither cause harmful interference to, nor claim protection from, MSS services and other primary radio services operating in adjacent bands. ATC operations will be subject to technical and operational requirements considered appropriate to mitigate potential interference; • complete applications as radiocommunication carriers will need to be submitted to seek authorization to operate an ATC mobile system as an integral and indefeasible part of the MSS service offerings. Specific information will be required as part of the applications to demonstrate adherence to policy, operational and regulatory principles; • spectrum area licences will be issued for ATC systems and will be subject to spectrum fees. NOTE 2: See Industry Canada - Notice No. DGTP-006-04 - Spectrum and Licensing Policy to Permit Ancillary Terrestrial Mobile Services as Part of Mobile-Satellite Service Offerings (May 2004). ETSI 39 ETSI TS 102 584 V1.1.2 (2010-XY) 4.2 Main issues addressed by DVB-SH 4.2.1 Differences in propagation characteristics between satellite and terrestrial channels For frequency below 3 GHz, the only significant effects are those caused by the environment close to the user terminal as atmospheric effects are negligible (see ITU-R Recommendations P.676 [25], ITU-R Recommendation P.618 [31], ITU-R Recommendation P.531 [32] and ITU-R Recommendation P.680 [33]). For DVB-SH, three main mobile environments may be considered: • rural environment: the propagation is mainly affected by the vegetation. The coverage is mainly provided by the SC; • urban environment: the propagation is mostly affected by dense buildings or other constructions with height of 4 storeys or more. The coverage is mainly provided by the CGC; and • suburban environment: representing an intermediate case with medium density of buildings, lower structures (2 to 3 storeys) and roads which are wider than in an urban environment. The SC and the CGC contribute to the desired coverage. Small villages may be treated as suburban areas where, if the population density is low, satellite may be the only source for service provision. For both LMS and terrestrial mobile channels, the effects are conventionally divided into three types according to the scale of distances to be considered: • path loss at large scale (very slow fluctuations): the signal suffers variations due to modifications in the geometry of the propagation path. This loss is usually assumed to be proportional to dn, d being the distance from the transmitter to the user terminal and n being an empirical exponent, based on theory and measurements. For satellite channels n=2 while for terrestrial channels typically n=4; • shadowing at mid-scale (slow fluctuations): it corresponds to amplitude variations due to nearby obstacles on the ground. The signal suffers variations due to obstructions caused by buildings, trees etc., the scale here being similar to the dimensions of these obstacles; and • multipath fading at small scale (fast fluctuations): the scale of the variations of the signal is about one wavelength, as a result of the constructive or destructive addition of multiple paths. For wideband systems, it is necessary to consider the multipath fading as frequency selective. The satellite propagation channel in DVB-SH is generally non-frequency selective. This assumption is very accurate for 1,5 MHz, pretty correct for 5 MHz and fairly correct for 8 MHz channelization. It means that the satellite propagation channel can be considered as a single complex multiplicative process. Instead the terrestrial repeaters propagation channel is considered to be frequency selective because of their respective coherence bandwidths. A frequency selective fading is classically characterized through a Power Delay Profile (PDP) which gives the relative time of arrival, the relative power and the type (Ricean or Rayleigh distribution, spectrum) of each group of unresolved echoes (also called tap). These PDP are then used to parameterize Tapped Delay Line (TDL) models. For SFN operation between satellite and terrestrial OFDM (SH-A, SFN), all contributions have to be taken into account. In summary, the LMS channel has very different characteristics compared to the terrestrial one. In particular, the channel can be considered as non frequency selective and link margins are not as large as for terrestrial networks. More specifically, the satellite distance from the user is basically constant within the beam and the satellite power limitations make the link margin bounded to a value typically ranging from 5 dB to 15 dB. 4.2.2 LMS channels (in the L and S bands) The LMS channel models can be grouped into three classes: 1) empirical models are obtained from experimental data. They are very close to reality for the environment type in which the measurements have been done but are difficult to generalize to other environment types; 2) statistical models are based on the use of canonical statistical distributions. Like empirical models, statistical models are applied to environment classes (rural, suburban, urban, etc.). The subsequent classification problem ETSI 40 ETSI TS 102 584 V1.1.2 (2010-XY) is not straightforward, since an environment classified as urban in some countries may look a little more like small town elsewhere. Therefore statistical models are also difficult to generalize; and 3) physical models rely on a deterministic modelling of the propagation phenomena (reflection, diffraction, refraction), but also of the considered environment. These models have been efficiently used for planning purposes in terrestrial radio-communication or broadcast networks. In the following the main models belonging to the two first categories are presented, the third being mainly used for terrestrial networks. 4.2.2.1 Empirical Models The ERS (Empirical Road Side) model was elaborated based on numerous experiments performed since 1983 by J.Goldhirsh and W.J.Vogel [i.2]. This model mathematically describes attenuation from tree-lined roads, multipath in mountainous and tree lined roads. It also provides a frequency scaling relationship for static tree attenuation from UHF to K Band. This model is recommended by ITU for rural environment (see ITU-R Recommendation P.681-4 [12]). Examples of ERS model link margin calculations for S-band (2,2 GHz) are reported in table 4.5. Table 4.5: Additional link margin versus elevation angle for rural landscapes according to the ERS model, at fc = 2,2 GHz Elevation angle Margin for Margin for Margin for 80 % availability 95 % availability 99 % availability [dB] [dB] [dB] 41,5° 5 12 20 39° 6 13,5 21,5 30° 9,5 17,5 26,5 25° 12,5 20 29 It can be seen that by using an ERS model, calculation based on instantaneous signal availability results in a link margin that, for availability greater or equal than 95 %, is incompatible with practical satellite capabilities. As shown in Clause A, by using the DVB-SH waveform features (powerful FEC, long time interleaving at physical layer or inter-burst upper layer FEC) one can achieve high availability with smaller margins. This is possible as short term link unavailability is recovered through physical FEC and LL-FEC. Thus the DVB-SH link margins shall be based on a more elaborated satellite channel model able to represent the first and the second order statistics of the fading/shadowing process. Therefore the ERS model is not recommended for DVB-SH network planning. 4.2.2.2 Statistical models (for narrow-band signals) The statistical channel models illustrated in the following clause are capable of generating time series of the LMS signal amplitude fading/shadowing process through a propagation model that can be readily implemented in a computer simulation or in a laboratory test environment. The statistical LMS propagation model is characterized by parameters that have been derived through synthetic time series matching with experimental data obtained in different LMS propagation environments. The experimental data matching corresponds to first and second order statistics of the signal attenuation, thus is in principle suitable to our DVB-SH performance assessment scope. 4.2.2.2.1 Single-state models Statistical models for satellite channels make the assumption that the received signal is composed of two parts: a coherent part associated with the LOS path and a diffuse part arising from multipath components. Table 4.6 summarizes the best known statistical satellite mobile channels in increasing order of complexity. Table 4.6: Statistical satellite mobile channel models Model Coherent part Correlation Diffuse part Rice Constant Zero Rayleigh Loo [i.4] Log-normal Variable Rayleigh Corazza [i.5] Log-normal Unity Log-normal, Rayleigh Hwang [i.6] Log-normal Zero Log-normal, Rayleigh ETSI 41 ETSI TS 102 584 V1.1.2 (2010-XY) The Rice model assumes that the LOS component is only affected by Rayleigh distributed multipath and is characterized by the carrier-to-multipath ratio (C/M or Rice factor K). The Loo model assumes that the LOS is affected by lognormal shadowing (typical for tree shadowing) which makes the instantaneous C/M time variant. The Corazza model assumes that both the LOS and the multipath components are affected by the same lognormal shadowing. Thus in the Corazza model the C/M is constant although both the LOS and the average multipath power components are lognormal distributed. The Hwang model also considers lognormal process affecting both the LOS and the multipath component but with total decorrelation between them. The Hwang model has been shown to include the Rice, Loo and Corazza models as special cases. As DVB-SH is supposed to operate in different mobile environments, where channel conditions are time variant, a single-state channel model is considered inadequate. 4.2.2.2.2 Two-state (Lutz) model The Lutz model [i.3] is a statistical model which represents the channel by a two states Markov chain. The "good" state occurs when a dominant line-of-sight component is received. In that state, the channel can be considered as a Ricean channel. The "bad" state occurs when no LOS component is received. In this state, the channel can be modelled as a Log-normal Rayleigh channel without coherent component. The parameters associated with each state and the transition probabilities from one state to another are empirically derived. This approach allows the generation of time series representing different environments. However, two-state channels were considered not enough to represent the variety of propagation conditions experienced in the different environments. 4.2.2.2.3 Three-state (Fontan) Model A further refinement of the Lutz model is represented by the Fontan model [16] which includes a three-state Markov chain: • state 1: line-of-sight (LOS); • state 2: moderate shadowing; and • state 3: deep shadowing. The statistical model adopted for all three states is a Rice/lognormal distribution with parameters that have been experimentally derived from propagation campaigns performed in different European locations. The model parameters are reported in [16] for different environments, elevation angles and locations. Being recognized as the most accurate statistical LMS channel model available today, encompassing the widest set of environments and elevation angles, this model has been adopted for the DVB-SH performance evaluation detailed in Clause A.7. Figure 4.4 shows a number of time series generated by using the Fontan's model for various environments and satellite elevations angles at S-band. For open area the received signal power fluctuations are rather limited except for the 20 degrees elevation. The situation is radically different for S-band intermediate tree-shadowed (ITS) environment. In this case the multi-state channel nature is evident in particular up to 40 degrees elevation. When no LOS signal is present large shadowing fluctuations on top of multipath fading are visible. Suburban areas shows similar two-state nature with longer LOS condition compared to the ITS case. Urban environment keeps a remarked on-off channel nature even at high satellite elevation angles due to the presence of close buildings. ETSI 42 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 4.4: Examples of Fontan's channel model time series 4.2.2.2.4 Quasi-static Channel Model The use of DVB-SH for direct satellite operation reception by hand-held terminals operating in quasi static conditions with user cooperation calls for an extra channel model. This channel model corresponds to the line of sight state of the Fontan model described before. Being the terminal quasi static, the resulting channel can be described by a very slow Ricean fading process (see Clause A.7). Being the fading correlation time larger than the physical layer or upper layer decorrelation capabilities, link availability can be computed according to first order fading statistics as described in clause 11. ETSI 43 ETSI TS 102 584 V1.1.2 (2010-XY) 4.2.3 Terrestrial Channels Path loss: several models have been developed to describe path loss for a single terrestrial transmitter in both urban and suburban environments in the past [i.1]. Two of them have been used in contexts close to the DVB-SH one: a) the COST 231-Hata classically used for terrestrial broadcast network planning; b) the Xia Bertoni model used for the assessment of UMTS performances (see TR 101 112 [i.24] and ITU-Recommendation M.1225 [11]. Shadowing effects: for a single terrestrial transmitter, shadowing is generally considered to be lognormal distributed. This lognormal law is characterized by its median value (corresponding to the path loss) and by its standard deviation or location variability, σL. Location variability is generally higher in suburban areas than in urban areas. For outdoor users in urban and suburban environments, the value of σL recommended by ITU-Recommendation M.1225 [11] is 10 dB at 2,2 GHz. However, in urban environment, this value can be slightly pessimistic: models described in [i.1] give a value of about 8,1 dB in urban environment and of 9,5 dB in suburban environment. Nevertheless, other references that are used by Mobile Telephony Operator also propose 8 dB for suburban areas (see clause 11). On the other hand, it is acknowledged that the value of 5,5 dB for σL is commonly used by broadcasters [13], [19] and [i.21]. So the proposed values for network planning in clause 11 are σL = 5,5 dB and σL =8 dB, for the "broadcasters' approach" and the "cellular approach" respectively. TM-SSP plans to publish refreshed recommendations on this parameter as soon as experimental results are made available. Physical layer performance assessment: Single cell environment: for physical layer performance assessment in terrestrial environment DVB-SH in common to DVB-H has selected the TU6 model [18]. The Typical Urban 6-paths model (TU6) is proven to be representative of the typical mobile reception with Doppler frequency above 10 Hz. In TU6 the channel model consists of tapped delay line (TDL) made of 5 delayed plus one non delayed path each affected by Rayleigh distributed fading with different average power. The Rayleigh fading bandwidth is dependent on the mobile speed. In addition to TU6 we also considered the so called Pedestrian Outdoor and Pedestrian Indoor (PIPO) channels described in Clause A.13 which are better suited to represent pedestrian type of terrestrial channels. Although very popular the TU6 channel shows limitations when used to represent SFN type of networks, indoor operations and low speed mobile conditions. Low fading effects (also called large scale fading) modelled for example with lognormal fading has not been included in the simulation setups for the evaluation of the terrestrial reception. These effects are covered by the margin added by the network planning. For the satellite reception the slow and very slow fading effects are covered by the LMS models. The evaluation of the fading characteristics for terrestrial networks typical for DVB-SH is subject of several on-going projects and field trials. In the future these models may be used to provide complementary performance data. It should be remarked that Rayleigh and Rice channel models definitions are also used in the context of DVB-T/H but they refer to two different contexts as explained detailed in Clause A.7. Multi-cell environments: the DVB-SH terrestrial re-transmitter network is characterized by Single Frequency Network (SFN) architecture. This characteristic must be taken into account when assessing the system performances in realistic propagation conditions when coverage is fully or partly ensured by gap-fillers as it is the case in urban and suburban environments. Contrary to the satellite-only case, the terrestrial propagation channel must be considered as frequency selective. Therefore, the assessment of waveform performances must be performed using a TDL model parameterized with a Power Delay Profile giving the stationary impulse response of the propagation channel. This PDP for such SFN configurations can be obtained through two types of methods: • for a given site for which a 3D representation is available, a site-specific deterministic tool can be used (ray launching or ray tracing tools). For a generic urban/suburban context, a macro-cellular geometric configuration is set for a given cell radius. Then, given a position of the user terminal, the channel SFN Power Delay Profile (PDP) is obtained by combining elementary PDP coming from each gap-filler taking into account the path losses from the gap-fillers to the user terminal. This second approach allows the cell radius to be tuned and requires an elementary PDP repeated for each gap-filler. Two standardized PDP corresponding to a macro- cellular single emitter case have been used in contexts close to the DVB-SH one: - ITU Vehicular A recommended by ITU-Recommendation M.1225 [11]. This PDP has been used for UMTS performance assessment; ETSI 44 ETSI TS 102 584 V1.1.2 (2010-XY) - GSM-TU6 as defined in [18]. This PDP has also been used for DVB-H performance assessment; • for path loss, the Xia-Bertoni model gives average results between dense urban and suburban cases of the COST 231-Hata model. Therefore this model is preferably used for such an analytical approach that requires a unique cell radius for the whole SFN configuration. A SFN channel model for DVB-SH physical layer performance assessment is illustrated in Clause A.13.1. . 4.2.4 Satellite specific issues Another important element in the DVB-SH transmission chain to the user is the satellite payload that together with the land mobile satellite channel greatly contributes to the end-to-end system performance. 4.2.4.1 Satellite Payload architectures Satellite payloads for mobile broadcasting systems comprise two major elements: a feeder-link receive section and a high power S-Band transmit section. The outstanding EIRP requirements obviously make the last element the most demanding and worth of a dedicated discussion. The payload architecture of a high power transmit section largely depends on the required service performance (coverage, EIRP, number of beams, etc.) and on the satellite constellation typology (GEO/HEO). Systems aiming at wide coverage regions are based on medium/large on-board antenna apertures which are composed of light-weight deployable rigid reflectors (these technologies are typically available up to about 5 meters of reflector diameter at S-band). On the contrary, regional or multi-regional coverage areas as seen from a GEO/HEO are of very limited angular extension and called for a large deployable reflector of typically 12 meters projected aperture diameter, at S-band. EIRP requirements and the difficulty of embarking large deployable antennas lead in most cases to a demanding RF power requirement that generally exceeds the generation capabilities of single amplifiers available for space applications and "power combining" must be adopted. The approach consists in adding up a plurality of amplified replicas of the desired signal individually amplified by highly-efficient/high-power devices such as Travelling Wave Tubes Amplifiers (TWTAs). The problem dimensionality is further increased if power re-configurability is required to match varying and un-predictable market demand of transmitted capacity (i.e. EIRP) among the different beams, or to adjust in-orbit the beam coverage with the use of Beam Forming Network (BFN) or Ground Based Beam Forming (GBBF) antennas. To meet these objectives, the distributed amplification concept plays a fundamental role. The general concept foresees several power amplifiers working together not only to achieve the total amount of power otherwise non-achievable with a single source, but also to constitute a single power pool to which each signal can get its own amount in a variable and reconfigurable manner. This leads to the question of whether the satellite can or cannot be used in non-linear mode as further discussed in the next clause. More details about typical satellite payload architectures are provided in Clause A.1. 4.2.4.2 Nonlinear Payload Distortion Effects For assessing payload distortion impacts on DVB-SH waveforms, the optimal operating point (OBO) of the TWTA must be evaluated. Details about the satellite nonlinearity impact and the TWTA operating point optimization methodology is provided in Clause A1.2. For this, two main satellite payload configurations should be distinguished (see Clause A.1.1): 1) single multiplex amplification per TWTA (network and polarization combining); 2) multiple multiplexes amplification per TWTA (antenna spatial combing and multibeam satellite). As shown in Clause A1.3 in the first case, the satellite TWTA can be driven close to saturation if the TDM waveform is selected i.e. SH-B. In this case, SH-B may have a power advantage of 1 dB to 1,5 dB with respect to SH-A. In the second case, since each satellite TWTA is used to amplify several multiplexes, these amplifiers have to be backed-off away from saturation in order to avoid the creation of intermodulation products. Therefore the use of SH-B is not so advantageous in terms of link-budget for this class of payloads. ETSI 45 ETSI TS 102 584 V1.1.2 (2010-XY) 4.2.4.3 Phase Noise Aspects In the DVB-SH direct path (i.e. satellite-to-user) system phase noise is dominated by the contributions from the satellite local oscillators used for on-board frequency conversion and the user terminal RF front-end Local Oscillators. This is because of the usually high frequency band used for the feeder link, the need of frequency conversion agility and the space environment. Table 4.7 reports a typical/worst-case Ka- to S-band satellite transponder DVB-SH phase noise mask which has been used to derive the impact of phase noise on DVB-SH waveform. Table 4.7: S-band DVB-SH phase noise mask (Satellite) 10 Hz 100 Hz Hz 10 kHz 100 kHz 1 MHz 10 MHz Phase noise -29 -59 -69 -74 -83 -95 -101 (dBc/Hz) There exist simplified approaches to compute the impact of the phase noise residual phase error on the digital demodulator i.e. [i.7] for OFDM or [i.8] for TDM. In practice the phase noise that can not be tracked is dominated by the high frequency components of the phase noise i.e. the components from the equivalent phase estimator phase noise bandwidth BN to the baud rate. In SH-B two pilots groups having length L = 80 symbols and separated by LTOT / 2 = 1088 symbols being LTOT the Rs PL slot length. Consequently the equivalent feedforward phase estimator can be computed as BN = kHz which LTOT corresponds to BN = 1,8 KHz for the 4 Mbaud case. The DVB-SH phase noise mask of table 4.7 produces a typical amount of 4,5° rms untrackable phase noise due to the almost flat phase noise PSD between 1 KHz and 100 KHz. This untrackable error should be not causing any appreciable impact on coded QPSK performance. The impact of phase noise error depends also on its dynamic compared to the physical layer FEC block size. Typically for the same standard deviation "fast" phase noise has much less impact compared to "slow" phase noise being the speed related to the FEC block duration. For this reason when higher than QPSK modulations are used, it is recommended to simulate the phase noise impact on the end-to-end system rather than just basing impact estimation on the simplified approach described above. 4.2.4.4 Satellite induced Doppler Shift Satellite induced Doppler shift manifests itself as a change in signal frequency which cycles through a range of frequencies and which reaches a given ± maximum value. This cycling results in a smooth and readily predictable frequency shift as a function of time. This effect should not be confused with the Doppler spread induced by multipath and the receiver speed, except in SFN mode. In SFN mode, the satellite-induced Doppler must be corrected (locally) by the CGC, otherwise it manifests itself as a Doppler spread in the hybrid reception coverage. Clause 7.5 gives recommendations for this correction. In considering Doppler shift here it is tacitly implied that any Doppler shift in the uplink can be pre-compensated for at the transmitting station. Unfortunately such pre-compensation approaches are not feasible for the downlink where the Doppler shift varies as a function of geographical location within the coverage footprint. For GEO satellites the Doppler shift is very small and presents no problems for the receiver. For the case of a GEO system with an inclined orbit (up to around 5 degrees) the Doppler Shift may be of the order of ±250 Hz. For HEO and MEO satellite constellations the downlink Doppler shift at S-band can be of the order of tens of kHz and requires that the receive can handle any initial frequency shift at acquisition and can track the frequency shift during the orbit pass. Figure 4.5 shows typical ground tracks of HEO systems targeted for European service based upon the parameters given in table 4.8. ETSI 46 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 4.5: Ground track of typical HEO orbits Table 4.8: Representative parameters of typical HEO Orbits Orbit Orbit Semi- Apogee Apogee Perigee Perigee eccentricity Inclination Argument Range of Minimum Period major Radius height Radius height (degs) of Elevation number (hrs) Axis (km) (km) (km) (km) Perigee Angles of (km) (degs) (degs) satellites 8 hr 8 2 027 33 163 26 785 7 378 1 000 0,636 63,435 270 55 to 90 8 Molnya 12 hr 12 26 561 45 804 39 426 7 307 929 0,7249 63,4 270 60 to 90 4 Molnya 63 degs 12 42 164 53 480 47 102 30 847 24 469 0,2684 63,4 270 at least 3 Tundra 60 55 degs 24 42 164 55 656 49 278 28 672 22 294 0,32 55 270 65 to 90 3 Tundra The maximum value of S-band downlink Doppler shift and the maximum rate of change of this Doppler shift with time are given in table 4.9. Table 4.9: Typical maximum Doppler shift parameters for downlink at S-Band Orbit Min elevation angle Maximum Doppler shift Maximum rate of change of (degs) (KHz) Doppler shift, (Hz/s) 8 hr Molnya 50 17 2,0 60 15 2,0 12 hr Molnya 50 22 1,8 60 21 1,8 63 degs Tundra (24 hr) 50 3,9 0,35 60 3,9 0,3 55 degs Tundra (24 hr) 50 4,8 0,3 60 4,3 0,3 MEO 10 11,0 4,5 GEO, stable orbit Any Very small Very small GEO, inclined orbit (5o) Any 0,25 0,02 ETSI 47 ETSI TS 102 584 V1.1.2 (2010-XY) 4.2.5 DVB-SH Impairments Countermeasures 4.2.5.1 LMS Mobile Channel Impact Mitigation To mitigate the impact of LMS propagation impairments, two specific techniques are recommended: exploit diversity and optimize demodulator robustness. 4.2.5.1.1 Exploitation of Time and Space Diversity The DVB-SH standard provides inherent methods for both time and space diversity: Method 1: Time diversity It is supported in DVB-SH by the highly scalable Physical time interleaver. In addition, a solution based on LL-FEC, with possible splitting of the interleaving function between the physical and the link layers is further detailed in clauses 6 and 7. Method 2: Space diversity (simultaneous reception of SC and CGC) For the SH-A (SFN), the space diversity is intrinsic in the demodulation process as the isofrequency satellite and terrestrial OFDM signals will be combined at the user terminal antenna. For the SH-B and SH-A (MFN) cases, the user terminal needs to separately demodulate the satellite and terrestrial signals and may implement different combining options depending on its capability (clause 7.6). In addition, the DVB-SH standard supports space and time diversity from "extrinsic" methods: Method 3: Multiple satellite space diversity The same combining strategy for MFN of Method 2 applies to this case too. For SFN combining there are restrictions. Method 4: Multiple receive antenna space diversity It has particular benefits in terrestrial conditions, mainly indoor (quasi-static). For the satellite direct reception antenna diversity may bring valuable performance improvement for (very) slow mobility conditions if the separation is sufficiently large (e.g. 1 meter or more) which is often difficult to support in practice, except for vehicular applications. Polarization diversity (e.g. horizontal/vertical linear or left-handed/right-handed circular) may be beneficial as soon as the reflected components still carry sufficient power. 4.2.5.1.2 Robust Demodulator Operations In a mobile fading channel, it is important to have very robust signal re-acquisition after signal loss as well as a fast and reliable signal state tracking. Although this aspect is also common to terrestrial standards such as DVB-H some specific aspects must be addressed by DVB-SH: • satellite operations with limited link margins compared to terrestrial systems; • demodulator operation at lower SNR than terrestrial systems. Guidelines for the demodulator implementations and performance for this aspect are given in clause 10. 4.2.5.2 Nonlinear Channel Impairments Countermeasures In case of multiple-carriers amplification, the use of linearized TWTA can bring appreciable advantages. Similar performance improvement can be obtained by on-ground pre-distortion to compensate for the satellite TWTA nonlinear characteristic. In case of single-carrier amplification, the use of linearized TWTAs provides limited performance improvement in particular for QPSK with low code rate as the TWTA will be operated in the vicinity of its saturation (where the linearizer has little effect). However, linearized TWTA may provide appreciable benefits for 16APSK in particular if the operating point corresponding to a given modulated signal drive is not negatively impacted by the linearizer presence. ETSI 48 ETSI TS 102 584 V1.1.2 (2010-XY) For single carrier TWTA operation, 16-APSK being multi-ring can benefit from simple static pre-distortion techniques which are able to mitigate the constellation warping effect [i.16]. Dynamic pre-distortion techniques are particularly suitable to reduce the clustering effects of higher order modulations such as 16APSK. Their benefit for QPSK and 8PSK is instead limited. 4.2.6 Receiver characteristics This clause discusses, in general terms, the architectures of DVB-SH receivers. The detailed Reference Designs for the terminal classes and their recommended performance are given in clause 10. DVB-SH receivers must be able to: • receive seamlessly a signal when moving between satellite(s) and CGC(s) coverage; • implement battery saving schemes when necessary; • exploit diversity mode with several receiver branches in parallel when appropriate; • co-exist with other active radio functions (2G, 3G, IEEE 802.11 [26], Bluetooth, etc.) embedded in a terminal; and • be relatively immune to high-level signals from various cellular network base stations (2G, 3G) or/and ISM systems. Two different receiver architectures can be distinguished according to the DVB-SH waveform options, respectively the OFDM/OFDM (SH-A see figure 4.6) and the TDM/OFDM (SH-B see figure 4.7) system architectures. The SH-B architecture encompasses the SH-A architecture in the sense that SH-B receivers can be used in an SH-A configuration. The converse is obviously not true. Receivers designed for the SH-A architecture are in general intended to be used with SFN between satellite and CGC. However, it is recommended that they should be also compatible with an MFN configuration (transmissions of the OFDM satellite signal and its OFDM CGC counterpart on two different sub-bands). 8,7,6,5MHz 1.7MHz DVB-SH 8k,4k,2k 1k TPS QPSK 16QAM Time Front end Slicing MPE-FEC + BB TS IP RF Extended Tuner Turbo MPE De-interleaver MPE-FEC decoding Antenna Radio SH-A Demodulator DVB-SH IP Decapsulator NOTE: The dual tuner may have a common front-end and downconversion. Figure 4.6: SH-A Receiver architecture ETSI 49 ETSI TS 102 584 V1.1.2 (2010-XY) 8,7,6,5MHz 1.7MHz DVB-SH Front end TPS 8k,4k,2k 1k + QPSK 16QAM Dual Tuner RF TDM/OFDM BB Time terrestrial Turbo Time De-interleaver decoding MPE-FEC Slicing TS IP Extended 8,7,6,5MHz 1.7MHz MPE MPE-FEC Signalling Pilots PL-slots Field QPSK 8PSK 16APSK Time satellite De-interleaver Antenna Radio SH-B Demodulator DVB-SH IP Decapsulator Figure 4.7: SH-B Receiver architecture 4.2.6.1 Vehicular reception constraints Vehicular receivers must work at high moving speeds and stay longer in a satellite-only reception mode. However, vehicular receivers can exploit the following features not generally possible for receivers used in a handset: • the terminal has adequate power supply able to support more complex receiver processing; • the terminal allows better antenna diversity (order 2 or more) implementation (form factor and antenna spacing); • one antenna can be optimized for satellite reception (directivity and matching polarization); • Low Noise Amplifiers (LNAs) can be integrated with the antenna(s) to reduce sensitivity loss (the underlying assumption is that the vehicular receiver does not coexist with embedded Telecom modems in the same electronic equipment and therefore the RF filtering protection for the LNA is removed or relaxed, therefore improving noise factor); • larger memory can be embedded in the receiver so that longer Physical Layer interleaving can be supported; • the higher speed of the terminal allows a better exploitation of time diversity (either at Physical Layer or at Upper Layers), at equal memory resource. 4.2.6.2 Handheld reception constraints (pedestrian) In most common cases the channel conditions are that produced by a pedestrian user (< 3 kmph). Due to the relatively low speed, continuity of service is in general achieved by increasing the link margin, rather than by increasing the time interleaving depth. When in satellite-only reception mode, some cooperation may be required from the user, i.e. to maintain good LOS with the satellite. Some challenges are associated to the specificities of handheld terminals. These include: • antenna diversity (order more than 2 is very challenging); • small battery requires an efficient power saving management; • antenna gain is in general low (can be less than -3 dBi); • antenna polarization is most often linear and not optimized to satellite reception; • embedding telecom modems like GSM or 3G inside terminal without reducing the satellite receiver sensitivity; • RF filtering, antenna design rules and compactness constraints have an impact on the achievable receiver sensitivity and immunity to high level blockers coming from the terminal; ETSI 50 ETSI TS 102 584 V1.1.2 (2010-XY) • memory limitation may, in some architectures, not allow the support of a large Physical Layer interleaver. 4.2.7 Terminal High-level architecture In theory, all types of terminal can embed large memory and all options defined above. Furthermore, no limitation is imposed on the number of embedded multi-band or/and multimode 2G, 3G communication modem(s) or other wireless features like Bluetooth or WiFi. Also GPS could be integrated in a terminal. In practice, compactness, supply source and price considerations split terminal in categories. Categories chosen in DVB-H Implementation Guidelines [i.21] and in EICTA/TAC/MBRAI-02-16 [14] are kept (see clause 10). Figure 4.8 shows the terminal high-level architecture for all classes of terminals. Terminal sub-systems will be designed according to their terminal categories and manufacturer marketing requirements. antenna 1 front end demodulator 1 diversity host multimedia & tuner 1 combiner processor & & video and audio antenna 2 front end demodulator 2 decoding & tuner 2 channel decoding Host memory antenna N front end demodulator N Memory & tuner N DVB-SH receiver Man Machine Interface & display power supply & power management optional 2G/3G Telecom module optional connectivity & peripherals Figure 4.8: Terminal high-level architecture 4.2.8 Compatibility with Existing DVB-H System Features 4.2.8.1 Battery-powered receivers This clause deals with receivers embedded in handheld "convergence" terminals and pocketable digital TV receiver devices (see clause 10 for definitions of receiver categories). For these devices, battery saving is essential. 4.2.8.1.1 Time slicing The DVB-H group has defined time slicing as a key mechanism for battery saving purposes. DVB-SH receivers also use this mechanism for the same purposes. Burst size and off time, maximum burst duration, power saving calculations are based on same assumptions than DVB-H. 4.2.8.1.2 Low-power Microelectronics State-of-the-art technologies like the newest RF BiCMOS or SiGe for tuner and CMOS 65 nm process for channel decoder and sometimes full CMOS process for one chip solutions have significantly reduced the contribution of the front-end in the overall terminal power consumption, compared to what it used to be some years ago. 4.2.8.1.3 Antenna Diversity (for S-band) Operation at S-band enables the use of antenna diversity, even in small handheld terminals. Antenna diversity does increase power requirements since more than one receiving chains operate in parallel. However, this power consumption increase can be kept acceptable compared to the gain in QoS improvement with the use of new low-power technology and time-slicing mechanism as stated above. In addition, a signal quality indicator could be used to turn off or reduced diversity processing when this is not required. ETSI 51 ETSI TS 102 584 V1.1.2 (2010-XY) 4.2.8.2 Service Switching time (Zapping time) Zapping time is an important subjective quality criterion for the user in a multi-program offer. Due to the combined effect of time-slicing for battery-life conservation and time-diversity techniques to overcome long fadings, possible degradation of zapping time may be incurred if appropriate counter-measure techniques are not applied. In DVB-SH it should be noted first that zapping time may be different depending on the coverage (CGC or Satellite) and the particular length and structure of the time interleaver chosen for the propagation channel in each coverage. In the CGC coverage, DVB-SH zapping time, normally, does not exceed 2 s on average. In the satellite coverage, one can distinguish between good reception conditions (e.g. about > 5 dB above threshold) and critical reception conditions. A "smart" terminal can optimize zapping time according to its measured reception margin. In good reception conditions, such terminals can immediately play out any source data without waiting for maximum FEC protection to be fulfilled. This strategy is made possible by special sending arrangements of FEC parity and source data allowed by DVB-SH. Techniques for transitioning between "immediate" play out strategy (in good reception) and maximum error protection (in mobility) may be needed to be implemented in the terminal. Clauses 6 and 7 describe the use of these techniques, for the link and the physical parameters respectively, and also discuss their possible side implications. 4.2.8.3 Support for Variable Bit Rate (VBR) and Statistical Multiplexing (Statmux) The use of VBR and statistical multiplexing (Statmux) of the different services transmitted in a single TS allow either bandwidth saving or better video quality. The actual gain provided by VBR depends on the system configuration, on the types of contents and on the number of services multiplexed statistically on the TS. The main element impacting VBR implementation with DVB-SH is the combination of the Time interleaver together with Time Slicing: • with short interleaving (i.e. slightly exceeding the length of the Burst), VBR is supported with time slicing thanks to the "delta-T" mechanism defined for the same purpose in the DVB-H standard; • when long interleaving is used, Burst content (and hence "delta-T" information) is spread over several Bursts. This adds complexity in handling of Time Slicing, but does not preclude the use of VBR nor reduce its bandwidth saving gain. VBR implementation is described in more detail in clause 6.3.4. 4.2.8.4 Multiple QoS support 4.2.8.4.1 Introduction This clause deals with the support of different quality of service in a DVB-SH environment. The overall quality of service results from the combination of different layers. Streaming protection is mainly achieved by physical and link layers, whereas physical and application layers achieve file delivery protection. Depending on the chosen systems configurations, different options are possible at different layers as presented in table 4.10. Table 4.10: Configuration options for the different classes Terminal option Parameters Relevance Class 1 Physical layer Rate ++ Link layer Rate; depth ++++ Application layer Rate, repetition ++ Class 2 Physical layer Code rate; interleaver ++++ Link layer Code rate; interleaver + Application layer Rate, repetition ++ ETSI 52 ETSI TS 102 584 V1.1.2 (2010-XY) 4.2.8.4.2 At physical layer DVB-SH does not provide variable quality of service at physical layer. The following parameters must be configured according to the expected performance of the most demanding service and as a function of the worst channel model scenario: • code rate; • interleaver configuration, in particular interleaving depth. 4.2.8.4.3 At link layer DVB-SH does provide at link layer a per-service protection based on MPE sections. It is possible to configure a streaming service with a long interleaving while a "push" service may be configured with no protection at the link layer since the protection is provided at the application level. 4.2.8.4.4 At application layer DVB-SH does not provide any specificity for the application layer since the interface with the "application world" is IP. However, the DVB-SH application layer is fully compliant with all CBMS specifications and, in particular, with the file delivery specified in document [Content Delivery Protocols]. It is also expected that other techniques may be applied to the streaming services such as scalable video encoding. Scalable video encoding must be coordinated with the link layer protection since the former is based on a differential protection applied to two parallel streams making the video stream. 4.2.8.4.5 Recommendations The procedure displayed in figure 4.9 should be followed when configuring the parameters of the FEC protection.              Figure 4.9: Procedure for choosing FEC protection parameters Several configurations are possible while offering the same useful throughput but not always the same performance: • protection at physical layer only compared to joint physical and link layer protections; • different splits of the joint protection. The network operator, depending on the deployed network and receivers, will choose the physical layer configuration. The service operators will then be able to choose if they want to activate the link layer and application layer protections, trading performance versus available bandwidth. ETSI 53 ETSI TS 102 584 V1.1.2 (2010-XY) Differentiation between services at application and link layer may usually happen in the following circumstances: • when different kinds of services are broadcasted in parallel: this is for instance the case between the streaming application (which will highly benefit from the link layer protection) and the file download application (which may benefit better from the content delivery protection and then will less use the link protection); • when, for a specific service, different levels of protection are required: this is the case of scalable video encoding which requires two levels (or more) of protection to protect better the main stream than the others. 4.2.8.5 Service announcements in hybrid networks This clause deals with announcement of services within the DVB-SH context. 4.2.8.5.1 Introduction Signalling within the DVB-SH system follows the layered approach described in figure 4.10. Figure 4.10: Layered signalling approach According to this layered approach, a terminal learns all information about the network and services at first switch-on time in the following steps: • radio configuration: the terminal can discover the physical layer parameters by scanning relevant frequencies and recovering either TPS (in OFDM waveform) or Signalling Field (in TDM waveform). These fields will deliver all information about the baseband configuration like FEC ratio, interleaver depth, etc.; • TS configuration: the terminal is then able to decode the MPEG2 flow. If required it will discover the SH services structure by reading the SHIP packets; • section configuration: the terminal will read the PSI/SI tables that will give all system related information to be able to derive from a target IP address the relevant PID: Given a target IP address, the IP Notification Table (INT) will deliver the relevant component tag of the target elementary stream; • then using the PMT and PAT, the relevant PID will be found; • programme configuration: the terminal is then able to read to the bootstrap ESG IP address to discover the list of ESG providers, choose one and then listen to the relevant IP address to discover its electronic service guide. The ESG has the format of an XML file describing what content can be watched when, typically . Then, using PSI/SI the relevant PID is fetched by the terminal. Compared to DVB-H, the only difference is the introduction of the SH service layer. The service announcement is performed, as in DVB-H, according to the Electronic Service Guide specification of the DVB-CBMS protocol suite. In general, there is nothing specific compared to the ESG used for the DVB-H, in particular for the SFN case that is very similar to the DVB-H case (no need for SH services). In the MFN case, the announcement of the local content and of its availability at particular locations will require the use of the "SH service" concept (see clause 6). ETSI 54 ETSI TS 102 584 V1.1.2 (2010-XY) 4.2.8.6 Handover issues This clause describes in general terms the handover issues in DVB-SH. DVB-SH has been designed to take advantage of the mechanisms for handover that have been defined in the DVB-H framework (see TS 102 470 [21]). In particular, the following principles are maintained in DVB-SH: • use of physical layer signalling: for instance, the TPS provides information such as cell_id to quicken the detection of the target network. A similar signalling mechanism has been defined for the DVB-SH TDM waveform; • use of spare time in time-slicing for scanning available frequencies: this is also possible in DVB-SH, even with long physical layer interleaving; • PSI/SI signalling: the same philosophy of signalling has been maintained, with some adaptations to cater for the hybrid architecture. The NIT describes the full network in a beam, including the target frequencies and waveforms. The INT describes the availability of the session on the target transport stream. Nevertheless, DVB-SH introduces also some specificities that are worth keeping in mind: • presence of the satellite "umbrella" cell that covers all the terrestrial cells; • terrestrial cells can be much smaller than in DVB-H, especially in S-band (less than a Km in radius); • terrestrial cells and satellite umbrella cell may use different time interleaver lengths for the same service (for SH-B and MFN-SH-A; • terminals with diversity reception have more than one front-end and demodulator; • coherency of PSI/SI signalling must be addressed carefully when code diversity combining (with SH-B only) is used. A detailed analysis of handover in DVB-SH is given in reference [i.9]. 4.2.9 Support of Other Frequency Bands The DVB-SH applicability should not be limited to above mentioned bands. The standard should be scalable to other frequency bands and channelization schemes that could be allocated to satellite broadcasting services. Possible other frequency bands to be supported by DVB-SH are the following: 3) Ku band: non planned FSS, planned FSS, planned BSS: 10,7 GHz to 12,75 GHz; 4) Ka band: non planned: 17,3 GHz to 20,2 GHz. BSS: 21,4 GHz to 22 GHz; 5) other bands: e.g. C-band. A high-level survey about the standard applicability to the above listed frequency bands above 3 GHz led to the following preliminary conclusions: • the SH-A configuration is definitely not suitable for A3GH applications as the OFDM waveform will not be able to cope with the higher Doppler spread of A3GH systems (in particular Ku and Ka-band) unless a drastic reduction in mobile user speed can be accepted; • the satellite TDM component of the SH-B configuration can also be exploited for A3GH bands with no expected impact on performance. In fact the mobile channel countermeasures adopted for SH-B are also valid for A3GH operations. The possible Ku/Ka-band receiver different phase noise mask compared to B3GH applications must be compatible with SH-B TDM pilots structure; • for the CGC other solutions must be used (different waveform compared to the SH-B) or different frequency bands (e.g. B3GH whereby SH-B OFDM waveform can be exploited). ETSI 55 ETSI TS 102 584 V1.1.2 (2010-XY) 5 System Configurations and possible Deployment 5.1 Definitions and concepts System: combination of physical elements (satellite, CGC, terminals, etc.) implementing the DVB-SH specifications. Radio configuration: defines the selected parameters of the DVB-SH waveform. Among available options we find: SH-A (OFDM/OFDM) and SH-B (OFDM/TDM), interleaving (physical, link layer), FEC ratios, modulation mode, hierarchical modulation. Frequency configuration: spectrum allocation to the satellite and terrestrial component. For the terrestrial transmitters, it is useful to distinguish between hybrid and non-hybrid frequency sub-bands. A non-hybrid frequency sub-band is defined as one that carries only local content. One main option in the frequency configuration is the choice between SFN and non SFN. In a SFN configuration, the hybrid sub-bands carry only the Common content. In a non SFN configuration, hybrid sub-bands include also those that carry a mix of Common and Local content. Topological configuration: used in conjunction with frequency configuration, it defines where the spectrum is available, on a satellite ("beam") or terrestrial ("cell") basis. For satellite, it gives the location and shape of the beams. For a terrestrial cell, it gives the centre (in latitude, longitude) and extension of the cell. Local Content Insertion configuration: defines the way the local content is inserted by the CGC transmitters, in the context of a non-SFN network. CGC Transmitter configuration: hardware options placed on the transmission site; it concerns the location (co-sited with a 3G base station, in a new specific site), the antenna (reused new) and the power. Receiver configuration: hardware options; it concerns essentially the available memory for physical and link layer interleaving (using class 1 and class 2 categories) and the antenna options (diversity, gain, etc.). System configuration: set of options retained for a DVB-SH system; this includes radio, frequency, topological, content, receiver and transmitter configurations. System transition: evolution of a system when one of its configurations changes. Not all transitions are relevant and/or foreseen in the system deployment. Compatibility: a receiver is compatible with "radio configuration X" when it can successfully decode the useful information using only a sub-set of the waveform elements allowed in this configuration. A lower reception margin or a more restricted coverage area is implied compared to that of a compliant receiver. Compliancy: a receiver is compliant with "radio configuration X" when it can exploit ALL the waveform elements allowed in configuration X to maximize reception margin and/or coverage area. 5.2 Configurations options definitions This clause discusses in more details on the options that are offered. 5.2.1 Radio configuration The air interface has many available parameters as reflected by the flexibility of the SH delivery system descriptor. Only three of them are discussed in this clause to provide visibility about the main classes. The complete set of possible scenarios is more complex than what illustrated hereafter. Choice of TDM (SH-B) or OFDM (SH-A) for the satellite signal: this choice is dictated mainly by spectral efficiency considerations (satellite, terrestrial, overall) and the satellite nonlinearity characteristics as discussed in clauses 4.1.3, 4.2.3.2 and C.1. Choice of between physical layer or link layer techniques to combat long duration fading: comparison of these techniques is given in clauses 6 and 7. The choice is dictated by the performance required in a long duration fading environment as well as the performance for achieving fast zapping; however, the cost and required foot-print of the memory to implement a long interleaver at the physical layer must be considered. The combination of a short physical ETSI 56 ETSI TS 102 584 V1.1.2 (2010-XY) interleaver with a long link layer interleaver is typically preferable for handheld terminals (see clause 10) whereas a long physical interleaver can suit transportable receivers like those installed in cars. When different modulation parameters and frequencies are used between the satellite and the terrestrial signal, it is possible to resort to different interleavers (physical and link layer) for the satellite and terrestrial signals hence extending available options for the system. Choice of Modulation and FEC: guidelines for this choice are given in clause 7. 5.2.2 Frequency configuration It is important to plan how the frequency can be used and, more importantly, reused. • firstly, the bandwidth of a frequency sub-band must be decided between the values allowed in the standard (1,7 MHz, 5 MHz, 6 MHz, 7 MHz and 8 MHz). This bandwidth is most of the time implied by the frequency band selection; • secondly, the choice between SFN and non-SFN must be made, on a beam-by-beam basis. In SFN case, a minimum of 1 sub-band is needed per beam, whereas in the non-SFN case this minimum is 2 sub-bands. Since the sub-band that repeats the Common content in the non-SFN CGC may need to be at different frequencies in different places, this minimum may be higher than 2; • thirdly, the remaining sub-bands can be allocated to local content transmission, on a per-cell basis. 5.2.3 Topological configuration This consists in allocating the sub-bands to the topological elements, i.e. the satellite beam(s) and the terrestrial cell(s). Adjacent beams must not have common frequencies for interfering avoidance reasons. This is the same for adjacent cells. However it is possible to reuse these frequencies: • a frequency used in a beam/cell can be reused in another beam/cell if this beam/cell is sufficiently separated from the first; • terrestrial cells can reuse an adjacent beam frequency provided that these cells are sufficiently far away from this beam. Actual frequency allocation to sub-bands may be quite complex due to possible interferences from the satellite adjacent spot beams. In particular, for reusing adjacent spot beam frequency, terrestrial reuse is sought in the centre of the spot (far from the borders) but this is not always possible. 5.2.4 Local content insertion configuration Two different techiques can be used for inserting Local content: Technique 1: Local content insertion by usage of the special hierarchical modulation: As shown in Figure 5.1 hierarchical modulation can be used by splitting content into two TS at the input of the terrestrial transmitter: the first TS is input to the primary interface of the terrestrial modulator; this TS is exactly the same as the one going to the satellite transmitter. The Second TS is input to the secondary interface of the terrestrial modulator to carry local contents. There is no specific need for synchronizing the two TS at the service level since they are essentially independent. However, there is a need to synchronise the two HP and LP streams at the SH frame level so that SFN can also be ensured. This latter synchronization is ensured is described in section 7.5.4. Such technique enables in principle to double the content at the expense of some interferences generated by the local content signal on the satellite primary signal. Such interferences can be assimilated to an extra additive noise and it is usually negligible compared to the terrestrial transmitter power. ETSI 57 ETSI TS 102 584 V1.1.2 (2010-XY)           Figure 5.1: Hierarchical modulation for local content insertion. Figure 5.2 provides a block diagram of a network using terrestrial transmitter to broadcast local contents, on top of the global content delivered by the satellite, by the use of a hierarchically modulated signal. Code rates in the High Priority (HP) stream carrying Global content and the Low Priority (LP) stream carrying Local content may be different. But in any cases, the satellite signal carries only the HP stream. Accordingly, as it can be seen in Figure 5.2, HP-hierarchical content is available in any part of the service area while the LP-hierarchical content should be available only in areas covered by terrestrial transmitters where some additional content is transmitted. Whatever the location of reception, the information carried by the TPS (Transmission Parameter Signalling : physical layer signalling) signals a hierarchical transmission even if the LP stream is missing in the received signal, under the satellite coverage and possibly terrestrial coverage where there is no additional content. High Priority Global Content Channel coding Hierarchical Low Priority Modulation Local Content Channel coding Figure 5.2: Hierarchical modulator for local content insertion. Figure 5.3 provides a block diagram of a receiver capable of demodulating hierarchically modulated signal and extracting both global and local content. High Priority Channel decoding Global Content Hierarchical Demodulation Low Priority Channel decoding Local Content Figure 5.3: Hierarchical demodulator supporting local content insertion. NOTE 1: Hierarchical modulation can be used also for other purposes. These cases are described in section 5.4 Technique 2: Local content insertion by usage of the partially available transport stream. Another possibility is to use the notion of partially available Transport Stream for service regionalization as specified in [21]. Using the SHIP signalling, the transmitters will forward only the relevant part of the TS. The satellite transmitter removes all the local content. The terrestrial transmitters remove the part of the local content they need not forward. The concept is exemplified in Figure 5.4. ETSI 58 ETSI TS 102 584 V1.1.2 (2010-XY) TS_id1 (satellite services) Adaptation TS_id1 (satellite + area 1 services) TS_id1 (all services) & distribution TS_id1 (satellite + area j services) TS_id1 (satellite + area N services) Mandatory for the signalling Left open to implementation Mandatory for the modulation Figure 5.4: service regionalization concept. Service regionalization is fully specified in [21]. 5.2.5 CGC transmitters configuration There are two main approaches for configuring the CGC network of transmitters: • the "high-density", "low-power" approach attempts to reuse, all or partially, existing 3G/2G transmitter sites, or to build an equivalent low to medium height type of transmitters network. These networks are characterized by transmitter towers of typically 30 meter high delivering from 200 W to 1 kW ERP for dense urban coverage in the range from 0,5 km (deep indoor) to 2 km (outdoor). • the "low-density", "high-power" approach attempts to reuse existing digital terrestrial TV transmitter sites, or to build an equivalent high-altitude transmitters networks. These networks are characterized by transmitter towers from 100 to 300-meter high, delivering from 1 kW to 4 kW ERP for typical coverage in the range from 5 Km to 7 Km. 5.2.6 Receiver configuration The standard identifies two receiver classes based on their capability to process time diversity (refer to clause 10 for the memory sizing guidelines for each Class). Class 2 receivers offer maximum flexibility in terms of time diversity. Class 1 receivers are compatible with the time diversity transmitted for Class 2 (refer to clauses 6 and 7 for the different trade-offs between memory complexity and time-diversity performance). 5.2.7 Configuration naming Among the various possible configurations described above, 4 are considered as critical in system deployment: Radio configuration, Receiver Class, SFN/MFN configuration, Local content handling. We propose the following nomenclature for a given configuration. ETSI 59 ETSI TS 102 584 V1.1.2 (2010-XY) SH w Cx SFN y LC z A: OFDM/OFDM B: TDM/OFDM HM: Hierarchical Modulation 1: SFN SR: Service Regionalization 1: class 1 0: MFN No: No local content 2: class 2 Figure 5.5: DVB-SH system configuration classification The rules for referencing the configurations are the following: • when the option is not set, the character is set to this value w, x, y, z; for instance if we want to refer to SH-A configurations in SFN, whatever the other parameters may be (Receiver class and Local content), we refer to "SH-A-C-x-SFN-1-LC-z"; • wildcards are always 1 character long and lower cap ("w", "x", "y", "z"); • exact values are either 1 character long in higher cap ("A", "B", "1", "2", etc.) or several characters long in either high or low cap ("hm", "sr", "HR", "SR", etc.) 5.3 Examples of system deployment Below is a non-exhaustive list of relevant transitions: • from SH-w-C-1-SFN-1-LC-z to SH-w-C-2-SFN-1-LC-z: this transition corresponds to a terminal upgrading in terms of memory, in particular by replacing the long upper layer with physical layer long interleaving; • from SH-w-C-x-SFN-0-LC-z to SH-A-C-x-SFN-1-LC-no: this transition corresponds to a bandwidth upgrading by moving from non SFN conditions to SFN conditions; • from SH-A-C-x-SFN-1-LC-no to SH-w-C-x-SFN-0-LC-z: it would make non sense downgrading overall bandwidth by changing from SFN to non-SFN unless terrestrial network is dense enough to enable important local content insertion that mitigates bandwidth loss. So this transition corresponds to an increase of local content by a strengthening of the CGC network; • from SH-A-C-x-SFN-1-LC-no to SH-A-C-x-SFN-1-LC-hm: when the terrestrial network is dense enough, it makes sense to introduce some local content on the hierarchical modulation. These transitions will be used in the following descriptions of system scenarios. 5.3.1 SH-A deployment scenario Below is one typical SH-A deployment. Step 1: early deployment of CGC • SH-A-C-1-SFN-0-LC-no configuration enables to deploy receivers with the constraint of limited memory, both on physical and link layers. Antenna diversity and physical short time diversity offer quite good performance under CGC coverage. Frequency reuse is possible between all CGC cells. Step 2: launch of a satellite with limited power: • no transition: LL-FEC is used for time diversity, depending on the link layer memory deployed in Step1, early terminals may remain compliant; • change to a stronger physical layer protection through SH-A-C-2-SFN-1-LC-no: early terminals are only "compatible" (inside CGC coverage, in good satellite only reception conditions); • provide differentiated contents through SH-A-C-x-SFN-0-LC-sr: early terminals continue to receive as before under the CGC and correctly receive satellite only under good reception conditions; ETSI 60 ETSI TS 102 584 V1.1.2 (2010-XY) • a transition to SH-B leads to the replacement of early terminals. The viability of this transition is therefore dependent on the latter installed base; • in the case of NGSO satellite, SFN-yes may no longer be possible. Step 3: launch a second generation terrestrial network we have 2 basic options • no transition; • increase local content insertion by increasing ratio between satellite and terrestrial bandwidth on the hybrid frequency through SH-A-C-x-SFN-1-LC-hm: early terminals continue to receive as before under the CGC and correctly receive satellite only under good reception conditions; in addition some local content can be received on the hierarchical modulation; Step 4: launch of a satellite with extended power • different Modulation and FEC parameters may be fine tuned to take advantage of the increase satellite margin. Note that SFN operation may limit the freedom to tune these parameters. 5.3.2 SH-B deployment scenario In a typical SH-B deployment, we have the following steps: Step 1: early deployment of CGC • SH-B-C-1-SFN-0-LC-z configuration enables to deploy receivers having limited memory constraints, both on physical and link layers. Physical short time diversity offer quite good performance under CGC coverage. Step 2: launch of a satellite with limited power capacity; 3 possible configurations are possible • no transition: LL-FEC is used for time diversity, depending on the link layer memory deployed in Step1, early terminals may remain compliant; • transitions to a new configuration with better time diversity through SH-B-C-2-SFN-0-LC-z: early terminals are only "compatible" (inside CGC coverage, in good satellite only reception conditions); • transitions to a new configuration with better local content insertion: transition through SH-B-C-x-SFN-0-LC- sr: early terminals are still compliant; more local content is available. Step 3: launch a second generation terrestrial network we have 2 basic options • no transition; • increase local content insertion by increasing ratio between satellite and terrestrial bandwidth on the hybrid frequency and go to SH-B-C-x-SFN-0-LC-sr. Step 4: launch of a satellite with extended power capacity • different Modulation and FEC parameters can be easily fine tuned to take advantage of the increase satellite margin; • depending on the relative satellite power, it could also be envisaged to transition to SH-A if local content insertion is not sufficient and satellite power is large enough. 5.3.3 Less likely transitions These are listed hereafter: • from SH-w-C-2-SFN-1-LC-z to SH-w-C-1-SFN-1-LC-z: this transition is relevant only when the network/market/use cases/receiver technology removes the need of long physical interleaver introduced in the earlier stage; • from SH-B-C-x-SFN-0-LC-z to SH-A-C-x-SFN-0-LC-z: the underlying logic is that a transition from SH-B to SH-A is accompanied by a transition from non-SFN to SFN. Note however that a transition from non-SFN to ETSI 61 ETSI TS 102 584 V1.1.2 (2010-XY) SFN is not always easy if the early CGC network takes advantage of the non-SFN characteristic and is not configured uniformly; • from SH-w-C-x-SFN-0-LC-hm to SH-w-C-x-SFN-0-LC-sr: it makes non sense decreasing terrestrial/satellite bandwidth ratio since the terrestrial network density can generally be improved over time and its bandwidth increased. 5.3.4 Forbidden configurations Configurations that are not technically possible are listed below: • SH-B-C-x-SFN-1: it is not possible with the SH-B radio configuration to have SFN since the modulation is not the same; however with tight synchronization between satellite and terrestrial signals, it is possible to have a code-combinable signal. This kind of detail is not listed here. • SH-A-C-x-SFN-yes-LC-rm: it is not possible in the SFN case to have different radio configurations between the satellite and terrestrial transmitters and so enable content removal. 5.4 Other “advanced” hierarchical modulation use cases 5.4.1 Introduction The following advanced use cases have been identified and listed in the table 5.1 hereunder. Case Name Area Comment 1 Quality Satellite This technique can also be used also for increase providing support of heterogenous terminals. 2 Capacity Satellite This technique is already used by increase commercial systems for adding additional video programs on top of audio programs. 3 Coverage Terrestrial ; This technique is already used by extension Satellite commercial systems 4 Zapping both time Table 5.1 hierarchical modulation “advanced” use case list In the next paragraphs and for each use case an application scenario example is derived and their gain evaluated. 5.4.2 Use case 1: quality increase The principle is to use a hierarchical 16QAM modulation configuration similar to a non-hierarchial QPSK one where the HP bit rate is maintained equal to the non-hierarchical QPSK stream while increasing the quality of the reception via an additional stream received on the LP. - In a first operational mode, the video quality improvement is obtained by transporting over the LP channel additional MPE-IFECprotection for HP stream (typically bringing MPE-IFEC code rate from 2/3 to 1/.2). - In a second operational mode, the video quality improvement is obtained by transporting over the LP channel an enhancement of video layers in using SVC technique. The flexibibility of SVC enables to bring different usages: o PSNR scalability: Assume that a base layer transmitted with a bit rate rBL yields an average PSNR of QBL for a certain video sequence. When an enhancement layer with an additional data rate rEL is ETSI 62 ETSI TS 102 584 V1.1.2 (2010-XY) received, the PSNR can be increased to QEL. The equivalent single layer AVC stream of same bit rate rBL + rEL would yield approximately same quality QEL. o Spatial scalability: Assume that the base layer provides video at QVGA resolution with an average PSNR of QBL at a bit rate rBL. The enhancement layer brings an additional data rate rEL which enables to provide roughly same quality at a larger resolution, e.g., VGA. - In a third operational mode, LP bit rate can also be used for providing support to heterogeneous terminals without resorting to simulcasting. The additional SVC layers enable spatial resolution extension. This use case is proposed to offer graceful degradation: on a low-protected TS, additional data can be exploited to improve content quality, either by sending enhancement layer data (as in scalable video coding), or by sending additional MPE-IFEC protection using a second source (see clause 6). The benefit on the signal quality will depend mainly on the position of the terminal under the satellite-only coverage:  under good reception quality, the terminal will benefit from both TS signals, the high priority that conveys usual video and IFEC protection, and the low priority signal that conveys enhancement layer data and/or additional MPE-IFEC source;  under bad reception quality, the terminal will benefit only from the high priority giving the basic video layer and primary source of MPE-IFEC protection. 5.4.3 Use case 2: capacity increase The principle is to use the HM and H.264 SVC video coding so that HP TS carries the AVC-compliant whereas LP carries additional SVC enhancement layers. Since the base layer requires less bit rate than the AVC layer at the quality of the highest SVC enhancement layer, this will turn into more available channels. A different code rate can be used on the HP to mitigate the HP penalty with regard to QPSK and therefore roughly keep same coverage and availability for dual HP and LP reception as for QPSK. On typical scenarios, using HM, we can provide: - 33% capacity increase (e.g. from 6 to 8 programs) with same quality as baseline during 1/3 of the total time (this is the due to LP reception penalty compared to HP stream); - 30% service disruption reduction due to better C/N requested level in HP compared to QPSK. - This comes at the cost of a quality degradation during 60% of service time since during 50% of total time, we receive the only HP stream. 5.4.4 Use case 3: coverage extension The principle is to benefit from intrinsic UEP from HM added to a complementary UEP with FEC (lower code rate) to: - extend HP coverage to size larger than 16QAM coverage: under this HP coverage the bit rate will be lower than the 16QAM bit rate; - limit the LP coverage degradation with regards to 16QAM coverage: under this LP coverage, the added bit rates of HP and LP will be greater or equal to the 16QAM bit rate. Using HM with the application scenario suggested enables the operator to provide: - same number of programs at base layer quality over an 10% increase of the baseline coverage; - same number of programs at same quality (base & enhanced layers) over 85% of the baseline coverage. So the cost of the extended coverage with lower quality is the diminution of the coverage with normal quality. ETSI 63 ETSI TS 102 584 V1.1.2 (2010-XY) 5.4.5 Use case 4: zapping time reduction In cases of systems that use long interleavers, the principle is the use of the LP channel with a shorter interleaver to transmit auxiliary information, resulting in faster resynchronization of the video and audio decoders. For example, the LP interleaver might be reduced to one second, resulting in an average interleaver delay of 1.5 seconds. It is proposed that the LP channel contain a copy of the audio channel. This would allow the audio portion of the multimedia stream to begin to play much sooner after a channel change request. Additionally, it is proposed that the LP channel contain periodic intra coded frames, known in the H.264 coding standard as SI frames. The average delay before the first video frame is half the time between SI frames. The consumer's experience of a channel change could therefore be the following: - after a fixed delay of approximately two seconds, the user gets full audio and low frame rate (1 Hz) video from LP; - this reception condition lasts for a period of 1.5 times the delay of HP interleaver; - after this delay, the HP de-interleaver can deliver full rate video to the user an the channel change time is over. This solution also gives the service provider the option to optimize the rate of IDR frames in both the HP and LP channels. Reducing the frequency of IDR frames in the HP channel (increasing GOP size) will result in either reduced HP bandwidth or improved video quality. Increasing the frequency of SI frames in the LP channel will provide still more improvements in user experience: - at zapping or after a very long signal interruption (when all data, HP and LP, are lost): when signal resumes, LP will be faster to display than HP; - some errors could happen in HP but not in LP that would then provide some diversity information1. Note that the above solution improves the zapping time only for terminals that are able to receive the LP channel. Assuming that the protection of the latter is weaker than on the HP channel, this might limit the effectiveness of the mechanism to few terminals in the coverage area. Another possibility is the following: The base layer operates with the desired frequency of IDR frames, while the enhancement layer contains less IDR frames which saves a significant portion of the total data rate. In this case, all terminals benefit from the reduced zapping time, albeit at a slightly reduced quality immediately after channel switching until reception of the first IDR in the enhancement layer on the LP channel. 6 Elements at Link and Service layers 6.1 Introduction DVB-H presents a layered system structure: equipments operating on a specific layer can easily interconnect to equipments operating on an adjacent layer. Acknowledging this approach, the DVB-SH reuses to the most extent the DVB-H link and service layer in order to achieve seamless interoperability with DVB-H and to benefit from all available DVB-H link layer features as well as the already developed DVB-H ecosystem. This layered approach is presented in figure 6.1: • a set of IPDC servers deliver IP streams, including the video streams; • these IP streams are encapsulated by a DVB-SH IP encapsulator; the latter performs IP to MPE encapsulation according to EN 301 192 [9], PSI/SI insertion and MPE-IFEC protection and delivers an MPEG2 TS for the DVB-SH modulator; 1 Statistics of such events is FFS. ETSI 64 ETSI TS 102 584 V1.1.2 (2010-XY) • the DVB-SH modulators delivers a radio signal ultimately received by the DVB-SH receiver which performs baseband demodulation and decoding and processes the MPEG2 TS in the link layer client; • the latter processes sections, MPE, MPE-FEC, MPE-IFEC, PSI/SI, and delivers an IP stream to the IPDC client; • the IPDC client processes the IP streams, for example to deliver the ESG, the security decryption and the video and audio play out.                     Figure 6.1: DVB-SH layered approach Key features of DVB-SH link and service layers are: • Support of Multi-Protocol Encapsulation: - DVB-H provides an IP multicast transport on top of MPEG2 Transport Streams (TS). To encapsulate the IP datagrams over MPEG2 TS, Multi-Protocol Encapsulation (MPE), (see EN 301 192 [9], clause 7) is applied. As the DVB-SH physical layer is also MPEG2 TS based, DVB-SH reuses MPE for the transport of IP datagrams over DVB-SH physical layers; - MPEG2 TS-based transport and MPE enable to reuse most signalling concept of DVB-H also for DVB-SH; - note that DVB-SH physical layer also supports the GSE but this is for further study. • Support of time slicing: - DVB-H uses the real-time parameters, specifically the Delta-t information, conveyed within MPE and MPE-FEC headers in order to inform the start of the next burst. DVB-SH re-uses this concept: each MPE, MPE-IFEC and MPE-IFEC section carried by the MPEG2 TS over DVB-SH physical layer includes the same Delta-t information; - this mechanism enables to power off the terminal during periods where no relevant bursts for this service are transmitted. This also enables hand-over even for receivers with a single demodulator in case the infrastructure appropriately synchronizes the transmitted TS; - in addition, time slicing enables the efficient support of variable bit rate services since Delta-t can be adapted for each burst size. This is one way to efficiently support statistical multiplexing. • Support of link layer protection: - DVB-H permits the use of link layer protection by applying MPE-FEC (see EN 301 192 [9], clause 9) to counteract terrestrial fading. DVB-SH also supports the use of MPE-FEC; - in addition, MPE-IFEC may be required in satellite coverage, especially with class 1 receivers; - by doing so, individual protection for each service is enabled. Depending on the service requirements and the physical layer performance, the transmitter can select from a variety of link layer parameters, e.g. using single burst MPE-FEC or multi-burst MPE-IFEC. Each FEC protection scheme can be fully configured to the service requirements thanks to a number of parameters; ETSI 65 ETSI TS 102 584 V1.1.2 (2010-XY) - simultaneous use of MPE-FEC and MPE-IFEC on the same elementary stream is for further study and therefore not allowed. • Support of IPDC features: - DVB-SH is fully compatible with the DVB IPDC specifications, enabling a fast deployment of services on top of DVB-SH physical and link layers through the reuse of the IPDC protocol stack. Recommendations on the use of IPDC specifications in a DVB-SH context are given in [CDP over SH IG]; - DVB-SH uses updated PSI/SI to convey system and program parameters. This enables smooth transition scenarios between DVB-SH and DVB-H networks, in particular for handovers: dual-mode receivers may receive content on one or the other technique seamlessly, provided PSI/SI are coherently signalled. The PSI/SI updates to support DVB-SH signalling are under study. As DVB-SH relies to the most extent on DVB-H link layer technologies, this clause will only highlight the additions and differences of DVB-SH link layer compared to DVB-H. In case information on specific aspects on the DVB-SH link layer can not be found in this clause, the DVB-H link layer specification has to be used. To address the modification and additions of DVB-SH link layer compared to DVB-H link layer, this clause is organized in three parts: • the first part introduces of a new link layer forward error correction referred to as MPE-IFEC. This new technology is motivated by the fact that DVB-SH networks without terrestrial repeaters can results in challenging propagation channels ("Land Mobile Satellite"). Signal outages in the order of several seconds may be quite frequent, in particular due to blockage. These outage durations generally exceed conventional physical or link layer error correction capabilities. The interleaving depth of such systems is generally in the order of several milliseconds for DVB-T or at most several hundreds of milliseconds with MPE-FEC. In order to counteract signal outages of several seconds, it is necessary to extend the protection capabilities of the forward error correction schemes. On may either use physical layer protection with very long interleavers or may use a link layer forward error correction scheme. DVB-SH provides both solutions: when transmitting to class 1 receivers, one should rely on a combination of short physical layer FEC and extended link layer FEC, referred to as IFEC and introduced in clause 6.2 MPE-IFEC of this clause; when transmitting to class 2 receivers, one may rely exclusively on a physical layer FEC with extended interleaving as described in clause 7; • the second one addresses the backward compatibility of the DVB-SH physical layer with existing DVB-H time slicing and related legacy features. These include Time-slicing signalling, power saving, variable bit rate support and statistical multiplexing support. Clause 6.3 Time-Slicing describes what level of compatibility (such legacy DVB-H features) is introduced by the DVB-SH physical layer for both terminal classes; • the third part deals with mobility aspects and IPDC features (clause 6.4 Mobility). 6.2 MPE-IFEC 6.2.1 Framework description The MPE-IFEC has been designed taking account: • the necessity to have a protection scheme to counteract the disturbances in DVB-SH transmission and reception environments; • to achieve this, the MPE-IFEC, contrarily to MPE-FEC, encodes over several time-slice bursts, which enables to increase the interleaver duration; • to address legacy requirements to existing DVB-H equipment; • to enable a flexible solution which permits service specific adjustments; • to provide the option to do further adjustments and optimizations during deployments. The features supported by the framework are presented below. ETSI 66 ETSI TS 102 584 V1.1.2 (2010-XY) • Compatibility with DVB-H link layer (MPE sections): MPE-IFEC is introduced in a way that it does not modify MPE section format, but only introduces one new sections type, the MPE-IFEC section. As MPE sections, MPE-IFEC sections convey almost identical real-time parameters. MPE-IFEC conveys exactly the same Time-slicing information as MPE. In addition, MPE-IFEC sections introduce additional information on the respective positions of MPE sections within the burst via the use of MPE_boundary bit (frame_boundary bit signals the end of the MPE-IFEC section part inside the burst). • Support of MPE-FEC: permits the concurrent use of MPE-FEC and MPE-IFEC sections in one burst. However the support of simultaneous MPE-FEC and MPE-IFEC decoding in DVB-SH is for further study and therefore the parallel sending of MPE-FEC and MPE-IFEC on the same elementary stream is not allowed. This choice can be done on a per elementary stream bases so that MPE-FEC can be applied, on the same transport stream, on one elementary stream while another has MPE-IFEC. MPE-IFEC also relies on the same Application Data Tables (ADT) as the MPE-FEC and the transmitter may use the ADT of the MPE-FEC as the ADST of the MPE-IFEC. As a consequence, the number of rows of MPE-IFEC ADST is the same as the number of rows of MPE-FEC ADT. In case of losses of MPE sections, the receiver may use MPE-FEC sections, (exclusive) or MPE-IFEC sections, depending on the time_slice_fec_identifier signalling, but not both simultaneously, to recover the missing data. • Support long interleaving: MPE-IFEC allows significantly enhanced performance (as measured by ESR5(20)) in LMS channels when compared to MPE-FEC as demonstrated by simulations. This performance benefits can be achieved as the MPE-IFEC encoding process spans several time-slice bursts. MPE-IFEC requires a burst numbering which is signalled in MPE-IFEC headers. • Support of different service requirements: The MPE-IFEC can be configured to enable a variety of configurations providing flexibility for the network operator. Guidelines are provided later in this clause on how to select these parameters. • Support of fast zapping: inter burst FEC protection is adversely affecting latency, and therefore also has influence on channel zapping times. This is due to the fact that for making use of all MPE-IFEC sections protection, the receiver must perform a late decoding and wait for reception of all MPE sections and MPE-IFEC sections relating to the encoding matrix to which a certain MPE section is assigned to. However, by sending MPE sections in each burst, immediate access and processing of these MPE sections is possible in case no errors have occurred. As reception continues, additional parity is received and MPE-IFEC protection can progressively protect the data by doing an early decoding. After some time, late decoding is possible. • Support a variety of FEC codes: the framework can support a variety of codes, currently Reed Solomon is the only code supported by DVB-SH, other FEC codes are for further study. The MPE-IFEC provides an inter burst protection. Compared to MPE-FEC, this is achieved by either: • the increase of the encoding matrix to sizes larger than one burst (one encoding matrix is filled during several successive bursts instead of a unique burst as in the DVB-H case); • the parallelization of the encoding mechanism (instead of using only one matrix, data are distributed over a number of B parallel matrices, among M possible); • the combination of the above both concepts. The framework encoding process is shown in figure 6.2. ETSI 67 ETSI TS 102 584 V1.1.2 (2010-XY) Every EP datagram bursts, EP datagrams encoding is calculated bursts for the the current ADTs Datagram Datagram k-D Burst Datagram k-1 Burst MPE-IFEC Process Burst k Datagram Datagram Burst k-D Burst k Each ADST is split in B parts, ADST Datagram each being placed in one ADT Burst size B ADT 0 ADT m ADT M-1 Datagram MPE-FEC Burst k-D (if used) Encoding Encoding Encoding iFDT 0 iFDT m iFDT M-1 S IFEC Burst Generation Sections coming from S iFDT Burst k and Time Slicing are interleaved inside the iFEC burst k Figure 6.2: MPE-IFEC encoding process The MPE-IFEC encoding process operates on a sequence of datagram bursts. Datagram bursts consist of a collection of datagrams, i.e. generally IP datagrams. Each datagram burst may be of different size and may comprise a different number of IP datagrams. For each datagram burst, a corresponding IFEC burst is created which encompasses MPE-IFEC sections from iFDTs of S encoding matrices. An MPE-IFEC section is comprised of a header, the parity symbols from one or multiple columns from the same iFDT, and a checksum. This IFEC burst is then multiplexed with all MPE sections of an original datagram burst, including the corresponding MPE-FEC sections if present. This multiplexing is done with the datagram just received in case the delay parameter D is set to D=0, or with a previously received datagram when D>0. This collection of MPE, MPE-FEC, and MPE-IFEC sections forms an MPE-IFEC time slice burst that is further processed. The sections within the IFEC time-slice burst are mapped to MPEG2 TS packets and then encapsulated in the time-slice bursts. Each datagram burst is mapped by an ADST (Application Data Sub Table) function on to the B ADTs (Application Data Tables) selected out of M parallel encoding matrices. Every EP bursts, an encoding process is carried out on one of the encoding matrices, generating from the ADT the corresponding iFDT (IFEC Data Tables) by applying an FEC encoding function. The resulting protection is achieved at the cost of some latency for generating and receiving parity data since the parity data is computed and spread over different bursts instead of one single burst in the MPE-FEC case. The link layer encoding key parameters are given below: • EP: this is the encoding period of the FEC process expressed in burst units. It refers to the frequency with which FEC is computed: an EP of 1 means that the encoding process occurs at every burst whereas an EP greater than 1 means that the encoding process occurs every EP bursts and the encoding matrix capacity is EP times greater so it takes EP times longer to fill the matrix with data. This EP normalizes all other parameters except D. • B: this is the encoding parallelization expressed in encoding matrix units. Every burst is split into B parts distributed over B parallel encoding matrices. So actual interleaving depth is B*EP bursts, or a computed FEC has correction capabilities spanning EP*B successive bursts. The B matrices used for the encoding are updated every EP bursts and the list is completely refreshed after EP*B bursts. ETSI 68 ETSI TS 102 584 V1.1.2 (2010-XY) • S: this is the depth of the FEC spreading factor; it means that produced FEC is interleaved over S iFDT. This enables to better protect the produced FEC. • D: this is the delay applied to the data. Since FEC is computed with the received data, the normal sending order (send data immediately) would imply sending the FEC after the data. This parameters influences the zapping quality but has impact on end-to-end latency (see clause 6.2.4). These four parameters enable to configure the encoding process in a flexible way. A typical configuration is obtained by setting EP=1 and B>1. This results in a sliding encoding process which enables to reuse several components of an MPE-FEC implementation: every received data burst is interleaved over B encoding matrices and one iFDT is computed applying the MPE-FEC Reed Solomon code. To familiarize the reader with the MPE IFEC concept, figure 6.3 provides an example encoding and sending process with EP=2, B=9, S=2, and D=0. The horizontal axis represents the M=11 encoding matrices, numbered from 0 to 10, and the vertical axis represents the continuous datagram and MPE-IFEC time slice burst numbers (right column from 0 to 33). ETSI 69 ETSI TS 102 584 V1.1.2 (2010-XY) Encoding matrices 0 1 2 3 4 5 6 7 8 9 10 0 0 B (9) 1 1 2 2 0 EP (2) 3 3 0 4 4 0 1 5 5 0 1 6 D (0) 6 1 2 7 7 1 2 8 8 2 3 S (2) 9 9 2 3 10 10 3 4 11 11 3 4 12 12 S (S*EP=4) 4 5 13 13 4 5 Time-slice bursts Datagram bursts 14 14 5 6 15 15 5 6 16 16 6 7 17 17 6 7 Filling time 18 18 7 8 (B*EP=18) 19 19 7 8 20 20 8 9 21 21 8 9 22 22 9 10 23 23 9 10 24 24 10 25 25 10 26 26 27 27 28 28 29 29 30 30 0 31 31 0 32 32 0 1 33 33 0 1 mapping of datagram burst (line #) over ADT (column #) encoding event of EM (column #) 5 carriage of FDT (column #) in time slice burst (line #) Figure 6.3: One example (EP=2, S=2, B=9) Figure 6.3 can be read either horizontally or vertically: • Horizontally: - the horizontal reading gives a representation of the sending logic:  the first two columns give the impact of the parameter D on the datagram burst sending: if D is set to 0 (here this is the case), the datagram burst is sent at the same time as the MPE-IFEC time slice burst. For instance, datagram 7 is sent with burst 7. Should the D parameter be greater than 0, then the datagram burst k would be sent in burst k+D; ETSI 70 ETSI TS 102 584 V1.1.2 (2010-XY)  for a given MPE-IFEC time slice burst number, if there is an orange cell on the corresponding line, it means that some FEC excerpted from the corresponding FDT is sent in the burst. For instance, at burst 7, MPE-IFEC sections from FDT 1 and 2 are sent. The number of "source" FDT interleaved in the current burst is dictated by the S parameter (here set to 2). The choice of the FDT matrix number is dictated by ifdt_function in MPE-IFEC specifications [attachment]; - the horizontal view indicates also how the datagram burst is mapped on the ADST:  those ADT that receive columns from the current datagram burst are represented by the cyan colour: for example ADT 0 to 8 receive data from datagram burst 0 and 1, ADT 1 to 9 receive data from datagram burst 1 and 2, etc.;  one can see the influence of the EP parameter: the list of ADT receiving data is "updated" every EP datagram burst received. The number of the ADT used for mapping is given by adt_function in MPE-IFEC specifications [attachment]. • Vertically: the vertical reading gives a representation of the encoding logic: - every EP bursts, there is an "encoding event" indicated by a cross: one particular matrix encodes the data (it generates the iFDT from the ADT). After the completion of the encoding, the ADT can be freed from its data so that this ADT can be used to host new datagram bursts again. For instance, encoding matrix 4 has an encoding event at MPE-IFEC time slice burst 9. Its data can be reset after then; indeed this ADT it is not used during the 6 following bursts, then it is again filled starting at burst 20; - the encoding event creates FEC information that is transmitted during a successive number of bursts that is equal to S*EP. For instance, FDT 4, computed before burst 9 is sent, is sent during bursts 10 to 13 in 4 pieces; - since all other parameters are normalized by EP, the change of list of encoding matrices happens every EP bursts and the list has completely be renewed after EP*B bursts, which is also the time it takes to fill a matrix. For instance ADT 10 starts to be filled at burst 4 and stops at burst 21, 17 bursts later, which makes 18 bursts overall. 6.2.2 Usage in the context of the DVB-SH In the DVB-SH context of the current implementation guideline release, only RS(255;191) (see EN 301 192 [9], clause 9.5.1) is used as encoding scheme. The MPE-IFEC framework is instantiated on this code with the following mandatory parameters: EP = 1, B ≥ 1, S ≥ 1, D ≥ 0. Again, to provide some insight, we discuss an even simpler example with B = 3 and S = 1 where a completely lost burst can be recovered with a link layer FEC code rate of only 3/4. Table 6.1 shows, for each burst, what happens at transmitter and receiver sides and represent the event of burst "n" complete loss: • on transmitter side: - the first row represents datagram bursts, the data to be transmitted; - second row represents the 4 parallel encoding matrices used by the link layer: ADT are filled by received datagram bursts and FDT are computed from the ADT; - third row represents the actually sent MPE-IFEC time slice bursts made of datagram bursts and IFEC bursts derived from one unique FDT. As S = 1, the IFEC burst is constructed from a single FDT; • on receiver side: - first row represents received time slice MPE-IFEC burst, possibly with their losses; - second row represents the 4 parallel decoding matrices: ADT are filled by received datagram burst and FDT are filled by received MPE-IFEC burst; - third row represents the data to be sent to the higher layer, whether it may be received "lossless" or recovered by FEC decoding. ETSI 71 ETSI TS 102 584 V1.1.2 (2010-XY) Table 6.1: example of complete burst recovery Step Transmission side Reception side n                                                                             Transmission of datagram burst n; previously computed FDT 3 is sent in MPE-IFEC time slice We assume previous bursts kM, during a number of burst equal to M-B, the recently encoded and reset ADT does not receive any data. When data are again inserted from burst k and ADT m, the following relationship is valid: adt_index(k+B-1,0)=m, or (k+B-1)[M]=m, or there exist an "i" such that k = M*i + m–B+1 The ADT will then be filled by next B bursts, so bursts k, k+1,…k+B-1 We have presented as an example in figure 6.6 the ADT 0 after first 20 datagram bursts have been received. In this figure, we can see that the last columns to have been "shifted" in the ADT 0 (the ones that are on the rightmost part), are coming from burst 20, and the first ones to have been shifted in are on the leftmost part (due to the "push" from right to left of newer columns) and are coming from burst 15, which respects the above relationship. Figure 6.6: ADST to ADT mapping (ADT view) It can be observed in this figure that the source ADST columns are grouped by "sub-block". This is due to the "shifting" function that inserts a certain number of columns from the same burst. Note also that the source ADST columns indexes are not continuous but separated by B (for example see column indexes coming from ADST 15: 5, 11, 17, 23, 29). ETSI 77 ETSI TS 102 584 V1.1.2 (2010-XY) The adt_column function More generally, a complete ADT ready for encoding has the following structure presented in figure 6.7. floor(C/B) floor(C/B) floor(C/B) floor(C/B)+1 floor(C/B)+1 floor(C/B)+1 0 1 … … … … floor(C/B) … … … … … … … … … (B-C[B])*floor(C/B) .. .. .. .. .. .. .. … … … … C-3 C-2 C-1 0 k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 1 1 1 1 1 1 2 2 2 2 2 2 … B B B B B B B … B B B B B B B B B B B B B B - - - - - - - - - - - - - - - - - - - - - C C C C C C C 2 2 2 2 2 2 2 1 1 1 1 1 1 1 [B] [B] [B] [B] [B] [B] [B] 1023 Figure 6.7: Generic ADT structure The ADT is comprised of B sub-blocks, each sub-block being consisting of a number of columns coming from the same datagram burst and corresponding ADST: • Each of the first (B-C[B]) sub-blocks hosts floor(C/B) columns from same ADST; the floor(C/B) ADST columns of the kth block, k∈[1;B-C[B]], are the columns j, j∈[0;C-1], such that k = B-j[B], so the formulation of the column numbers is the following: For k = 1:1:B-C[B], for l = 0:1:floor(C/B)-1, j(k,l)=l*B+(k-1) • Each of the last C[B] sub-blocks host floor(C/B)+1 columns of the same ADST; the floor(C/B)+1 ADST columns of the kth block, k∈[B-C[B]+1;B], are the columns j, j∈[0;C-1], such that k = B-j[B], so the formulation of the column numbers is the following: For k= B-C[B]:1:B, for l = 0:1:floor(C/B)-1, j(k,l)= l*B+(k-1) and j(k, floor(C/B))= floor(C/B)*B + (B-k) Hence, the shifting operation in the MPE IFEC specification for each ADST to ADT mapping in case of EP=1 may be realized as deterministic positions inside the ADT when the time to encode the matrix has arrived. These positions are given by the following formula: adt_column(j) = (B–(j[B]+1))*floor(C/B) + max(0;(C[B])-(j[B]+1)) + floor(j/B) The use of such adt_column formula is easier than implementing the "shifting" method since it establishes a bijection between ADST and ADT columns. An ADST can be then considered as a collection of C "pointers" to the ADT columns. This approach is presented in figure 6.7 where the column number of the original ADST are presented below the ADST number (this is the pointer origin) and the column number of the destination ADT are presented at the top of the column (this is the pointer destination). Mapping ADST results in setting the pointers destination to the correct ADT column using the adt_column function and this update occurs only once instead of B times with the original shifting mechanism. The advantages of this approach are twofold: • This helps storage optimization since storage is done inside the ADTs rather than on the ADSTs, each ADST being a list of C pointers to C columns in B ADTs. More memory implementation discussion that uses this pointer approach can be found in clause 6.2.6 Memory requirements. • This helps the decoding since when an ADT has been decoded and some of its columns been corrected, the ADST is automatically "refreshed" because the pointer points now to a correct column. More decoder implementation discussion using this concept can be found in clause A3.4.2. ETSI 78 ETSI TS 102 584 V1.1.2 (2010-XY) Padding: One ADT columns may not necessarily host data, but also padding bytes and padding columns. Due to the construction of the ADST function that positions padding columns at the end of the ADST, these padding columns always occur at the end of the "sub-blocks" and therefore padding columns are inserted inside the ADT at different positions, but certainly not at the end of the ADT as it is usually the case in MPE-FEC encoding. Assuming in our example that the 4 last columns of each ADST are concerned by padding (the first one is a data column with some data at the top of the column, whereas the 3 last ADST columns are complete padding columns), the ADT will have the following aspect whereby the shaded area represents padding. Figure 6.8: ADT aspect with interleaved padding columns The generation of the Reed Solomon operates on a matrix of 191 data columns, so there is an additional padding of 191- C columns so that resulting ADT is as shown in figure 6.9. Figure 6.9: Final ADT aspect The padding is equivalent to a code shortening as defined in EN 301 192 [9], clause 9.3.3.1. Code shortening is required to obtained the signalled ADT size K. 6.2.3.8 FDT generation and code rate computation As soon as the ADT has been constituted, the FEC encoding function is applied to generate the iFDT using the Reed-Solomon code (255, 191). The FEC columns can be as high as 64 but usually only a fraction of these FEC columns is actually transmitted over the air (the first ones), which is equivalent to code puncturing as defined in EN 301 192 [9], clause 9.3.3.2. Puncturing is required to obtain the signalled iFDT size N. With EP=1, for each processed datagram burst, exactly one encoding matrix is processed, i.e. the iFDT is generated. Definition: the actual code rate is computed on a per encoding matrix basis based on the structure of the ADT and the iFDT. So for each encoding matrix m, we have: nof _ data _ columns(m ) code _ rateactual (m ) = (nof _ data _ columns(m ) + nof _ fec _ columns(m)) ETSI 79 ETSI TS 102 584 V1.1.2 (2010-XY) The actual code rate is usually taken as near as possible to a code rate target, and expected to be the same for all encoding matrices: ∀m ∈ [0; M − 1], code _ rateactual (m ) ≈ code _ ratet arg et . This target code rate is used to compute the number of FEC columns within each iFDT:  1 − code _ ratet arg et  nof _ fec _ columns = ceil nof _ data _ columns *  .There are at least two ways for computing this  code _ ratet arg et    number of columns: Method 1: a fixed number of FEC columns is transmitted in each IFEC slice burst by assuming, for all m, nof_data_column(m)=C' where C'≤C and C' is fixed. Whatever the real completion of the ADT, the transmitted number of RS columns will remain the same. In CBR situation (nof_data_columns~C), this will generate a code_rateactual almost equal to code_ratetarget. But in case of VBR, it will make the code rate variable, possibly by large means, depending on variations of nof_data_columns datagram burst by datagram burst and encoding matrix by encoding matrix: when nof_data_columns is equal to C, code_rateactual is almost equal to the target value code_ratetarget, but when nof_data_colums is small compared to C, then code_rateactual may become quite low compared to code_ratetarget, nearing 0). Method 2: another way is to compute the nof_fec_columns based on the actual nof_data_column(m) and target code rate. This ensures a quasi constant code_rateactual, whatever the actual nof_data_columns. The precision of the difference between actual and target is a function of the nof_data_column and is better when this latter increases. These two options for the selection of the RS columns and their impact on code_rateactual are presented in figure 6.10. Code rate actual for the two methods as a function of nof_data_columns 70,00% 60,00% 50,00% 40,00% constant code_rate_actual 30,00% constant nof_fec_columns 20,00% 10,00% 0,00% 1 4 7 10 13 16 19 22 25 28 31 34 37 40 nof_data_columns Figure 6.10: Different strategies for RS columns generation In general for correct support of VBR and statistical multiplexing, the second option is recommended. The first options is recommended when a pure CBR traffic is protected. In our typical case, we assume a code_ratetarget of 2/3. Based on  1− 2/3 the first option, we will assume that the number of FEC columns is equal to ceil 37 *  = 19 .  2/3  ETSI 80 ETSI TS 102 584 V1.1.2 (2010-XY) 6.2.3.9 IFEC burst generation Once the RS columns are generated inside one FDT, these columns are stored and interleaved before being inserted inside a time slice MPE-IFEC burst. This is where the S parameter is used: the MPE-IFEC sections inside one IFEC burst are coming from S successive FDT and the index position of one section inside the IFEC burst enables to derive the original FDT. This spreading effect is shown in figure 6.11. It can be seen that, in every IFEC burst i+4, there are: • 5 IFEC sections coming from FDT i+3. • 5 IFEC sections coming from FDT i+2. • 5 IFEC sections coming from FDT i+1. • 4 IFEC sections coming from FDT i. More generally, given the MPE-IFEC time slice burst number k', k'∈[0;kmax-1], the IFEC section index i, i∈[0;18], the functions ifd_index and ifdt_column enable to derive the original iFDT index and iFDT column: ifdt_index(k',j)=(k'-j[S]+M)[M] ifdt_column(k',j)=j If nof_fec_columns(i) gives the number of FEC columns present in FDT i, the sending will be as follows: • in the nof_fec_columns(i)[S] first IFEC bursts, we send floor(nof_fec_columns(i)/S)+1 IFEC sections; • in the remaining S-nof_fec_columns(i)[S] IFEC bursts, we send floor(nof_fec_columns(i)/S) IFEC sections. One representation of IFEC burst i+4 is presented in figure 6.11. ETSI 81 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 6.11: Sending arrangement of IFEC burst i+4 ETSI 82 ETSI TS 102 584 V1.1.2 (2010-XY) Please note that the section index may not be consecutive and there could be index discontinuities when an iFDT has "run out of available columns". Such situations may occur in VBR situations with a fixed code rate: one ADT i has fewer data columns than the following ADT i+1 so that their corresponding iFDT have also different FEC columns available for transmission. Take the current example where the number of column is 19 on all iFDT except iFDT i where only 18 columns are available. The IFEC bursts composition are represented hereafter: Table 6.4: Source of MPE-IFEC sections in different IFEC bursts Source iFDT i-3 i-2 i-1 i i+1 i+2 i+3 i+4 IFEC burst (19) (19) (19) (18) (19) (19) (19) (19) i+1 4 5 5 5 - - - - i+2 - 4 5 5 5 - - - i+3 - - 4 5 5 5 - - i+4 - - - 3 5 5 5 - i+5 - - - - 4 5 5 5 … The MPE-IFEC time slice burst i+4 will have only 18 IFEC sections with the following indices. iFDT i+4 index 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 16 17 18 i i i i i i i i i i i i i i i i i i iFDT source + + + + + + + + + + + + + + + 3 2 1 3 2 1 3 2 1 3 2 1 3 2 1 Figure 6.12: IFEC burst I+4 in case of section discontinuity It is also to be noted that the interleaving over the S IFEC bursts will smooth IFEC burst variations. More scenarios on usage of the section indices are discussed in clause 6.2.7 MPE-IFEC usage scenarios. 6.2.3.10 MPE-IFEC section header In addition to the static configuration of the time_slice_fec_identifier, real-time signalling is conveyed in the MPE-IFEC headers. Figure 6.13 provides an overview on the syntax of the MPE-IFEC section and its header. ETSI 83 ETSI TS 102 584 V1.1.2 (2010-XY)    iFEC_header iFEC_data_bytes CRC iFEC_header iFEC_data_bytes CRC  iFEC_header iFEC_data_bytes CRC iFEC_header iFEC_data_bytes CRC iFEC_header iFEC_data_bytes CRC32 burst_number      fdt_index section_index    fdt_column  max_iFDT_column ≤    iFEC_burst_size ≤  section_length   FDTFDT Prev_burst_size  datagram_burst_size Figure 6.13: MPE-IFEC header and payload structure For our particular example, the indices have the following information: • burst_number = k', where k' ∈[0;249]. • section_index = j, where j' ∈[0;18]. • The iFDT index is obtained as m = ifdt_index(k', j). • IFEC_data_bytes are obtained from iFDT(m) and correspond to iFDT columns ifdt_column(k', j). • max_iFDT_column = floor(255*max_iFDT_column(m)/N)=floor(255*19/64)=75. • IFEC_burst_size = 19 (total number of IFEC sections included in this IFEC burst). • section_length = 1 037. • prev_burst_size = datagram_burst_size((k-D-j[jmax]-1+kmax)[kmax]). Usage of these fields is multiple and several examples are provided in clause 6.2.7 MPE-IFEC usage scenarios. Real-time parameters according to table 6.5 are inserted. These real-time parameters usage are detailed in clause 6.2.3.11 Burst sending arrangement as they relate to the sending arrangement within an MPE-IFEC time slice burst. Table 6.5: MPE-IFEC real-time parameters Syntax Number of bits Identifier real_time_parameters () { delta_t 12 uimsbf MPE_boundary 1 bslbf frame_boundary 1 bslbf prev_burst_size 18 uimsbf } ETSI 84 ETSI TS 102 584 V1.1.2 (2010-XY) 6.2.3.11 Burst sending arrangement Considerations on the burst, Time-slicing and Delta-t parameter In the MPE-IFEC specifications [MPE-IFEC], there is a reference to an (MPE-IFEC) time slice burst ([MPE-IFEC], clause 2.3.7). This MPE-IFEC time slice burst has the same definition as the burst in DVB-H, as specified in EN 301 192 [9], clause 9.1]: "a burst is a set of sections delivered on an elementary stream. Between two consecutive bursts there is a period of time when no sections are transmitted on the particular elementary stream. Each burst indicates the start time of the next burst within the elementary stream". This definition of burst includes potentially sections of different kinds, like MPE, MPE-FEC and MPE-IFEC, so that the burst is by nature an open concept. In [MPE-IFEC], the IFEC time-slice burst is therefore used to designate all the sections, including MPE, MPE-FEC and MPE-IFEC, that age timely grouped. For simplification and clarity reasons, we will in the following refer to the IFEC time-slice burst by the simple term burst, not to confuse it with the DVB-H time-slice burst that is referred by the term time-slice burst. Another kind of burst has been defined in MPE-IFEC specifications [MPE-IFEC] called the IFEC burst. This includes the only MPE-IFEC sections. The terminology is presented in the table below. Table 6.6: Terminology Name in the present Concerned section Name in DVB-H Name in IFEC document kinds Burst All burst Burst or MPE-IFEC time slice burst Time slice burst MPE and MPE-FEC Time slice burst N/A IFEC burst or MPE-IFEC MPE-IFEC N/A IFEC burst or MPE-IFEC burst burst Contrarily to the burst, the definition of Time-slicing given in the same document ("Time Slicing: Method to deliver MPE sections and MPE-FEC sections in bursts") refers to the only MPE and MPE-FEC sections, excluding any additional types of sections (like MPE-IFEC). Additionally, the definition of Delta-t given in the clause 9.10 makes a clear reference to a time-slice burst ("The field indicates the time (Delta-t) to the next Time Slice burst within the elementary stream"). Note that Delta-t information provided by MPE-IFEC sections must be exactly the same as Delta-t information provided by MPE and MPE-FEC sections since, otherwise, the Delta-t provided by different sections within same burst-MPE/MPE-FEC on one side, MPE-IFEC on the other side could point to different positions within the next burst and hence create inconsistent signalling. This is why the definition of the Delta-t field in MPE-IFEC sections is exactly the same as the definition of the Delta-t field in MPE and MPE-FEC sections: "Delta-t is a time offset indicating the time from the start of the transport packet carrying the first byte of the current IFEC section to the start of the transport packet carrying the first byte of next time-slice burst". Therefore, the burst (union of the time-slice burst and the IFEC burst) and the time-slice burst may not coincide in time due to the exclusion of IFEC sections from time-slice burst definition while, at the same time, their Delta-t signalling points to the same temporal structure, the time-slice burst. Some examples are given below to clarify, in the case when no MPE-FEC is used in the elementary stream (please note that simultaneous used of MPE-FEC and MPE-IFEC is not allowed and is for further study). ETSI 85 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 6.14: Burst sending arrangement (w/o MPE-FEC) In figure 6.14, we can see three successive bursts of the same elementary stream (bursts i to i+2): • Delta-t signalling throughout the same burst always points to the first MPE section of the following burst, and not the first MPE-IFEC section, because Delta-t has the same definition, whatever the section, and always refers to the time-slice burst; • as a consequence of this signalling, if the burst begins with MPE-IFEC sections, they may be missed, unless the delay T between their sending and the actual sending of the time-slice burst is small compared to the Delta-t unit (10 ms). • the time-slice burst and the burst do not necessarily coincide in time. Many configurations are possible and, in this example, the only burst i+3 sees the matching between both. Recommendations: Since the burst can differ from the time-slice burst but the Delta-t signalling is in any case related to the time-slice burst, the following recommendations are given for cases when no MPE-FEC decoding applies: • At least one MPE section must be positioned in the vicinity of each burst beginning so that no MPE-IFEC sections are missed. The vicinity is considered such that the delay between the burst and the time-slice burst beginnings is below 1/10th of the Delta-t unit so 1 ms. Depending on the transmission bit rate, this time may allow to send an MPE-IFEC section or not. For instance with a bit rate of 2,4 Mbits, the time it takes to send 1 TS is 0,623 ms which allows sending 1 full MPE-IFEC section in the only case of 256 rows. For simplification, the first section should therefore be an MPE section. • Since the only MPE-IFEC section conveys burst numbering information, at least one MPE-IFEC section should be positioned close to the beginning of the burst, such that the receiver may detect the burst number as early as possible while receiving the burst, and position immediately the received MPE sections in their correct ADT, without waiting for reception of late MPE-IFEC sections. Note that this recommendation is not a hard one since it is always possible to receive MPE sections without previously knowing their burst number: these sections are stored until their burst number can be determined by the receiver, by receiving a later MPE-IFEC section conveying this burst number or other means, for instance based on timing. ETSI 86 ETSI TS 102 584 V1.1.2 (2010-XY) • After the beginning of the burst, the MPE and MPE-IFEC may be freely interleaved during a time interval not exceeding max_burst_duration signalled in time_slice_fec_identifier. After max_burst_duration, no more sections (MPE, MPE-IFEC) must be present. The receiver must ignore these sections received lately and can switch off the receiver, hence enabling power saving. • To detect the end of the burst, the receiver must check that the end of the time-slice burst and IFEC burst are reached. For that purpose, the receiver checks the received sections flags and looks whether one of the conditions presented in table 6.7 are respected for all lines. These conditions are the results of time-slice burst sending arrangement rules found in EN 301 192 [9], clause 9.10 ("All sections carrying Application data datagrams of a given MPE-FEC Frame shall be transmitted prior to the first section carrying RS-data of the MPE-FEC Frame (i.e. sections carrying Application data datagrams shall not be interleaved with sections carrying RS-data within a single MPE-FEC frame).[…] Within an elementary stream, sections delivering data of different MPE-FEC Frames shall not be interleaved. […] Note that for each MPE-FEC Frame, MPE sections are delivered before MPE-FEC sections") and MPE-IFEC time-slice burst generation and sending arrangement rules found in MPE-IFEC specifications [MPE-IFEC], clause 3.6. Table 6.7: Conditions for end of IFEC burst detection Condition 1 Condition 2 Condition 3 Condition 4 Condition 5 MPE An MPE section An MPE-FEC An MPE-IFEC Current time has Another service with a section has section has been exceeded previous PID has been table_boundary been received with an Delta-t plus found, signalling set to 1 has been received. MPE_boundary max_burst_duration. change in the received. set to 1. service MPE-FEC An MPE or MPE- - An MPE-IFEC reception. (this is for FEC section has section has been information only been received with received with an Time-slice burst since MPE-FEC the MPE_boundary cannot be frame_boundary set to 1. present at the set to 1. same time as MPE-IFEC in the ES) MPE-IFEC An MPE-IFEC - - IFEC burst section with frame_boundary set to 1 has been received. This table can be used as follows on a particular case where the IFEC burst is made of a first MPE section, then one IFEC section, then all remaining MPE sections, then all remaining IFEC sections. • In case no section at all is lost: - Time-slice burst end will be known with the last MPE section (condition 1). - IFEC burst end will be known with the last IFEC section (condition 1) • In case some MPE sections, including the last, and some IFEC section, but not all, are lost: - Time-slice burst end will be known with the first MPE-IFEC section received after the last MPE section (condition 3). - IFEC burst end will be known with the last IFEC section (condition 1). • In case some MPE-IFEC sections, including the last, are lost: - Time-slice burst end will be known with the last MPE section (condition 1). - IFEC burst end will be known with the time out (condition 4). ETSI 87 ETSI TS 102 584 V1.1.2 (2010-XY) • Note that one consequence of these rules is the need for the receiver to potentially stay on after last MPE section has been received (to receive MPE-IFEC sections if any) or after last MPE-IFEC section has been received (to receive MPE and MPE-FEC sections if any). This specific behaviour must be applied on a per stream basis: for those streams where no IFEC protection is applied, the same rules as DVB-H will apply (when the last MPE or MPE-FEC frame has been received, the receiver will not have to wait for other MPE- IFEC sections and so will be able to switch off immediately). 6.2.4 Parameters selection 6.2.4.1 Introduction This clause provides explanations and recommendations for MPE-IFEC parameter selection. To configure the MPE-IFEC, the operator needs to fix the 3 main parameters of the MPE-IFEC framework, namely B, S, D and the code rate: D is selected based on heuristics and the impact on zapping time, B and S are selected based on latency (B+S) and the code rate is selected on performance criteria. All performance results are measures at interface R1 shown on with a D equal either to 0 or B+S. The reader will see how these parameters can be easily established depending on the sought performance and the channel behaviour. For that purpose: • we first discuss the D parameter and its dependence with other B and S in clause 6.2.4.2 Recommendations on D; we show that D can be selected between only two typical values; • we then discuss how B and S can be derived in clause 6.2.4.3 Recommendation on B and S from link layer latency and code rate; • we finally show how link layer latency and code rate can be selected based on simulation abacuses in clause 6.2.4.4 Selection of M and code rate; • we also give a possible global procedure in clause 6.2.4.5 Parameter overall selection logic. We introduce the following latency definitions: • "encoder latency/delay": time between reception, at the encoder side, of an IP packet and actual emission of the last (data or IFEC) section related to this IP packet; the data section is the one carrying the IP packet whereas the IFEC section has correction capability on the ADT column(s) where the IP packet is located; the encoder latency can be computed over a set of IP packets belonging to one burst; we then talk of "burst encoder latency":  MPE section sending time  encoder_latency(IP_packet) = last MPE - IFEC section sending time  − IP_packet arrival time    encoder_latency(burst ) = max (encoder_latency(IP_packet) for all IP_packets belonging to the burst ) • "receiver latency/delay": this is the time between delivery of an IP packet at the decoder side and reception of the first information on this IP packet, whether it is the MPE data section carrying this IP packet or an IFEC section providing parity for an ADT column that includes data bytes for this IP packet. The receiver latency can be computed on a complete burst, we then talk on the burst receiver latency.  MPE section arrival time    receiver_latency(IP_packet) = IP_packet delivery time − first MPE - IFEC section related to an ADT column   that includes data bytes for this IP packet arrival time    receiver_latency(burst ) = max(receiver_latency(IP_packet) for all IP_packets included in the burst ) • "end-to-end latency/delay": time between delivery of an IP packet at the decoder side and reception of the IP packet at the encoder site. The end-to-end delay is equal to the encoder latency, the transmission delay and the receiver latency. So end-to-end delay varies only due to the receiver latency since encoder latency and transmission delays are considered as fixed. In steady state, the end-to-end delay is equal to its maximal value, but during non steady states (like zapping) this end-to-end delay may be shortened due to the possible reduction of receiver latency. ETSI 88 ETSI TS 102 584 V1.1.2 (2010-XY) 6.2.4.2 Recommendations on D B,S and D are related through the function giving M = B + max(0,S − D) + max(0, D − B) . This M, in addition to giving the number of encoding matrices, gives the receiver latency and the memory requirements: • Assuming the simplification that burst are regularly repeated with a temporal period of burst_repetition_interval, the receiver latency at the jitter free interface is equal to burst_repe tition_int erval* M . The jitter-free interface is the R1 interface when late decoding is used. • Assuming that ADT and FDT sizes are given by ADT_size and FDT_size, the memory required for complete IFEC decoding is equal to (ADT_size + FDT_size ) * M . These formula show the importance of the parameter D and its impact on user perception (via the latency) and terminal sizing (via memory), all other things being equal. For instance for a B equal to 6 and a S equal to 4, we obtain the curve presented in the figure 6.15. M(D, [B=6,S=4]) 10 8 6 M 4 2 0 0 5 10 D Figure 6.15: M(D, [B=6, S=4]) It can be seen that: • when D is equal to 0, M is rather large and equal to B+S=10; this configuration corresponds to the situation when the MPE-IFEC sections are always sent after the data they protect, which is good for error correction performance. But the high M value implies long jitter-free receiver interface delay and large memory requirements; • a minimum value of M is obtained for D taken in the range [4; 6]. Obviously this enables to reduce memory requirements in receiver and latency since the receiver needs to wait less for MPE-IFEC section reception at the jitter-free interface compared to when the sections are not sent with the data they protect. However the MPE-IFEC sections will be mixed with the datagram bursts, which reduces performance because an error affecting a burst likely affects at the same time, the data it conveys, and the FEC that corrects the errors. Increasing D will also increase end-to-end latency; • after the value 6, M increases again. Data and MPE-IFEC are still mixed, up to the situation when D=B+S (10 in our case). In this case, the data and IFEC sections are again not mixed, which provides good performance results. The receiver jitter-free latency is the same as D=0 but since the IFEC always precedes the data, this enables "early decoding" during zapping (see clause 6.2.5.2 Zapping time performance). The influence of the D parameter on different criteria is presented in table 6.8. ETSI 89 ETSI TS 102 584 V1.1.2 (2010-XY) Table 6.8: Influence of the selection of D parameter on different criteria D value Performance Receiver latency Encoder latency Receiver memory D=0 + - + 0 D=Dmin_sizing - 0 0 + D=B+S + + - 0 NOTE 1: Dmin_sizing is the lowest D that minimizes the M function. For instance, on our previous example, Dmin_sizing = 4. NOTE 2: "+" means the criteria is matched with a good performance, 0 means a neutral performance and "–" means a worse performance. NOTE 3: Performance is the correction capability at the jitter free interface; receiver latency includes both jitter-free and non jitter-free (during zapping); all other criteria are classical. NOTE 4: D=B+S provides a good compromise in performance, including receiver latency during zapping, at the detriment of encoder latency. This configuration should be used for contents not delay constrained whereas D=0 should be kept for delay constrained contents, at the detriment of receiver latency and zapping time. 6.2.4.3 Recommendation on B and S The code rate is defined on a per encoding matrix basis by the formula combining, for a given encoding matrix m, the number of ADT data columns (adt_data_columns(m)) and the number of FDT FEC columns (nof_fdt_fec_columns(m)). The target code rate is defined in clause 6.2.3.8 FDT generation and code rate computation. We also assume, as in this clause, that the same target code rate applies to all encoding matrices. Code_rate can be selected with complete freedom within the range authorized by C (using padding) and R (using puncturing). If we take our current example, C is fixed to 40 but the number of data columns is on average 37. Depending on the number of FEC columns (from 0 to 64), 64 code rates from 1 to 0,36 are possible as presented in clause 6.2.3.8 FDT generation and code rate computation. However, an exhaustive simulation campaign determined that the configuration offering the best performance  B  (B,S,code_rate)actual follows the rule   ≈ code_rate actual . Based on this assumption, near optimal parameters  B + S  actual Bopt and Sopt can be derived from target FEC rate and B+S for cases where D=0 by using the following formulas: S opt = ceil((1 - code_rate) * (B actual + S actual )) B opt = (B + S) − ceil((1 - code_rate) * (B actual + S actual )) Since B and S can be easily derived from the couple (B+S), which is equal to M when D=0 and noted M(0,B,S), and the code rate, we will use these two parameters in the following. 6.2.4.4 Selection of M and code rate Simulation campaigns have enabled to derive the rule between M(0,B,S) and the code rate for a given ESR5(20) fulfilment, on a per channel basis. These complete results are presented in the clause A.12. We take the example of a suburban channel with figure 6.16. ETSI 90 ETSI TS 102 584 V1.1.2 (2010-XY) 60 50 Interleaver Length (s) 40 30 20 10 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Supported Bitrate (Mbps) QPSK 16QAM NOTE: M(0,B,D)=B+S for D=0 and D=B+S; the curve are valid for these two values of D. Figure 6.16: SH-A, Class 1 - QPSK 1/2 and 16QAM ¼ - LMS-SU - 50 kmph - 63 dBW EIRP Satellite Figure 6.16 can be interpreted as follows: • all points on the curve match an ESR5(20) fulfilment of 90 %; • all points above have a better quality; • all points below have a lower quality. The curve is then an iso-ESR5(20) one. The iso-ESR5(20) curve has always the same shape with a slope that is increasing as code rate increases, an inflexion point and then an asymptotic point. In the example given, the following points can be found: • first value is around 1,7 Mbps/5 s (interleaver length directly related to M); • the inflexion point is at 2,55 Mbps/25 s; • the asymptotic value is at 2,75 Mbps/50 s. Two different approaches can be followed for determining which parameters must be selected for a particular system: • the terminal is memory constrained (compared to the asymptotic value) to a certain M(0,B,S), for instance 25 s: the selected code rate is the code rate found at the intersection of the iso-ESR5(20) curve and the horizontal line at the target M(0,B,S); in our case the resulting useful bit rate is 2,5 Mbps, which turns into a code rate of 2/3 because the total bit rate in 5 MHz QPSK1/2 GI=1/4 is 3,357 Mbps. These bit rates are measured at the R2 interface at MPEG2 TS layer; • the terminal is not memory constrained (compared to the asymptotic value), the objective is to maximize the capacity and so position at the inflexion point of the curve. In our case, the target capacity could be 2,7 Mbps (code rate 4/5); the M(0,B,S) is found at the intersection of the curve and the vertical line at target code rate. In our example the value is 50 s. A mobile system is also designed to work in a set of different propagation channels. The selection of the link layer parameters must be done for the worst case, typically LMS-ITS. Once selected, the configuration should perform well under less challenging channels. We give the complete example below. ETSI 91 ETSI TS 102 584 V1.1.2 (2010-XY) We start with an LMS-ITS dump excerpted from clause A.12. 60 50 Interleaver Length (s) 40 30 20 10 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Supported Bitrate (Mbps) QPSK 16QAM Figure 6.17: SH-A, Class 1: - 16QAM 1/4 and QPSK 1/2 - LMS-ITS - 50 kmph - 63 dBW EIRP Satellite If we are with a M(0,B,S) equal to 25, this gives a code rate of 0,44. We can see immediately on this figure that the performance will be much higher than the target in LMS-SUB (we are above the curve). Finally, since we are in the hybrid frequency in that case, we can also check the performance under TU6 channel: in clause 6.2.5.1.4 TU6 for hybrid frequency we show that a configuration for LMS-ITS or LMS-SU performs perfectly under TU6 channels. 6.2.4.5 Parameter overall selection logic So the general parameter selection logic is presented in figure 6.18. Worst channel model selection (LMS-x) Select required quality (1-ESR5%) Is the receiver memory constrained ? YES NO Select (B+S) compliant with memory Select target fec ratio Define fec ratio Define (B+S) (based on simulation/trials results) (based on simulation/trials results) Compute Bopt=f(B+S,fec ratio) Sopt=f(B+S, fec ratio) Figure 6.18: link layer parameter selection logic ETSI 92 ETSI TS 102 584 V1.1.2 (2010-XY) This selection logic is valid for D values such that MPE-IFEC sections are not transported in the MPE-IFEC same time-slice burst as the MPE they are protecting, so for D=0 and D≥B+S. For intermediate values, the performance will be slightly less requiring higher B+S and/or lower code rate and is for further study. This procedure is used to recommend parameters in clause 6.2.5.1.3 Recommended class 1 configuration. 6.2.4.6 Recommendations on signalling parameters The following recommendations are made on the way to configure the signalling fields: - max_burst_duration shall not exceed the repetition_interval to avoid confusing the receiver reception; 6.2.5 Simulated performance Link layer simulation configuration elements can be found in clause A.6. All detailed simulations results can be found in clause A.12. We give in this clause a summary of the clause A.12 results. 6.2.5.1 ESR5(20) performance This clause excerpts some results from clause A.12. The logic for these results is the following: • targeted quality: - for LMS channels, ESR5(20) fulfilment of 90 % is sought; - for terrestrial (TU6) channels, ESR5(20) fulfilment of 99 %, equivalent to a FER of 1 % is sought; • memory constraints: - depending on terminal category (handheld or vehicular), the memory constraints are different; vehicular terminal can be considered to have "unlimited memory" whereas handheld receiver can be considered to have memory limitations; • code rate: - the target code rate is the one that enables to support similar spectral efficiency as the class 2 receivers (full protection at physical layer); for instance, a QPSK1/2 and 16QAM1/4 will have a target link layer code rate of 2/3; - if the quality criteria for a maximum B+S value can not be fulfilled, then the code rate is reduced until the quality is achieved. The objective of the simulation procedure is to find the ((B+S);code rate) combination that enables to meet expected quality of service under the double constraint of (B+S) stays below maximum memory and code rate does not drop below 30 % of capacity loss. This exercise is performed for different channel environments and receiver categories. 6.2.5.1.1 LMS-ITS 50 kmph Please refer to clause A.12 for details. There is always an asymptotic value for the interleaving duration, around 25 s to 30 s: increasing the duration beyond this value does not bring any help. The recommended configuration for the link layer is therefore given in table 6.9. ETSI 93 ETSI TS 102 584 V1.1.2 (2010-XY) Table 6.9: Recommended link layer configuration in ITS (50 kmph) Hybrid Frequency C/N used for Recommended B+S Recommended Code modulation simulation rate SH-A 11,2 dB 25 ½ (QPSK1/2 and 16QAM1/4 16QAM2/7 GI=1/4) SH-B 12,3 dB 25 2/3 (TDM 8PSK1/3, QPSK1/2) Additional configurations are for further study, in particular sensitivity of this results as a function of C/N. 6.2.5.1.2 LMS-SU Please refer to clause A.12 for details. An interleaver of 25 s gives the asymptotic performance. Minimum recommended configuration is given in table 6.10. Table 6.10: Minimum configuration in SU (50 kmph) Hybrid Frequency C/N B+S Code rate modulation All >8 25 ¾ SH-A (QPSK1/2, <8 25 2/3 16QAM ¼) 6.2.5.1.3 Recommended class 1 configuration Based on previous results and procedure explained in clause 6.2.4 Parameters selection, the following class 1 configurations are proposed. Table 6.11: Recommended class 1 configuration Hybrid Frequency Threshold C/N modulation SH-A 12 SH-B 11 For C/N below the threshold, the following configuration shall be chosen B+S=25 and code rate = ½ whereas for C/N above the threshold, B+S=25 and code rate = 2/3 can be chosen. For services which require lower end-to-end latency, lower values of B+S may be used. The appropriate configuration can be derived according to the above procedure (in clause 6.2.4 Parameters selection) and using simulation results in clause A.12. The parameters can be freely chosen provided the required memory, as computed in clause 6.2.6 Memory requirements, does not exceed the class 1 memory capability as given in clause 10. 6.2.5.1.4 TU6 for hybrid frequency This clause refers only to the SFN case: having selected a class 1 configuration, we know its performance under LMS-ITS and LMS-SU channels. We need now to check also the performance over the terrestrial TU6 channel since the same MPE-IFEC configuration is also used over the terrestrial repeaters because of the SFN nature of the network: will the user experience different quality while performing a hand over between the hybrid and non-hybrid channels under the same CGC network (the hybrid frequencies do have an IFEC protection, which is not supposed to be the case of a non-hybrid frequency)? To evaluate the performance difference, we assume that CGC is working at C/N that enables ESR5(20) fulfilment of 99 % without MPE-IFEC and we need to check at this C/N if the hybrid signal, having additional MPE-IFEC protection but lower physical layer code rate, will have lower, same or better quality than non MPE-IFEC protected ones. ETSI 94 ETSI TS 102 584 V1.1.2 (2010-XY) As an example, excerpted from clause A12, we use a terrestrial configuration in QPSK1/3 that has an error-free C/NNHF at 3 kmph of 3,5 dB and 1,5 dB at 50 kmph. We now check the performance of the hybrid frequency (still at 3 kmph/50 kmph respectively) but with lower physical code rate (1/2) and the selected link layer configuration (e.g. 2/3), for a 25 s interleaver. It can be checked in clause A12 that the MPE-IFEC hybrid frequency will need 1 dB more than equivalent physical layer configuration at 3 kmph (the required C/N is 4,5 dB) and 2 dB at 50 kmph. So the channel in hybrid frequency will require a slightly higher C/N than the equivalent one of the non hybrid frequency to ensure same quality as the non hybrid one. The MPE-IFEC compensate partly the higher physical code rate. Note also that, due to the SFN between satellite and CGC, this added C/N can be compensated in areas where the CGC coverage starts to be less efficiently received. Such SFN gain is introduced in clause 11. 6.2.5.1.5 TU6 for non-hybrid frequency We check the interest of the MPE-IFEC for non-hybrid channels (content that is not broadcasted over the satellite, only sent over the terrestrial network). For this purpose, we display the useful bit rate as a function of the C/N in clause A.12 and see with MPE-IFEC how dBs can be traded for bandwidth. This figure shows that a network operator, we can trade capacity with dB: • in QPSK1/3: - 20 % capacity with 1 dB of C/N. • in QPSK1/2: - 15 % capacity with 1 dB of C/N (for comparison, with MPE-FEC we would need 33 % for 1 dB gain); - 30 % of capacity with 2 dB. • in 16QAM1/5: - 13 % capacity with 1 dB of C/N; - 25 % of capacity with 2 dB of C/N. • in 16QAM1/4: - 25 % capacity with 1 dB of C/N. Generally speaking, the short physical interleaver performs well in TU6 environment and MPE-IFEC enables to provide intermediate C/N operational points in between the values, in a similar way as, but better than, MPE-FEC. 6.2.5.2 Zapping time performance 6.2.5.2.1 Introduction This clause deals with zapping performance for only class 1 terminals. For a more general zapping performance discussion, including class 2 please refer to clause 6.3.2 Zapping time impact. However, some definitions are given or recalled that are also applicable for both classes. After these definitions, we explain the concept of zapping in an MPE- IFEC context and then give a complete example. These are conceptual considerations, zapping simulation results is for further study. 6.2.5.2.2 Definitions The definitions are either new ones or precision on already given definitions in the light of the newly defined. Précised definitions: Late decoding: refers to the techniques used for decoding with maximum protection. The receiver latency is always maximal, the IP packet being delivered to the video decoder only when the maximum protection is achieved (received). Late decoding time implies the end-to-end delay is always maximized whatever the situation because the receiver ETSI 95 ETSI TS 102 584 V1.1.2 (2010-XY) latency is maximized. Late decoding is explained for class 2 in clause 7.3.3.5.4 and for class 1 in clause A.6. Only class 2 can afford late decoding and fast zapping. Early decoding: refers to the set of techniques used for decoding with less protection than the maximum one. The receiver latency may be reduced compared to the late decoding case, the IP packet being delivered only when the minimum parity has been received and not all the parity. This enables to lower end-to-end delay and accelerate IP packet delivery in case of good reception condition but this induces variable end-to-end delay when more parity is requested during fading events. Early decoding for class 2 is explained in clause 7.3.3.5.5 and for class 1 in clause A.6. Early decoding is requested for class 1 for supporting fast zapping. New definitions: Zapping instant: time when the user selects a new program. Zapping time/delay: delay between the zapping instant and the time this program is actually received by the video decoder (we take into account only the link layer latency); by definition zapping time is mostly equal to the receiver latency. Zapping time can depend on many parameters like air interface configuration (physical and link layer interleaver: do we use a uniform/late profile, is D>0, etc.) but also on local reception condition (is reception quality good?) and receiver strategy (does the receiver supports fast zapping?). In lossless reception situation, zapping time can be as short as a burst reception for terminals applying fast zapping (class 1 with any D value but preferably >0 and class 2 with a uniform late profile), or as long as encoder latency for terminals without any fast zapping support (class 1 with late decoding strategy or without the capability to recover the parity later-on and class 2 with uniform long profile). In bad reception condition, zapping time can exceed these values and last as long as the signal blockage. Fast zapping: techniques used for reducing zapping time to acceptable delay even in the presence of long interleavers, be they at physical or link layer. They generally involve specific waveform configuration (uniform late for class 2, D preferably >0 for class 1), terminal specific behaviour (late decoding for class 2 and early decoding plus parity recovery for class 1). Not taking into account other delays in the system (video decoding, etc.), the fast zapping time can be as low as the time to receive a single burst. • In class 1 context, early decoding is usually used for fast zapping: under good reception condition, it is always possible to display content immediately since datagram bursts are transmitted as usual. The receiver can watch the video without waiting for late decoding latency, but may not have all the parity information to sustain a potential loss. Air interface configuration (use of D parameter) and parity recovery techniques can help the receiver to recover the parity progressively so that after a certain period, the receiver has recovered same full protection as in late decoding. • In class 2 context, fast zapping refers to the interleaver uniform/late configuration in conjunction with late decoding that enables displaying first image immediately under good reception condition; parity is progressively recovered and after the encoder latency, full protection is ensured. This technique is described in clause 7.3.3.5. Parity recovery: technique used by a receiver applying early decoding to progressively recover the same level of protection as late decoding one. For the link layer, different techniques are presented in clause 6.2.5.2 Zapping time performance. Note that parity recovery is not requested for the uniform/late physical interleaver since fast zapping is compatible with the late decoding as explained in clause 7.3.3.5.4. 6.2.5.2.3 Principles for zapping time analysis in class 1 The typical following scenario happens: • the end user zaps by selecting one PID on a list, the PID being chosen via the ESG; • the receiver decodes the MPEG2 TS and looks for corresponding PID; • once it has found the corresponding PID, it starts receiving MPE and MPE-IFEC sections; • with the included Delta-t, it can delineates the time-slice bursts and achieve power saving; • it stores the received sections in their respective bursts, maps them on their respective ADTs and iFDTs so that the Encoding Matrices are progressively "populated"; • after a certain delay, the receiver can decode and deliver IP packets to the video player. ETSI 96 ETSI TS 102 584 V1.1.2 (2010-XY) The zapping delay depends on this delay and different receiver behaviours are possible: • late decoding: - the receiver will wait for all the parity before decoding; - the waiting delay will depend on the sending arrangement:  it will be equal to M when D=0;  it will lower to a minimum value when D equals Dmin_sizing;  for larger D, the delay will be kept to this value; - for instance in the example case (B=6, S=4), the initial delay is 10 when D=0 and 6 when D=10. Thanks to this delay, all the parity coming from the data part can be processed and, due to the fact that the FEC parity precedes the data, the FEC is also available; - however, with late decoding, fast zapping is then not possible (we need to wait for Dmin_sizing): we need to resort to early decoding; • early decoding: - the receiver does not wait for all parity before decoding; the receiver can start decoding before parity has been received, for instance as soon as it has received a burst. Obviously, this has a cost since the protection may not be enough to protect against normal impairments; - as bursts are received, if D is large enough, since FEC is sent before data, FEC will be accumulated to provide some protection in conjunction with previously received burst (for FEC computation, current burst is interleaved with B-1 previous bursts). However, full protection is not achievable since the protection of the currently received burst depends also on following B-1 bursts (the interleaving is a convolutional one); - for achieving normal protection, the receiver must perform a late decoding and, for this, has to wait for same delay as in late decoding, for instance 5 burst in addition of the current one; - this additional waiting time can be "spread" over time thanks to recovery techniques and/or split with an initial small additional buffering. During this "spreading time", the end-to-end delay will not be constant and will increase from the initial value to the maximum one. So the interface is not jitter-free at the beginning. One technique for achieving this spreading is presented in [i.38]: the data from the link layer is delivered with a reduced rate to the IP layer. A media decoder may be informed to slow down the media playout. More precisely, after zapping, the data signal partially corrected by the MPE-IFEC with a slightly slowed rate than the transmitted rate is played out; the receiver corrects the signal by the MPE-IFEC, and, after a certain period, inversely proportional to the slowing rate, switches to the fully protected signal, at the transmitted rate. [i.39] suggests that 20 % of speed reduction is hardly perceivable by the user, allowing full protection and display at the transmission rate after 25 s, for a B equal to 6 (we need to recover B-1 bursts in addition of the initial one, so 5 bursts and, with a repetition interval of 1 burst per second, this leads to 5 s recovered in 5/0,2 = 25 s). Such techniques may require modification to existing media decoders, and especially audio ones; - it is also important to mention what happens if the receiver detects that during the slow down, the FEC is not sufficient. Then an immediate rebuffering may be reasonable. To summarize, correctly decoding with a long interleaving imposes a buffering delay that is a function of D and can be lowered to values typical of B-1 in addition of the current burst reception. However in many cases, the burst received at zapping instant is correct and can be forwarded to the media player for immediate decoding and display. This action is referred to a "fast zapping". If the terminals plays the signal at normal speed, the IFEC decoding will never benefit from the B-1 datagram bursts that are also used to create the parity. So techniques for spreading the delay are considered. EXAMPLE: We give hereunder a first analysis of these techniques, assuming only MPE-IFEC is used and excluding therefore MPE-FEC. If we consider an encoding depth of B, a FEC spreading of S and a given code rate, we have the following theoretical characteristics during the zapping time on MPE-IFEC time slice burst k: ETSI 97 ETSI TS 102 584 V1.1.2 (2010-XY) • at zapping instant, if D=0 or D>B+S, the MPE-IFEC time slice burst k is received without any protection; typical probability of having an erroneous burst before MPE-IFEC decoding is in the order of 20 % to 30 % in worst LMS-intermediate-tree-shadowed cases but only 8 % in LMS-suburban; • for subsequent MPE-IFEC time slice bursts, more protection is received that enables better recovery as, in average, more data and FEC for the corresponding encoding matrices is received through early decoding (decoding without having received full parity, either FEC or DATA): - if D≥B+S, between zapping instant (MPE-IFEC time slice burst k) and MPE-IFEC time slice burst k+B+S-1, useful MPE-IFEC sections are progressively received up to full reception when MPE-IFEC time slice burst k+(B+S-1) is received); - between zapping instant (burst k) and burst k+B-1, MPE sections are received that progressively fill the ADTs up to a state of partial completion at burst k+B-1 that then stagnates at this level: even if we wait indefinitely, the ADTs will never be completely filled, which could be considered as a loss, because the missing columns are coming from datagram burst to be received after currently received datagram burst. The level of achievable completion in ADT depends on the initial buffering time DBuf: the more the initial time, the better; • after a number of burst, the received FEC columns can compensate for the missing ADT columns and can even provide some additional protection for "real" losses. In a typical case (B=6, S=4), during 4 bursts, no error can be corrected but starting with the 6th burst, the correction capacity starts to be positive and a complete burst loss is can be sustained at the 7th position; • the number of artificially missing columns due to non completion of the data part remains but decreases quickly when the initial buffering increases. So it can be interesting to increase slightly the initial buffering time to ensure that a good level of protection is ensured. In any case, a complementary buffering time is needed to reach full protection (40 columns in each ADT in average). So during B+S bursts, FEC protection increases regularly but does not reach maximal capacity unless some additional buffering of B-1 is activated. • the initial buffering before starting to deliver IP packets, as counted in number of bursts, Bufinit, can be increased (Bufinit>1), this has a good impact on protection of the stream during zapping time; • complementary buffering can be done by slowing the playout rate by a slowdown_rate factor: the slowdown_rate is the percentage of speed reduction at the output of the IFEC decoder (TR 102 377 [i.21] suggests that 20 % can be supported); • total duration to recover the data parity (missing datagram burst to fill completely the ADTs) is then (B-Bufinit)/slow_down_rate. All these delays have been represented in the following figure as a function of (B+S) and taking for initial buffering delay Dbuf a value of 4 (systematic delay of 4 burst). For instance for (B+S)=10, we recover the initial data after 6 bursts, then the FEC and the final data are completely recovered after 10 bursts. So we can display first image 4 bursts after zapping and be completely protected after 10 burst. Another example for (B+S)=20: after 12 bursts, we have, thanks to the initial delay of 4 bursts, already a good protection over the data. We must receive 20 bursts to benefit from full FEC parity and the final protection is achieved after 45 bursts. Please note the these delays are not additive: they are time thresholds. ETSI 98 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 6.19: main zapping delays 6.2.5.2.4 Zapping performance with measured quality during the zapping period 6.2.5.2.4.1 Zapping performance in class 1 Introduction The zapping performance for class 1 has been carefully studied in [i.38] to which the user shall refer for more details. Over DVB-SH, transmit data and parities are delayed. Delay results from the aggregation of multiple effects including, among others, network latency and jitter, receiver buffering, and random access point picture acquisition. This leads to possibly important channel change delay !!!             •        Figure 6.20: zapping delay with introduction of MPE-IFEC Zapping performance metrics: To evaluate numerically this delay, we define performance metrics: ETSI 99 ETSI TS 102 584 V1.1.2 (2010-XY) • Channel change delay: This is the duration between the instant of the channel change request and the instant when MPE-IFEC delivers the first Intra-coded (I) frame to the player (TCH). Tch measures only the delay introduced by the decoder - MPEG2-TS packet acquisition, buffering time required for RTP and video decoding not taken into account. Tch = AD + (m − 1).Ts + ε where,  AD = ADST Delay = delay between data arrival and data decoding,  m = index of the burst containing the first decodable I frame (m  1)  Ts = service period, ε = the tune time depending on the location of I frame in the time-slice • Cumulated Freeze Time: an indicator of the subjective video quality. Cumulated time during which the decoder was unable to deliver a displayable frame over 20s observation windows (CFT). N −1 J −1 CFT =  F j ,n .Δt where, ( ) n =0 j =0  N = number of Group of Pictures (GOPs) over 20 seconds, J = number of frames per GOP  F j ,n = 1 if frame j is non-displayable, otherwise 0  Δt = duration of the frame                                                Figure 6.21: zapping performance metrics concept Reducing zapping time techniques: In order to limit the delay introduced intrinsically by the MPE-IFEC technique, the fast zapping technique depicted in section 6.2.5.2.3 is introduced whereby received data bursts are delivered to the upper layers without waiting for reception of all data and parity (this is called the “early decoding mode”) while the data transmitted to the upper layer is slightly delayed in order to fill the MPE-IFEC buffer until the steady stae is reached (this is the late decoding mode) when all data and parities related to a datagram burst are normally retrieved by the MPE-IFEC decoder before decoding a current data burst. Early decoding mode implementation is based on a new parameter, referred as the ADST delay and denoted by AD. It fixes the buffering time of a received data burst before being delivered to the upper layers. According to AD MPE-IFEC operates in an early or late decoding mode. ETSI 100 ETSI TS 102 584 V1.1.2 (2010-XY)               ε ε    Decoding mode LATE EARLY Transmission mode D=0 D = B+S D=0 D = B+S ADST delay : AD = (B+S) . Ts + tmax = B . Ts + tmax < (B+S) . Ts + tmax < B . Ts + tmax Recovery quality Maximal correction capacity Partial correction capacity at Steady State    From the first displayed T = (B + S) . Ts T = B . Ts T = AD . Ts T = AD . Ts picture, Steady State achieved at T       Figure 6.22: Trade-off between the reduction of channel change delay and the subjective video quality with Early decoding mode. Zapping time simulation set up The support simulations is made of: - a video streaming server delivering Constant Bit Rate, GOP duration = 1 second, H.264/AVC format, for 8 TV programs multiplexed in Ts = 1 second; - a set of radio features of OFDM transmission with 2K FFT mode, Guard interval of ¼, QPSK ½ , 5 MHz channel, LMS-ITS or LMS-SU at 50 km/h at SNR = 11.2 dB and TU at 3 km/h and SNR = 7dB. - MPE-IFEC sizing for maximal performance at steady state in Late mode: the Goal is to guarantee 90% fulfillment of ESR5 leading to the interleaving depth B+S, where optimal values for B and S are derived from B+S and code rate CR = B / (B+S) as follows: LMS-ITS : B = 9, S = 11 (B+S = 20) - for CR = 0.45 (LMS-ITS = 91.6% - LMS-SU = 100% - TU = 100%) LMS-SU : B = 4 & S = 3 (B+S = 7) - for CR = 0.5 (LMS-ITS = 92.8% - TU = 100%) Note: Simulations shows same results for both modes D = 0 and D = B+S Simulated zapping timeresults: Simulations shown the contribution of the three parts in the change delay: AD and a variable part (m-1).TS + ε:  In LMS-SU and TU environments, in 90% of channel change cases, the first error free I frame is displayed after the first received burst: (m-1).TS + ε is measured to a value lower than the GOP duration (1 second).  In LMS-ITS environment, the first received bursts are more frequently non-displayable since severe burst losses occur more often in the first seconds after the channel change. ETSI 101 ETSI TS 102 584 V1.1.2 (2010-XY) Delay (s) in 90% LMS-ITS dimensioning: LMS-SU of cases dimensioning: (B,S) = (9,11) (B,S) = (3,4) LMS-ITS LMS-SU & TU LMS-SU & TU Tch AD + 1.131 AD + 0. 555 AD + 0.319 Table 6.12: simulated channel delay. Simulated video quality results: Video quality performance split into different 3 behaviour regions with regards to AD:  For D = 0  In A : change delay is ok, video quality is nok;  In B : compromise between change delay and video quality  In C : video quality is ok,, change delay is nok.  For D = B+S:  In A : compromise between change delay and video quality  In B&C : video quality ok, change delay nok but quality degraded in the first 20s because errors in the first bursts are very prejudicial compared to D = 0. In any case, D = B+S more sensible to errors than D = 0 just after the switching request (first seconds of video). For larger B+S values, IFEC enters in the steady state later (late decoding): 20 s in LMS-ITS instead of 7 s in LMS-SU for D = 0, 9 s in LMS-ITS instead of 3 s in LMS-SU for D = B+S.    3.19s 2.95s  1.05 s 0s      Early mode   Late mode D=0 Early mode Late mode D = B+S Figure 6.23: video quality as a function of ADST delay in LMS-SU (B,S,D) = (3,4,0) and (3,4,7) and 90% of cases ETSI 102 ETSI TS 102 584 V1.1.2 (2010-XY)    3.78s  3.22s 2.09 s 1.9 s      Early mode   Late mode D=0 Early mode Late mode D = B+S Figure 6.24: video quality as a function of ADST delay in LMS-SU (B,S,D) = (9,11,0) and (9,11,20) and 90% of cases Conclusion: Experimental analysis of the channel change performance with respect to the delay and the cumulated freeze time showed that for typical DVB-SH channels (TU and LMS) the value of the ADST delay can be chosen in order to minimize the channel change delay while maintaining an acceptable level of subjective video quality (minimal cumulated freeze time). However the quality level obtained is not optimal since MPE-IFEC decodes the data with a reduced correction capacity. In order to offer the lowest channel change delay with the better subjective video quality to the end-user, the [CDP IG] proposes a mechanism to enable a transitioning between an “immediate” play and a maximum error protection realized via a progressively switching between the early decoding mode and the late decoding mode of MPE-IFEC. 6.2.5.2.4.1 Zapping performance in class 2 Introduction The main focus of this chapter is the behavior of the DVB-SH Class 2 physical layer. It is investigated what are typical delays between a switch-on or zapping command and the availability of a first I-Frame at the video decoder input buffer. Different events of switch-on and zapping are described. In addition to that, a reference composition of a DVB- SH MPEG-TS including time sliced services is introduced in a fairly abstract way, leaving exact numbers to be filled by experts on DVB-SH multiplex composition. Therefore, only ranges of values and their corresponding symbols are given. Cases for switch-on and zapping S1) Acquisition of a new TS; this assumes that the parameters like bandwidth, FFT, etc. are known to the receiver, e.g. using the delivery descriptors available throughout any other DVB or DVB-SH link Z1) Decoding of a different elementary stream within the same TS, but within a different SHService Z2) Decoding of a different elementary stream within the same TS and within the same SHService Z3) Decoding of a different elementary stream within the same TS under the assumption of full multiplex decoding (including large de-interleaver memory to store the entire de-interleaver). Cases S1 and Z1 might be similar as the de-interleaver memory is empty at the moment of the event, while cases Z2 and Z3 can benefit from previously-stored interleaver units within its deinterleaver memory. Special attention is therefore laid on cases S1 and Z1. ETSI 103 ETSI TS 102 584 V1.1.2 (2010-XY) Zapping time simulation set up Figure 6.25: Channel amplification as simulated (here with C/N = 12 dB); channel off-time between seconds 20 and 40; blockage hold-over time (amber), switch-on time (green) For simulation, an AWGN-channel with on-off characteristic is introduced. All on- and off-events are of larger duration than the time interleaver such that either the interleaver is completely filled (before the off-event to evaluate the blockage hold-over time) or completely empty (before the on-event to evaluate the switch-on time without any prior knowledge). The time between the off-event and the first error after the FEC decoder is measured (see amber arrow) – this value indicates how long the physical layer is able to deliver error-free output even after running into a blockage event. The time between the on-event and the last erroneous packet after the FEC decoder is measured (see green arrow) – this value indicates how long it takes for the physical layer to deliver error-free output again after switching- on. Configuration Table ID Comments 2 Table A10.11 1 - 3 Table A10.12 1 10/12/10/8/4 4 Table A10.11 1 40/0/12/8/4 5 Table A10.16 1 10/18/11/8/4 6 Table A10.12 1 5/18/11/4/4 8 Table A10.26 7 Table 6.13: class 2 zapping time cases Zapping time simulation results Configurations with 16QAM R=1/5 ETSI 104 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 6.26: Switch-on times for 16QAM with R=1/5 Figure 6.27: Blockage hold-over times for 16QAM with R=1/5 Configurations with different spectral efficiencies ETSI 105 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 6.28: Switch-on and blockage hold-over times for spectral efficiency ~2 bps/Hz Figure 6.29: Switch-on and blockage hold-over times for spectral efficiency ~1.3 bps/Hz ETSI 106 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 6.30: Switch-on and blockage hold-over times for spectral efficiency ~0.7 bps/Hz Conclusions Different Class-2 interleaver configurations and its behavior for different switch-on and zapping instants have been analyzed to determine the performance for a given spectral efficiency. The following conclusions can be derived: 1) For a given spectral efficiency, e.g. 0.8 bit/s/Hz resulting from 16QAM, R=1/5, different interleaver profiles result in significant performance differences. a. While the uniform interleaver has the highest performance to overcome blockages, its switch-on behavior is rather bad, even in high C/N conditions. b. The uniform/late interleaver as defined in IG-V1 has best switch-on performance but a lousy blockage resistance – therefore its use is discouraged also for IG-V2 simulations c. A carefully designed uniform/late interleaver (with R=0.8 in the late part, config03) benefits highly from fast switch- on (e.g. 2.2 sec at 8 dB C/N) while allowing long blockages (e.g. 5.0 sec at 8 dB C/N) 2) Comparing at similar spectral efficiencies these 16QAM results with QPSK results, the performance of QPSK (here simulated with R=1/3 and a spectral efficiency of 0.7 bit/s/Hz) offering less throughput is comparable to the 16QAM, R=1/5 case: For the carefully designed uniform/late interleaver (16QAM, config03), the switch-on times at 8 dB C/N is 2.2 sec while overcoming 5.0 sec of blockages. For the carefully designed uniform/late interleaver (QPSK, config06), the switchon times at 8 dB C/N is lower (1.2 sec) but the blockage performance is worse (4.1 sec). 3) The statements made during the Call for technologies on the switch-on performance have been fully confirmed by these simulations: a. According to the instantaneous C/N, the switch-on time is reciprocal to the received C/N at switch-on. 4) For higher spectral efficiencies (e.g. 16QAM, R=1/3 offers 1.3 bit/s/Hz), it is shown that the uniform/late profile still offers good performance at switch-on and for blockage protection. 5) With even higher spectral efficiencies (e.g. 16QAM, R=1/2 offers 2.0 bit/s/Hz), the time interleaver shuffles the missing code bits (which are either lost in the blockage or have not been received before the terminal switch-on), such that: a. The switch-on performance is not acceptable ETSI 107 ETSI TS 102 584 V1.1.2 (2010-XY) b. The blockage hold-over performance is not acceptable c. The use of long physical layer interleaver with code rates 1/2 and higher is not recommended 6) If using Class-2 concepts with uniform/late profile, it is recommended to keep the code rate below 2/5. If the targeted spectral efficiency can not be achieved, a switch to a higher-order modulation scheme is recommended. 6.2.6 Memory requirements 6.2.6.1 Introduction This clause gives the memory requirements for the receiver. We first explain in clause 6.2.6.2 Memory sizing function how the MPE-IFEC memory can be sized in a class 1 terminal by defining a "memory sizing function". We then introduce how this memory can be used in a typical implementation using the bijection between the ADST and ADT introduced in clause 6.2.3.5 ADT mapping, distinguishing between constant bit rate and variable bit rate IP flows. We describe an implementation applying for CBR. VBR can always be supported with this implementation but the memory may be significantly higher. Memory management optimization are possible that enable keeping the memory as low as for CBR traffic but this is for further study. 6.2.6.2 Memory sizing function Hypothesis 1: we consider that ADST columns are stored in ADT via the adt_index and adt_colums functions (see clause 6.2.3.5 ADT mapping). As a consequence, there is no need for budgeting memory for ADST and memory is essentially budgeted for Encoding Matrices, ADTs and FDTs. NOTE 1: One could say that at least 1 ADST buffer could be envisaged for absorbing current burst for which burst number is unknown and which cannot be mapped on ADT/FDT. This is true but, as the reader will see in the rest of the clause, optimizations could probably be possible to store this burst in ADT/FDT columns not already used without affecting quality of the decoding. So the conclusion remains the same, the budget is considered for the only ADT and FDT. Since we budget only ADT and FDT, memory requirements are directly a function of the number of encoding matrices, which is a function of parameters B,S and D via the M function, and of ADT and FDT individual sizes: memory = (ADT_size+ FDT_size) * M(D, B,S) where M(D,B,S) = B+ max(0,S− D) + max(0, D− B) . Hypothesis 2: we assume that memory is the sum of all ADT and FDT sizes counted by their matrix weight in bytes and zeroing any pointer structure (that will be used in the document below) or ancillary information bits. This approach will give precise order of magnitude but actual implementation may require more memory capacity. ADT and FDT memory is sized by their number of rows, T, and their number of columns, C for the ADST and ADT and R for the FDT. Since different optimizations are possible, for the parameter C we need to precise between ADT and ADST (Cadst and Cadt). The list of parameters required for sizing the memory is given below: • T is usually selected by the user but minimum values are fixed by the data volume per datagram burst since the maximum number of columns C is fixed to 255. T must be selected amongst {256, 512, 768 and 1 024} but T must be greater than ceil(datagram_burst_size/255). A default working value for T is 1 024. • C is a function of the traffic profile and its instantaneous variations datagram burst by datagram burst: variable datagram burst will occupy an ADST that must be sized to accommodate its peak size: ∀k ≥ 0, T * C ≥ datagram_burst_size(k ) + 1 => this sizes the Cadst parameter. • Normally Cadst should be used for ADT sizing so that Cadt = Cadst.. This is the natural option suggested by the specification. But terminal memory management optimizations can enable differentiation between both. These implementations are for further study. So in the following C=Cadt=Cadst. • When the code rate is fixed, it is a direct function of Cadt => this sizes the R parameter such that: 1 - code_rate R = Cadt * according to clause 6.2.3.8FDT generation and code rate computation. code_rate ETSI 108 ETSI TS 102 584 V1.1.2 (2010-XY) These Cadt, T and R parameters enable to derive memory requirements: T * Cadt * M(D, B,S) Memory = . code_rate NOTE 2: Cadst is not considered in the equation; this is extremely important since it enables to de-correlate the IFEC memory sizing from the actual instantaneous variations of the traffic which is measured through the Cadst. More considerations on the case where Cadt and Cadst are different is for further study. NOTE 3: Memory will need to be sized for a minimum code rate since, the lower the code rate, the higher the memory requirement. 6.2.6.3 Implementation aspects 6.2.6.3.1 Introduction Some implementation aspects of the memory are described in the following, in particular the memory requirements. Basically two opposite scenarios can be described: • a first scenario will size the memory based on peak traffic (Cadt = Cadst); • a second scenario will size the memory on averaged traffic (Cadt may differ from Cadst).. 6.2.6.3.2 Approach based on peak traffic In the first approach based on peak traffic, the memory requirement is to use M(D,B,S) times the memory required for a maximum burst size as given by the product T*Cadst. Implementation-wise, each ADST is constituted of a list of Cadst pointers, each pointer addressing one of the C columns of one of the M ADTs via the adt_index and adt_column functions. If one ADST column is padded, the memory is still reserved inside the corresponding ADT, leading to potential memory waste. In addition, there must be R columns for the FDT, R being equal at least to 1 - code_rate Cadst * , leading also to wastes on the FDT part. This is illustrated in figure 6.31. code_rate C C ADST k-1 (pointers to ADT columns) ADST k (pointers to ADT columns) C R C R R C R C T Figure 6.31: Implementation 1 (single pointer) On this figure, we can see that the padding column is greyed but its memory is "physically" reserved. The main drawback of this implementation is particularly apparent in case of VBR where it can lead to excessive memory since, for a lot of datagram bursts, the ADST will be made of many padded columns. If we consider the typical example of clause 6.2.3.1 Introduction, we have on average 37 columns. If we reserve the full 191 columns, this means we will have a waste factor of 5! 6.2.6.3.3 Approach based on average traffic Another approach is to use the averaging effect of ADT mapping. ETSI 109 ETSI TS 102 584 V1.1.2 (2010-XY) k + B-1 Hypothesis 3: It is assumed that ∀k ≥ 0,  ADT_columns(k ) ≤ L where ADT_column(k) is the number of data k columns in ADT(k) excluding the padding columns from the counting. L  1 - code_rate  We then define C ADT = and R ADT = ceil Cadst *   . The memory can then be sized based on the averaged B  code_rate   CADT rather than the peak CADST, using a double pointer approach illustrated in the figure below: first pointer connect ADST columns to ADT columns (via standard ADST to ADT mapping functions adt_index and adt_column) whereas second pointer removed padding columns by pointing only the ADT data columns to the physical memory. ADST k-2 ADST k-1 ADST k ADST k+1 Figure 6-21: implementation 2 (double pointer) Therefore the real physical memory can be sized on the average bit rate and not on the peak bit rate. This information is given to the receiver via max_average_rate_over_B field found in the time_slice_fec_identifier. This field introduced in  i = (n - n [EP ]+ (EP −1))+ (B -1)     Bi i = n - n [EP ]   [i.33] with following definition: C B = max  , where  Tn       ∀n ≥ 0 - Bi is the size of ith Time Slicing burst in MPE section payload bits - Tn is the time from the transport packet carrying the first byte of the first MPE section in burst/frame n-n[EP] to the transport packet carrying the first byte of the first MPE section in burst/frame (n-n[EP]+(EP-1)+B) within the same elementary stream. It is then quite easy to cnvert this information into the number of columns. 6.2.6.3.4 Memory evaluation We provide some memory quantitative evaluations. For the average MPE traffic, we use the parameter max_average_rate, given by the time_slice_fec_indentifier. In order to represent the burst traffic fluctuations, we introduce a parameter called traffic_variations_ratio which represents the burst traffic variations around this average value. ∀k ≥ 0, C b (k ) ≤ C b_ifec * (1 + traffic_variation_ratio ) . This parameter can be important for VBR traffic (more than 100 %). Even in the case of CBR, there are small variations of traffic around a medium value and the ratio is not null. Hypothesis 4: we assume also that a repetition_interval (average time distance between two successive bursts) can be given. This can be compute by averaging over a sufficient number of bursts. The number of rows nof_rows is given by T found in the time_slice_fec_identifier (we take 1 024 in the following). ETSI 110 ETSI TS 102 584 V1.1.2 (2010-XY) Cadst is then sized by the average traffic plus the traffic_variations_ratio and given by following formula:  max_average_rate * repetition_interval * (1 + traffic_variations_ratio )  C adst = ceil     nof_rows * 8  c) Implementation based on peak traffic: Cadt=Cadst d) Implementation based on averaged traffic:     ceil max_average_rate * repetition_interval * sizing (B, S, D )      C adt = ceil  nof_rows * 8  sizing (B, S, D )        The final CADT must be taken as the minimum of the two values - min (Cadt ; Cadst ) -.  fec_ratio  In any cases, R = ceil  1 - fec_ratio * C adt     For both cases, assuming for B and S the Bopt and Sopt, we can derive memory as a function of (B+S), code_rate and max_average_rate: (B + S)   ( )   Bopt ; Sopt  M(D; B;S) code_rate    (C adst ; C adt ; R ) max_average_rate     repetition_interval  The resulting bit rates are given in figure 6.32 for some typical maximum average bit rates values (128, 256, 512) and with a maximum value of code rate of 50 %. Memory (fec ratio = 50% ) 35 30 25 20 128 Mbits 256 15 512 10 5 0 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 (B+S) Figure 6.32: Memory requirements in CBR For a 250 kbps CBR stream, based on simulated performance given in clause 6.2.5, we recommend values presented in table 6.14. ETSI 111 ETSI TS 102 584 V1.1.2 (2010-XY) Table 6.14: Typical memory requirements for a CBR stream Modulation B+S (bursts) Code rate (%) Memory (Mbits) LMS-ITS 16QAM 25 40 13 O-QPSK 20 55 10 TDM 15 65 8 LMS-SU all 10 66 5 NOTE 1: These values are excluding any memory for the MPE-FEC decoding. NOTE 2: These values are applicable for CBR. For VBR, one must ensure that the burst size does not exceed C*T. Memory optimizations for supporting efficiently VBR are for further study. 6.2.7 MPE-IFEC usage scenarios 6.2.7.1 Definition Equal Error Protection (EEP): case where a video stream made of several individual layers (e.g. SVC) is protected equally on all its individual layers, id est each of the individual layer bears same FEC protection as the others. NOTE: This case opposes to the UEP case. MPE-IFEC sources: these are sets of elementary streams transporting MPE-IFEC sections coming from the same encoding matrices and having coherent signalling. These elementary streams may be located on same or different TS, same or different frequencies, same or different radio technologies. Unequal Error Protection (UEP): case where a video stream made of several individual layers (e.g. SVC) is protected unequally on the different individual layers, id est each individual layer bears a potentially different FEC protection than the others, more important layers being usually better protected than less important layers. NOTE: This case opposes to the EEP case. 6.2.7.2 Introduction This clause describes typical scenarios of MPE-IFEC usage in a DVB-SH environment, leveraging on the signalling flexibility. We distinguish two basic scenarios, one using a single source of MPE-IFEC and another one using two sources, the latter being for further study. 6.2.7.2.1 Single MPE-IFEC source This is the basic scenario that has been used throughout the document. The system is delivering an IP flow sent in the format of MPE sections, protected by MPE-IFEC sections sent together with the original MPE data sections over the same frequency. This scenario applies for instance to SFN cases where the satellite and CGC are operating on the same modulation and frequency, forcing exactly the same content at the section level to be transmitted under the different areas (satellite only, hybrid satellite and CGC, CGC only). In a constant bit rate situation, the number of MPE and MPE-IFEC sections is constant from an IFEC time-slice burst to another IFEC time-slice burst. However, when a variable bit rate traffic must be supported, the number of sections present in each burst can vary dramatically. This clause describe how these variations are handled by the sender and how the signalling helps the receiver coping with them. Assuming the general case of variable bit rate traffic, the amount of MPE-IFEC section is limited by the following factors: • the used code rate: as explained in clause 6.2.3 A practical example, maintaining a target code rate will enable a maximum number of MPE-IFEC sections in iFDT is equal to:  1 - code_rate target  nof_fec_columns_code_rate(i ) = ceil nof_data_columns(i ) * ;  code_rate target    • the capacity of the IFEC time-slice burst, usually computed in number of MPEG2 TP that turns into a number of MPE-IFEC sections once the capacity to send the MPE sections has been used. We assume this capacity to be nof_fec_columns_capacity(i). ETSI 112 ETSI TS 102 584 V1.1.2 (2010-XY) It is assumed that the two variables will not differ significantly, in particular because the IP encapsulator will change nof_fec_columns_capacity(i) to follow the variations of nof_fec_columns_code_rate(i). However some adjustments are possible, especially if the nof_fec_columns_capacity(i) is slightly superior to nof_fec_columns_code_rate(i). It would then be useful to use the resource to send an additional MPE-IFEC section. For that purpose, the max_iFDT_column must be sent to a value superior or equal to the largest nof_fec_columns_capacity(i) plus additional margin multiple of S. The procedure to constitute the IFEC burst can be the following: • for each IFEC burst, list sections eligible for inclusion inside the burst according to the interleaving mechanism explained in clause 6.2.3.9 IFEC burst generation; • for that purpose, pick in the S previous iFDT the required number of sections until one of the following conditions is reached: - no more IFEC sections are available (in that situation an MPE-IFEC section index discontinuity may happen as explained in clause 6.2.3.9 IFEC burst generation); - no more capacity is available in the MPE-IFEC time slice burst; • once the list of MPE-IFEC sections has been established, insert them in the MPE-IFEC time slice burst and set their real-time information in the header: - set the burst number to current value; - set the section index to the current value; - set the iFDT index equal to the source iFDT; - set max_iFDT_column equal to the value computed from target_code_rate as explained in clause 6.2.3.8 FDT generation and code rate computation; - set IFEC_burst_size equal to total MPE-IFEC size; - compute real-time parameters, including Delta-t (the latter in exactly the same way as in MPE as explained in clause 6.2.3.11 Burst sending arrangement). Typical algorithm supporting such procedure are for further study. 6.2.7.2.2 Multiple MPE-IFEC source Scenarios allowing multiple sources of MPE-IFEC sections are also possible and presented as examples below: - scenario 1: MPE and MPE-IFEC sections are sent over the satellite link and complementary MPE and MPE-IFEC sections are sent on the terrestrial link. This complement is expected to provide enhanced performance at the border between the satellite and the terrestrial coverage. Synchronization between the satellite and the terrestrial links is required and already described in [21]. - scenarion 2: the hierarchical modulation is used to distribute complementary MPE and MPE-IFEC sections to the HP modulation. It is well known that some LMS channels like LMS-SU behave like on-off channels whereby receivers will be able to either receive error-free bursts or not receive at all the bursts with typically 2 consecutive bursts being erased. It would therefore be beneficial for receivers, during the “good state”, to receive additional IFEC sections that would make the protection even stronger. Following recommendation apply to multiple MPE-IFEC sources: - the multiple MPE-IFEC sources must be mixed at the receiver before being processed by the MPE-IFEC decoder and this mixing is such that: o duplicate MPE-IFEC sections on HP and LP must be removed; o order is not necessarily maintained; - concerning the MPE sections: o duplicate MPE sections on HP and LP must be removed; ETSI 113 ETSI TS 102 584 V1.1.2 (2010-XY) o order must be maintained. - The consequence on the transmitting site is the following: o the two sources MPE-IFEC sections must be received at the mixer within within the signalled delta_t (from first MPE section on HP) + max_burst_duration; o the last MPE-IFEC having frame_boundary set to 1 transmitted on HP must be the last MPE-IFEC section received therefore all MPE-IFEC sections transmitted on LP must be transmitted before the last MPE-IFEC section of the HP; o repetition interval on both HP and LP shall be equal and there shall be the same number of time slice bursts on both modulations;                                                 Figure 6.33: synchronization principle in hierarchical multiple MPE-IFEC sources usage 6.2.7.2.3 Layered IFEC-encoding in SVC context Introduction: In this scenario, different layers compose a final video stream such as in SVC. MPE-IFEC enables to encode differently each layer and we want to check that the combined usage of SVC and UEP can be beneficial. The practical scenario is targeting improvement of DVB-SH satellite coverage in rural outdoor environment. This is achieved by reducing the targeted minimal value of C/N (Carrier to Noise ratio), named C/N min required for having good quality of TV/video service reception for DVB-SH channels, on the main layer: - The good quality of TV/video service reception is defined as the 1-ESR5 criteria of 90%. - The broadcast operator wants to extend the broadcast coverage on a specified area, without increasing transmission power and without impacting the number of broadcasted services (modulation and channel coding rate remain constant), and impacting the less service quality. The broadcast operator wants to offer access to a basic video quality on an extended coverage, keeping the nominal video quality over the nominal coverage, all others parameters (power, number of services) being equal. - A nominal video service of bit rate Rs protected with a channeld coding of rate Rc is split into 2 complementary H.264 SVC services: a base service SB of bit rate RsB and an SNR enhancement service SE of bit rate RsE that increases the decoded video quality when combined with service SB. SB is protected using a channel coding rate RcB while SE is protected using a channel coding rate RcE. ETSI 114 ETSI TS 102 584 V1.1.2 (2010-XY) - Let α being the “base protection extension” factor, known and fixed by the operator with α> 1. Base service −1 −1 radio coverage is extended if different code rates are used: RcB = α ⋅ Rc with α > 1 or different interleaving (B+S) depths are used. Rc is the code rate of the original AVC stream. Due to the disadvantage of using longer interleaver on the base service, thus increasing the overall zapping time and memory usage, the first solution is favored, having different code rates for the base and enhanced layers while keeping the overall bit rate. Configuration setting IFEC configuration is configured such that ESR5 quality exceeds 90%: in a channel with radio parameters QPSK 1/2 - C/N=11.9dB, the selected IFEC parameters are CR = 2/3 & B+S = 9. This leads to a performance of 1- ESR5 = 92.8%. 1-ESR5 of LMS SUB QPSK 1/2 vs iFEC CR_ota    1,2  1   0.2 0,8 0.3 0.4 1-ESR5 0.5 0,6 0.6 0.7 0,4 0.8 0.9 0,2 0 4 5 6 7 8 9 10 11 12 C/N (dB)      Figure 6-22: 1-ESR5 (%) w.r.t Rc vs. SNR (dB) for B+S=9 Then extension factor α is selected: RsB and α is fixed, RcB is computed and RcE is selected using the curves below. Los s in high quality cove rage , Rs =705k bps , Rc=2/3 1,1 1 0,9  m nt R nhance e   alpha=1,5 0,8 alpha=1,2     c_e 0,7     0,6 0,5 0 50000 100000 150000 200000 250000 300000 350000 400000 450000 500000 Rs _bas e Figure 6-23: sizing UEP with IFEC . ETSI 115 ETSI TS 102 584 V1.1.2 (2010-XY) Three configurations are tested : A) Rsb = 100kbps, ext =1.5  RcB = 0.44, Rce = 0.72, RsE = 605kbps, Burst_period = repetition interval = 1s B) Rsb = 200kbps, ext = 1.5  RcB = 0.44 & Rce = 0.82, RsE = 505kbps, Burst_period = repetition interval = 1s C) Rsb = 200kbps, ext = 1.2  RcB = 0.55 & Rce = 0.72, RsE = 505kbps, Burst_period = repetition interval = 1s Performance results In terms of video quality, the following results are given:                                   Table 6-13: LMS-SUB (Rs = 705kbps)  ref AVC @ 38.1dB PSNR – Rsb=100kbps                                   Table 6-14: LMS-SUB (Rs = 705kbps)  ref AVC @ 38.1dB PSNR – Rsb=200kbps Note: We have only a light degradation of PSNR (less than 2.2 dB) that can be justified by the usage of more than 2 layers: in the example, 4 enhancement layers were encoded, which significantly reduces the SVC coding efficiency. These multiple layers are used for allowing a simple rate control in the video stream by means of dropping dynamically excessive layers, to compensate the absence in the SVC toolbox of means for providing at the encoder side CBR rate control that are needed for DVB-SH dumps usage. As it is well-known, the additional layers bring additional overhead and therefore PSNR degradation at equal bit rates. Therefore, the encoding performance could be significantly improved with adapted CBR control mechanisms and less enhancement layers as shown in [SW07-01]. We assume that this loss of 2.2dB would significantly be lowered with the addition of efficient rate control mechanism with 2 layers. In terms of C/Nmin we get following results. ETSI 116 ETSI TS 102 584 V1.1.2 (2010-XY) LMS SUB QPSK, 3 services, Rs=705kbps, Rc=2/3, alpha=1.2 & 1.5 1,2 1  0,8 1-ESR5 0,6 0,4    α  α 0,2        α   α  α α 0 4 5 6 7 8 9 10 11 12 C/N (dB) Ref AVC, Rc=0.66 Base ext 1.5, Rcb=0.44 Enh. w ith Rsb=200kbps, ext 1.5, Rce=0.82 Enh. w ith Rsb=100kbps, ext 1.5, Rce=0.72 Base ext 1.2, Rce=0.55 Enh. w ith Rsb=200kbps, ext 1.2, Rce=0.72 Figure 6-24: curve of 1-ESR5 (%) vs. C/N (dB) for B+S = 9 Conclusion: Therefore for Rsb = 200 kbps and α equal to 1.2, using Rcb = 0.55, Rce = 0.72, we observe a gain of  5 dB C/N gain in LMS sub urban environment on base service, while having a base service at 200kbps of acceptable quality (above 30dB PSNR for Foreman) and almost no loss on enhancement coverage. 6.3 Time-Slicing Time slicing is one of the key features introduced by DVB-H. It enables three important features: • to spare battery by powering off the receiver during intervals when no service is listened; • to support fast zapping; • to support variable bit rates and so statistical multiplexing in TDMA environment. The DVB-SH makes full usage of this time slicing information in order to provide same kind of feature support. However the way this is handled depends on the choice of the long interleaver for protection against long impairments (class 1 or 2). In the following we first provide insight on the way Time-slicing signalling is handled in DVB-SH and the impacts of this handling on power saving and VBR. In each impact case, we differentiate between class 1 and class 2 long interleavers. 6.3.1 Signalling For class 1, time slicing information is important since it enables the terminal to power off. Time slicing signalling is conveyed by data and FEC section headers in real-time parameters (taken from MAC destination bytes in MPE). Since the Time-slicing information is repeated in all sections, the probability to not receive this information is equal to the probability of losing all sections and, if ts_error_indicator is used, to the probability of having, for all section headers, at least one TS among the group conveying those header, erroneous. When such case happens, the receiver will loose Delta-t synchronization and will stay on until it can process a full section header and get a new Delta-t. ETSI 117 ETSI TS 102 584 V1.1.2 (2010-XY) Table 6.15: LMS-ITS ratio of lost burst (%) case 14 17 18 18 21 24 its its its its its its #burst 3 691 3 691 3 691 3 691 3 595 3 586 #errburst 1 424 1 641 1 251 833 1 069 1 096 #lost burst 978 1 193 788 300 24 27 #lost bursts / 26 32 21 8 1 1 #bursts (%) Table 6.16: LMS-SUB ratio of lost burst (%) case 24 24 27 30 31 34 37 77 80 80 sub sub sub sub sub sub sub sub sub sub #burst 3 667 3 586 3 667 3 667 3 667 3 594 3 590 8 939 3 583 8 984 #errburst 278 1 096 278 300 262 344 342 2 237 139 2 214 #lost burst 190 27 190 223 166 9 12 1 061 2 953 #lost bursts 5 1 5 6 5 0 0 12 0 11 / #bursts (%) Assuming typical case of LMS-ITS, we have (case 18) 21 % of burst that are lost for which no Delta-t can be retrieved. During 20 % of bursts, the terminal, unless some different strategy is applied, will go back to power on, leading to a possible power saving impact. This can be approximated by saying that 80 % of the time, the normal power saving is used while during 20 % of the time no power saving is used. This leads to a degradation of 20 % of power saving. For class 2 receivers, time slicing information is less important since the long interleaver makes the real-time parameters to be outdated. Other techniques such as the one presented in clause 7.2.3.3.1 enables to rely on other structures like the DVB-SH services signalled by SHIP. 6.3.2 Zapping time impact This clause addresses how zapping time is impacted by the DVB-SH. DVB-SH introduces long interleaving, at physical and link layer to counteract long fading experienced in LMS channels. Long interleaving can appear as contradictory with the zapping time since, to ensure protection of the stream, the receiver would need to wait for the full duration of the de-interleaving. Forcing such long zapping times is not acceptable from a user perspective, so different techniques are supported by DVB-SH to provide "fast zapping" while still ensuring good protection levels. For class 2 receivers, the physical interleaver can be tuned to provide fast zapping using uniform late profiles as described in clause 7.3.3.5. Choosing e.g. a 50/50 uniform/late profile, the zapping time is the time duration of one time interleaver burst which corresponds to roughly 200 ms on the physical layer, plus link layer delays and I-frame searching. The price to pay is that this burst must be received with a few dB more than what is the C/N threshold for this code rate (e.g. for rate 1/4 now with 4,4 dB instead of -0,9 dB as presented in clause 7.3.2.6.4). For class 1 receiver, selection of adequate link layer parameter (D) will also help reducing the zapping time and increase reception quality during zapping period. Such configurations enable to display the first well received burst immediately after physical interleaver latency assumed to be around 200 ms, plus the delay to receive the burst and search the I frame, so without waiting for the full redundancy to be received. As presented in this clause, the fast zapping provide immediate display in 80 % of the bursts in most stringent channels. The price to pay is that full redundancy has not yet been received and FEC recovery techniques such as the ones presented in clause 6.2.5.2 Zapping time performance must be applied to increase protection for following bursts. 6.3.3 Power saving impact One of DVB-SH key elements is the introduction of a longer time interleaver at the physical level that has a negative impact on the power saving. Two cases are presented below depending on which type of interleaver is used. ETSI 118 ETSI TS 102 584 V1.1.2 (2010-XY) 6.3.3.1 "Terrestrial" physical interleaver The linear part of the interleaver introduces a latency within the receiver, because of the convolutional nature, that adds a net time to the power on. The following formula can be used for deriving this information: time_on = acquisition_time+physical_interleaver_duration + burst_duration repetition_interval • burst_duration can be approximated by number_of_services time_on power_saving = 1 − repetition_interval • depending on repetition_interval the power saving will evolve; typical values are: - 1 s for repetition interval and 9 programs => burst_duration~111 ms; - acquisition_time~50 ms; - physical interleaver~200 ms; - So time_on~361 ms and power_saving~64 % instead of 84 % in DVB-H. • impact of loss pattern: - as referring to clause 6.3.1 Signalling we have power saving degradation of 20 % so 0,8*64 = 51 %. 6.3.3.2 "Long" physical interleaver In that case, the involved durations are much longer than "terrestrial" case, in the order of multiples of SH-frames. In that conditions, time slicing signalling cannot really be used for providing the relevant off times since the decoding and MPE header processing happens after a time de-interleaving that lasts longer than the Time-slicing signalling itself. However, when service synchronization between DVB-H and DVB-SH is activated, it is possible to recover partially the power saving gain. In service synchronization, bursts are grouped in DVB-SH services signalled by SHIP packet. DVB-SH services are fixed-sized and their repetition interval does not vary. The terminal is then able to pre-determine the off periods without any knowledge on the MPE Delta-t: Time-slicing is actually managed at physical level. This approach is explained in clause 7.2.3.3.1. • if we take the previous example and consider 3 DVB-SH services, we find: time_on = acquisition_time+physical_interleaver_duration + burst_duration time_on repetition_interval power_saving = 1 − time_on = acquisition_time+late_tap_interleaver + repetition_interval 3 • assuming same interleaver configuration as the one used for the class 2 terminal fast zapping in clause 6.3.2 Zapping time impact and same DVB-H service structure as in the short interleaver case in clause 6.3.3.1 "Terrestrial" physical interleaver, we have following values: - 1 s for repetition interval; - acquisition_time~50 ms; - physical interleaver (late taps only)~200 ms; - So time_on~583 ms and power_saving~42 % instead of 84 % in DVB-H. • impact of loss pattern: - this degradation of power saving is compensated by the "constant" service structure: there is no need to switch on and search for the next burst and, in case of lossy channels, the power saving is not impacted by the low pattern contrarily to the class 1 (power saving maintained at 42 %). ETSI 119 ETSI TS 102 584 V1.1.2 (2010-XY) 6.3.3.3 Summary For a typical configuration with 9 services of 111 ms duration each and a mapping of 3 services within 1 DVB-SH service, the achieved power savings are given in table 6.17. Please not that these values are given as examples, actual values will vary between configurations. Table 6.17: DVB-SH impact on power saving DVB-H (reference) DVB-SH class 1 DVB-SH class 2 Power saving clear 84 % 64 % 42 % Power saving lossy (ITS) N/A 51 % 42 % 6.3.4 VBR/statmux impact 6.3.4.1 Interest of statistical multiplexing with AVC and SVC [rezai] "Implementing Statistical Multiplexing in DVB-H"; International Journal of Digital Multimedia Broadcasting, Volume 2009 (2009), Article ID 261231, http://www.hindawi.com/journals/ijdmb/2009/261231.abs.html [SVCons] : http://telecom.esa.int/telecom/www/object/index.cfm?fobjectid=30341 Figure 6-34 shows a comparison of variable bit rate (VBR) coding with respect to constant bit rate (CBR) coding. CBR coding (dashed lines) results in a big fluctuation regarding video quality. VBR encoding (solid lines) allows for significant quality improvement of the most difficult scenes. Experiments indicate that when video is coded at a fixed quantizer step size (almost constant quality), the peak bit rate of a difficult picture frame may become more than 10 times the mean bit rate.             Figure 6-34: Comparison of CBR and VBR coding A joint data rate control for all services within a multiplex can improve the utilisation of the whole multiplex as indicated in the right diagram of Figure 6-35 Although traffic variations may be shaped and turned into constant bit rate traffic, this would be done at the penalty of the receiver buffering and so zapping time. In order to avoid this buffering time, it is important to support "real-time" variations of the traffic bit rate. DVB-H supports such variations by letting the burst vary in size from frame to frame at the expense of some loss in power conservation (ref [Rezaei]). DVB-SH also supports these variations in a similar way as DVB-H although exact way of supporting depends on physical layer interleaver choice (class 1 and class 2) because different classes do not "mix" DVB-H services in the same way. ETSI 120 ETSI TS 102 584 V1.1.2 (2010-XY)                Figure 6-35: Uncorrelated services in statistical multiplex; left without, right with common data rate control Experiments simulating all link layer processing steps for AVC coded video services (i.e. multi-protocol encapsulation, time slicing, multiplexing, and mapping onto MPEG2-TS packets) have shown that the statmux gain for DVB-SH is the same as reported for other already established broadcast systems like DVB-H. Moreover, results obtained in [SVCons] has shown that a better statmux gain can be achieved also for scalable video coding (SVC). Here, different layers of the scalable video should be transmitted with different error protection, e.g., using hierarchical modulation. Using SVC, statistical multiplex allows for fluctuation of the data rate within each layer, i.e., the available data rate in the high priority (HP) channel is shared among the Base Layer (BL) of all services. Exploitation of the features introduced by SVC, such as graceful degradation, only requires that the aggregated BL data rate for all services may not exceed the available capacity of the HP channel. Enhancement layers share the available data rate in the low priority (LP) channel(s), optionally also utilizing remaining data capacity of the HP channel. The typical statmux gain as a function of the number of SVC services within one multiplex compared to AVC services are shown in Figure 6-36 below. It should be kept in mind that such results are dependent on the video material used, the video quality threshold adopted and the accepted relaxation of the power conservation in receivers.                  Figure 6-36: Statmux gain vs. number of services (from SVConS project) 6.3.4.2 Terrestrial (short) physical interleaver With class 1 receives, the long interleaver is performed at the link layer. Since link layer also provides Delta-t information, the burst can freely vary in size and so VBR is fully supported: the addition of MPE-IFEC does not change or delay the reception and processing of the Delta-t signalling. VBR implies that useful traffic will vary from burst to burst, possibly in large proportions. For instance, the volume could multiply (or divide) by a factor of 2 between two successive bursts. Without any IFEC protection, a statistical multiplexing control would allocate dynamically the bandwidth to different streams so that compound of all streams stays within radio fixed capacity. This means that ETSI 121 ETSI TS 102 584 V1.1.2 (2010-XY) starting dates for burst (and Delta-t) will vary from burst to burst. The variable DATA burst sizes will make ADT more or less filled with DATA columns. Then different strategies can be applied to benefit from this gain of bandwidth: • FEC complement protection: this strategy consists in varying the FEC so that, in each transmitted burst, the volume of DATA plus FEC is always the same. By this means, the freed volume by statmux is used to increase FEC protection and link quality. This leads also to varying code rates, with values lower than the minimum specified. Of course the capacity will not be increased. In the sender operation of MPE-IFEC specifications [MPE-IFEC], clause 3.5, the limitating factor becomes the fec_burst_size computed so that burst size is fixed, the number of FDT FEC columns being as large as required; • fixed FEC code rate: this strategy consists in keeping FEC code rate fixed while increasing the capacity. Depending on the actual ADT size in data columns, a varying number of FEC columns is created so that code_rate_FDT = nof_ADT_data_column / (nof_FDT_fec_column + nof_ADT_data_column). In the sender operation of EN 302 583 [1], clause 3.5, the limitating factor becomes the number of FDT columns dynamically set by the code rate, the fec_burst_size being as large as the maximum size. Usual implementations maintain the bit rate of a group of individually services within a fixed setting. It is clear in that situation that the D parameter MUST be the same for all services, otherwise the different delay applied to services may lead to variable bit rate of the group of services and, possibly, exceed the setting as presented in figure 6.37. This does not prevent having, for instance, one group of statistically multiplexed services with a fixed setting and same D, and another group with another D.            Figure 6.37: influence of D in a group of statmuxed services To summarize, the terrestrial physical interleaver combined with a link layer interleaver provides sufficient flexibility for supporting VBR and statmux in possibly different ways, some similar to the DVB-H way. 6.3.4.3 Uniform long interleaver As expressed in clause 6.3.3.2 "Long" physical interleaver, the uniform long interleaver "breaks" the time slicing signalling. Unless power saving is not a key criteria of the system, in order to support VBR, DVB-H services must be grouped in "DVB-SH services" as presented in clause 7.2.3.3.1. Then it is possible to perform VBR among the services of a same DVB-SH service. The impact on the statistical multiplexing will depend on the size of the DVB-SH service and the number of DVB-H services grouped within, it will grow as the DVB-SH service becomes smaller in size. Actual performance of such a mapping is out of scope of this release. In case power saving is not a key criteria, there is no need to group DVB-H services in DVB-SH services and each DVB-H service can freely vary in size: the system is then completely compatible with DVB-H statistical multiplexing. 6.3.5 Conclusion Table 6.18 summarizes the impact of the DVB-SH system on the different features of the time slicing and related link layer features. ETSI 122 ETSI TS 102 584 V1.1.2 (2010-XY) Table 6.18: Impact on link layer Class 1 Class 2 MPE Signalling Ok MPE Signalling is obsolete, rely rather on SHIP Power saving ok but depends on Depends on mapping between DVB-H/SH repetition_interval services Degradation in erasure rather independent of channel burst loss statistics channel according to % due to fixed structure of completely_lost_burst Statistical multiplexing full support Depends on mapping between DVB-H/SH with time slicing services Class 1 has less impact / better support of the time slicing and related features: all key features of the link layer can be satisfied at the same time. The class 1 receiver can fully exploit the time slicing and its related features but bandwidth efficiency and power saving may be degraded in certain LMS channels. Class 2 can compensate the "obsolescence" of the time slicing via the mapping of DVB-H over DVB-SH services, but the mapping optimization goes in two different directions depending if power saving or statistical multiplexing is sought. Actual positioning will then depend on system constraints and in particular on terminal capacity to sustain more memory and more battery requirements: a class 2 terminal having no battery restriction and being always on will be as efficient with statistical multiplexing as a class 1. For other class 2 receiver, a proper balance has to be found between power saving and the gain of statistical multiplexing, dependent on the application scenario. 6.4 Mobility Mobility is supported in a similar way as DVB-H and subject of a specific implementation guideline. Specificities of DVB-SH mobility is the usage of hybrid_delivery_descriptor that is similar to a terrestrial and/or satellite delivery descriptor and usage of SDT service_availability. Such specification is for further study. 7 Physical Layer elements This clause summarizes the different building blocks used in DVB-SH. Most of the blocks have already been used within other DVB standards EN 302 304 [3] and EN 302 307 [6] and are considered to be sufficiently well described in the references. Those elements which have been introduced to DVB within the DVB-SH waveform are described with a high degree of detail. This clause is normative unless stated otherwise. 7.1 Overview to the physical layer elements An overview on the physical layer technologies of the DVB-SH system is given in tables 7.1 and 7.2. The different technology submodules are grouped as follows: (A) encapsulation, forward error protection, interleaving and frame adaptation; (B) OFDM modulation including TPS and reference signal insertion as well as Fourier Transform processing; (C) TDM modulation including Pilot field insertion and roll-off filtering. ETSI 123 ETSI TS 102 584 V1.1.2 (2010-XY) Table 7.1: Technology sub-modules and their descriptions (part 1) Category Technology sub-module Description Related features (A) Turbo code with block length Subset of 3GPP2 turbo code has High power efficiency. Block length of of 12 282 bits been selected as FEC scheme. 12 282 bits offers also high flexibility for time interleaver design. (A) Turbo code with code rates Wide range of code rates with High receiver sensitivity for low code between 1/5 and 2/3 stepping of approx. 1…1,5 dB in rates. Low code rates allow powerful terms of required energy per code time interleaver design and outage bit over N0. protection at physical layer. Higher code rates are selected when link layer protection is used. (A) Block code structure suitable Block structure allows to encode 8 Diversity combining is simplified. for MPEG-TS encapsulation MPEG-TS packets in one turbo Synchronization information for encoded word. Block code framing hand-over and combining can be is aligned to the SH framing. derived from the framing. (A) CRC for each MPEG-TS Additional error detection Allows support of error mitigation packet mechanism. techniques or link layer protection. (A) Flexible time interleaving Time interleaving of different length Flexible exchange of fading protection is applied at physical layer. between physical layer and link layer. (A) Short time interleaving Interleaver length is selected Additional link layer protection can be (approx. 300 ms) accordingly to available memory. selected to cope with fading channels, Class 1 receivers with reduced especially for the satellite. physical layer memory size are supported. ETSI 124 ETSI TS 102 584 V1.1.2 (2010-XY) Table 7.2: Technology sub-modules and their descriptions (part 2) Category Technology sub-module Description Related features (A) Long uniform time Best "channel averaging" for fading Combined with low code rates, interleaving channels and short random signal receiver provides high sensitivity. All (e.g. 10 to 15 s, or longer) blockages. Class 2 receivers with signal blockages are handled by the extended physical layer memory physical layer. Profile may have size are addressed. impacts on the access time after switch-on or recovery after blockage. (A) Long uniform/late time Interleaver profile optimized for short Combined with low code rates, interleaving zapping time and fast recovery after receiver provides high sensitivity. All (e.g. 10 to 15 s, or longer) temporary longer signal blockages. signal blockages are handled by the Reduced "channel averaging" for physical layer. Profile allows fast fading channels. Class 2 receivers access time after switch-on or recovery with extended physical layer after blockage due to the "late" burst memory size are addressed. contribution. (B) Pilot symbol aided OFDM Waveform identical to DVB-T with Allows reuse of existing OFDM changes in the TPS bit description. demodulators for DVB-T or DVB-H, with some adaptations. (B) Addition of 1 k mode Waveform for 1 k FFT length added Allows reuse of existing OFDM to support higher speeds and/or demodulators for DVB-T or DVB-H, smaller bandwidths. with some adaptations. (B) Addition of 1,7 MHz Waveform for L-Band channelization Allows reuse of existing OFDM channelization added. demodulators for DVB-T or DVB-H, with some adaptations. (C) Pilot symbol aided TDM Waveform derived from DVB-S2 Pilot symbol pattern is designed to with fixed framing and preamble support synchronization and tracking distance. also at very low C/N values. Regular pilot scheme allows the prediction of framing. (C) Modulations QPSK, 8PSK Waveform allows flexible choice of Different modulation orders allow to and 16APSK modulation independent of the efficiently use available satellite power, physical layer code rate. independent selection of parameters Modulations are derived from for satellite and CGC and support for DVB-S2. local content insertion. (B), (C) Different bandwidth for TDM Usually TDM and OFDM bandwidth Different channel bandwidths for TDM and OFDM is considered equal. In DVB-SH no and OFDM allow a higher flexibility such restriction applies, TDM and using the available bandwidth OFDM bandwidth can be selected resources. independently from the specified set. For code-combining the same SH framing applies to both modulations. 7.2 Turbo code, time interleaver and SH framing 7.2.1 Introduction This clause gives an overview on features which comprises of the encapsulation and encoding of MPEG-TS packets into turbo encoded words. Additionally, the time interleaving operating on turbo encoded words is introduced. 7.2.2 Turbo code 7.2.2.1 Introduction This clause will explain the characteristics of this FEC solution from the receiver side. Details are also given for combining received code bits from two reception chains (OFDM/OFDM non-SFN or OFDM/TDM) with identical or different puncturing pattern and/or code rates. The transmission side is described in the waveform document [1], therefore the focus of this clause will be (after some short introduction on the encoder) on the implementation of the turbo decoder. ETSI 125 ETSI TS 102 584 V1.1.2 (2010-XY) 7.2.2.2 Overview of the key elements 7.2.2.2.1 Top level description The turbo encoder is a block encoder, which works on a block length of LTC-input bits. The parameter LTC-input can be either 1 146 bits (signalling field) or 12 282 bits (payload). The turbo encoder consists of two Recursive Systematic Convolutional (RSC) encoders, each of the encoders producing an output of three bits. The first RSC produces the bits X, Y0 and Y1. The output of the second RSC encoder is called X', Y0' and Y1', respectively. The two encoders are connected by an intra-codeword interleaver (called 3GPP2 interleaver in figure 7.1). The interleaving instruction is given in EN 302 583 [1], clause 5.3.3. For each block length LTC-input, the instructions to shuffle the input bits are different; however this can be described by selection of one parameter and another small lookup table.                              Figure 7.1: Turbo encoder schematic 7.2.2.2.2 Parameters and numbers The number of output bits at the internal interface [X Y0 Y1 X' Y0' Y1'] is (independently of the selected code rate or puncturing pattern): NOutputs*(LTC-input+LTailBits) This results in the following parameters: • NOutputs = 6 (size of the internal interface [X Y0 Y1 X' Y0' Y1']) • LTC-input = 1 146 or LTC-input = 12 282 • LTailBits = 6 Using the above formula, it can be derived that for the payload block length LTC-input = 12 282, the number of bits at this internal interface is 73 728 bits. From these bits, only a fraction is used for transmission. This fraction is derived by the puncturing pattern for the data and the tail bit periods. An example is given in the next clause. 7.2.2.2.3 Examples for puncturing the [X Y0 Y1 X' Y0' Y1'] vector The method of puncturing the data and the tail periods of the turbo encoder is described in the waveform document [1] for code rates 1/N with 11 demodulated signals, the turbo decoder logic must be run only once after the combining of demodulated streams has taken place. In DVB-SH scenarios, either N = 1 or N = 2 demodulated signals are typically available at the receiver input: • TDM for satellite transmission, OFDM for terrestrial transmission; • OFDM for satellite and terrestrial transmission (in MFN mode). As far as antenna diversity is considered, the combining of more than one antenna may take place before or after the demodulator stage, resulting in one or two demodulators from the turbo decoders' perspective. 7.2.2.3.2 Implementation The turbo decoder itself can be considered to be made-up by various internal blocks. Their functionalities are: • Bit-wise deinterleaving: the bit-wise deinterleaving reverts the bit-wise interleaving of the transmitter. Dependent on the selected puncturing pattern ID, this stage works on block sizes between 18 432 bits and 61 440 bits which can be directly read from the deinterleaver memory. This bit-wise deinterleaving is invoked several times per turbo decoding process, dependent on the number N of demodulators active for the same code word. • Depuncturing: the depuncturing reverts the puncturing process of the transmitter. Dependent on the selected puncturing pattern ID, this stage works on the same input block size as the bit-wise deinterleaving. It is also invoked several times per turbo decoding process, if more than one demodulator is active. • Combining: the combining reads the LLR entries already present in the decoder input memory (corresponding the codewords from those demodulators that have already been combined) and adds the LLRs just read from the interleaver memory for one codeword received by one of the N demodulators. Hence a codeword from another demodulator is combined with the codewords of those demodulators that have been combined by this block before for the current turbo info word. The combined LLRs are then again stored in the decoder input memory. This process is repeated until the N codewords from the demodulators have been combined that are associated with one turbo info word. Note that this process can even be carried out without performance loss when one of the demodulators does not receive a signal: in this case, the LLRs from this demodulator are zero, hence the LLRs from the other demodulators are not changed by adding the zero-codeword from the inactive demodulator. ETSI 129 ETSI TS 102 584 V1.1.2 (2010-XY) • Decoder Input Memory (either input and working memory or both memories combined): this memory has the size of the non-punctured turbo encoder code word of 12 288*6 = 73 728 LLR values. Dependent on the architecture and the speed of the decoder logic, this memory has either to be instantiated twice (double buffering) or only once (see figure 7.4). This memory is read and written by the depuncturing stage several times per turbo decoding process, but read only once by the decoder logic when no other input is available and the decoder logic can start. • Decoder logic: the decoder logic is called once all available demodulated streams have been processed and combined on the memory. The output is forwarded to the next stage in the receiver, e.g. the CRC16 check and the decapsulation. Figure 7.4: Possible turbo decoder memory architecture 7.2.2.3.3 Maximum ratio combining and complementary code combining Combining of two signals is done by the accumulation of all LLRs available for one bit; it is a simple addition of LLR values. This is performed as depicted in figure 7.5. Please note that the figures do not represent exactly the puncturing patterns nor the code word structures introduced in DVB-SH, but should give an idea on the way combining works. Figure 7.5: Received LLRs from demodulator 1 (blue), received LLRs from demodulator 2 (green) Combined LLRs from both modulators (blue, green, orange) ETSI 130 ETSI TS 102 584 V1.1.2 (2010-XY) The different colours represent the LLRs received from different demodulators: • figure on the left hand side: the blue squares represent LLRs received from demodulator 1. This codeword is a valid codeword which can be decoded at the threshold (C/N)1 for the selected code rate (e.g. 1/2); • figure in the middle: the green squares represent LLRs received from demodulator 2. This codeword is a valid codeword which can be decoded at the threshold (C/N)2 for the selected code rate (e.g. 2/5); • figure on the right hand side: all squares in colour represent the combined LLRs available from both demodulators. This codeword is a valid code word which can be decoded at the threshold (C/N)3 for the combined code rate. The colours represent the following: - orange: these LLRs have been received by both demodulators and have been combined by pure addition of the LLRs. Addition of these LLRs is equivalent to maximum ratio combining of the demodulated signals; - green/blue: these LLRs have either been received through demodulator 1 or 2. Each green or blue square complements the other received code word. These LLRs are complementary code combined. Dependent on the choice of puncturing pattern IDs on the transmit side, the percentage of LLRs being maximum ratio combined or complementary code combined can vary. In general, maximum ratio combining of the complete codeword is a special case of combining codewords for the following configuration: • identical code rates on the different transmission paths; • identical puncturing pattern IDs on the different transmission paths. 7.2.2.4 Selection of turbo code rate together with link layer protection The DVB-SH standard offers a wide range of combinations for code rates both on the physical layer and the link layer. The overall spectral efficiency depends on both protection mechanisms and the selection of modulation order. A short computation example for the code rate selection is given here: • physical layer code rate: RPHY = 1/2; • link layer code rate: RIFEC = 2/3; • overall code rate: ROVA = 1/2 * 2/3 = 1/3. The proper choice of code rates and modulation orders is the key for the solution of the trade-off between: • high spectral efficiency (by selection of high code rates with lower protection); • high robustness in typical DVB-SH environments (by selection of low code rates with high protection). This choice is further complicated by the fact that the redundancy can be assigned seamlessly between physical layer and link layer. Some recommendations (for QPSK modulation order) are given here: • assignment of all redundancy to the physical layer (typical RPHY = 2/5 or lower): - should be chosen if long physical layer interleavers are used; - should be chosen if higher values of Doppler spread in OFDM have to be supported (see clause 7.3.1.3); • assignment of the larger part of the overall redundancy to the physical layer (typical RPHY = 1/2): - should be chosen if the link layer FEC protection is relatively low (typical RIFEC around 2/3); - should be chosen if long physical layer interleavers can not be used, e.g. due to memory constraints; - may degrade the C/N thresholds for error-free reception as higher C/N values for static and mobile reception are necessary; - impacts the satellite link budgets and terrestrial coverage planning; ETSI 131 ETSI TS 102 584 V1.1.2 (2010-XY) • assignment of the smaller part of the overall redundancy to the physical layer (typical RPHY = 2/3): - should be chosen if the link layer FEC code rates is relatively high (typical RIFEC around 1/2); - should be chosen if long physical layer interleavers can not be used, e.g. due to memory constraint; - severely degrades the C/N thresholds for error-free reception as higher C/N for static and mobile reception are necessary; - impacts the satellite link budgets and terrestrial coverage planning. According to the results presented in clause A.12, the selection of code rates on the physical layer has to be made according to the following criteria: • available satellite link budgets (given EIRP) and terrestrial coverage planning (given repeater EIRP and density); • physical layer margins necessary for certain reception scenarios and receiver classes; • desired spectral efficiency and desired reception quality. The choice to select the split of redundancy between the physical layer and the link layer will mainly be driven by the terminal classes addressed. The selection guide between solutions is for further study. 7.2.2.5 Processing at PHY to support Erasure decoding in UL Being the DVB-SH interface between Physical and Link layer, the MPEG-TS is the only mean to transport reliable error information (synchronized with the data payload) between these two layers. The DVB-SH frame supports an additional CRC16 error detection mechanism at the level of each individual TS. By checking this CRC16, the physical layer can detect erroneous TP and set accordingly the transport_error_indicator bit. This bit can then be used by following layers, in particular the link layer to optimize decoding process. Therefore, the mechanisms defined in ISO/IEC 13818-1 [8] for MPEG-TS packets should be used in order to signal the integrity of the MPEG-TS payload. Different algorithms are possible and one is proposed hereunder: The algorithm depicted in figure 7.6 takes both the CRC16 in the EHEADER and the various CRC16 over the User Packets (UP) into account. Figure 7.6: Proposed algorithm for decapsulation to support erasure decoding in UL ETSI 132 ETSI TS 102 584 V1.1.2 (2010-XY) 7.2.2.5.1 Overview on the proposed algorithm In total, the payload of one turbo code word consists of up to 9 (nine) CRC-checks, each of length 16 bit: • The first one to be considered is the CRC16 over the first 98 bit of the EHEADER. If this check fails, the EHEADER must be considered to be corrupted. No information on DataField Length (DFL) is available for this turbo code packet, therefore it can not be derived how many MPEG-TS packets had been encapsulated. No information can be forwarded to the link layer. • Up to 8 following CRCs16 have to be evaluated next, if the CRC16 over the EHEADER is correct and the SYNC byte has the expected value 47H. The DFL indicates the number of MPEG-TS packets in the DATAFIELD section. Over each packet of 187 bytes (UP), the CRC16 is calculated. The following applies: - if the CRC16 is wrong, the transport_error_indicator has to be set to 1, and the packet has to be delivered to the link layer processing; - if the CRC16 is correct, the transport_error_indicator remains untouched (it may have been set already by any other TS handler on the transmit side). 7.2.2.5.2 MPEG-TS packet format The format of the MPEG-TS transport packet is given in table 7.3. Table 7.3: MPEG-TS transport packet format The relevant flag to be set in case of a CRC16 failure on the user packet (UP) is the transport_error_indicator. 7.2.2.5.3 Generation of MPEG-TS null-Packet In case of a non-recoverable EHEADER, it is proposed to transmit a so-called MPEG-TS null packet. Although indicated by the name, the payload of a null-packet is not the all-zero sequence but defined as follows: Byte[0] Byte[1] Byte[2] Byte[3] Byte[4] Byte[5] … Byte[187] 47H 1FH FFH 10H FFH FFH … FFH Please note that all bytes between Byte [4] and Byte[187] are set to FFH. ETSI 133 ETSI TS 102 584 V1.1.2 (2010-XY) 7.2.2.5.4 Signalling of wrong MPEG-TS packet In case of an erroneous CRC16 over one of the user packets, the transport_error_indicator has to be set to 1. This is done by an OR-combination of the value 80H with the MSB (most significant bit) of Byte1. This helps to preserve any transport_error_indicator that has been set along the transmission chain. The way to set the transport_error_indicator is depicted in figure 7.7. Figure 7.7: Setting the transport_error_indicator flag 7.2.2.6 C/N performance values 7.2.2.6.1 Ideal performance in AWGN channel For AWGN channel an ideal receiver should have the theoretical performance given in table 7.4 for OFDM and table 7.5 for TDM. The values reported in table 7.4 and 7.5 exclude any LL-FEC. An ideal transmitter is assumed, as well as an ideal receiver with perfect synchronization and ideal channel knowledge. Pilot overhead in C/N is only taken into account what concerns the boosting of pilots in OFDM (0,3 dB) which reduces the Es/N0 of the data carriers with respect to the C/N. Any other overhead (signalling, guard intervals) is addressed in the spectral efficiency curves. Table 7.4: Theoretical C/N (dB) in AWGN channel for OFDM @ BER = 10-5 Code rate QPSK 16QAM 1/5 -3,6 0,7 2/9 -3,1 1,3 1/4 -2,5 1,9 2/7 -1,8 2,8 1/3 -0,9 3,7 2/5 0,1 5,0 1/2 1,4 6,8 2/3 3,5 9,7 Table 7.5: Theoretical C/N (dB) in AWGN channel for TDM @ BER = 10-5 Code rate QPSK 8PSK 16APSK 1/5 -3,9 -1,3 0,4 2/9 -3,4 -0,7 1,0 1/4 -2,8 -0,1 1,6 2/7 -2,1 0,7 2,5 1/3 -1,2 1,6 3,4 2/5 -0,2 2,7 4,7 1/2 1,1 4,4 6,5 2/3 3,2 6,9 9,4 ETSI 134 ETSI TS 102 584 V1.1.2 (2010-XY) 7.2.2.6.2 Ideal performance in Rice channel Only results for TDM (QPSK) over Rice channel are available and listed in table 7.6. The simulation assumes ideal channel estimation, perfect interleaving/deinterleaving (i.e. uncorrelated fading), the Rice factor K used is 3 dB. Table 7.6: Theoretical C/N (dB) in Rice channel (K=3 dB) for TDM @ BER = 10-5 Code rate QPSK 1/5 -3,4 1/4 -2,2 1/3 -0,4 1/2 2,2 7.2.2.6.3 Ideal performance in Rayleigh channel Only results for TDM (QPSK) over Rayleigh channel are available and listed in table 7.7. The simulation assumes ideal channel estimation and perfect interleaving/interleaving (i.e. uncorrelated fading), the Rice factor used is minus infinity dB (equivalent to a Rayleigh distribution). Table 7.7: Theoretical C/N (dB) in Rayleigh channel for TDM @ BER = 10-5 Code rate QPSK 1/5 -3,2 1/4 -2,1 1/3 -0,2 1/2 2,9 7.2.2.6.4 Ideal performance in burst erasure channel Only results for OFDM (QPSK) in burst erasure are available. The simulation assumes ideal channel estimation. Table 7.8 gives an example for the OFDM performance for the burst erasure channel. This channel removes parts of the redundancy directly at the input of the deinterleaver. One modulation (QPSK) and one code rate (1/3) has been selected to demonstrate the performance of the turbo code. The 1/3 FEC code rate can cope with a maximum erasure rate of (1- 1/3)*100= 66 %. Approaching this erasure rate the required C/N to achieve the wanted Word Error Rate (WER) is rapidly increasing to the C/N required for the uncoded case. Each turbo code word of rate 1/3 consists of 18 capacity units, which equals 288 interleaver units. The table has to be read as follows: • the first and second rows show the selected erasure rate; • the third row calculates the remaining effective code rate, i.e. the ratio of transmitted information bits over the received (non-erased) code bits: code rate / (1-erasure rate), after the applied burst erasure; • the fourth row displays the minimum C/N necessary to reach a word error rate (WER) of 10-3; • the fifth row displays the additional C/N necessary compared to AWGN performance ( = no erasures). Table 7.8: Performance of turbo codes on burst erasure channel (ideal interleaving) Erased interleaver units 0/288 24/288 48/288 72/288 96/288 120/288 144/288 Erasure rate in per cent 0% 8,3 % 16,6 % 25 % 33,3 % 41,6 % 50 % Effective Code rate 0,333 0,3636 0,4000 0,4444 0,5000 0,5714 0,6667 Minimum C/N req. for -0,9 dB 0,0 dB 0,6 dB 1,4 dB 1,8 dB 3,0 dB 4,4 dB WER=10-3 Additional C/N req. 0,0 dB 0,9 dB 1,5 dB 2,3 dB 2,7 dB 3,9 dB 5,3 dB compared to AWGN This table is a good indicator of the receiver performance in the AWGN and a channel experiencing a certain percentage of time signal blockage and ideal interleaving. For AWGN channel, the actual required C/N to achieve a ETSI 135 ETSI TS 102 584 V1.1.2 (2010-XY) certain WER results in the implementation loss achieved, whereas the block erasure channel allows to derive the implementation loss in burst fading channels. The latter also helps to evaluate the performance of the time interleaver described in clause 7.2.3. The interleaver profiles are designed such that - depending on the current reception condition and/or receiver memory - not all interleaver units are necessary to successfully decode the turbo code word. This allows to rapidly decode data on the physical layer in case of receiver switch-on or change of service, without waiting for a full interleaver delay before starting to decode. The same argument can be used to evaluate the performance degradation of memory-limited receivers that are forced to drop part of the transmitted interleaver units. The performance of the receiver in this burst erasure channel scenario is crucial for the switch-on time (zapping time) of the receiver; the expected behaviour is explained further in details in clause 7.2.3.5.4. 7.2.3 Time Interleaver 7.2.3.1 Introduction Time interleaving in general helps to counteract the effects of signal fading and shadowing/blockage on the transmission channel. Reception instants with bad signal quality can be counteracted with periods of time with good signal reception as their time distribution is over the FEC frame “randomized” by the time interleaver. To have successful FEC frame reception the % of erased (bad) symbols shall be lower than the FEC erasure correction capabilities (see clause 7.2.2.6.4). Interleaving always has to be aligned with the assigned code rate and modulation scheme used for transmission, as well as with the expected reception field strength: • if a system is working with relatively low link margins (difference between the received and the required signal power strength for error-free reception is lower than 4 dB), interleaving over a long time span (e.g. 10 s) normally will improve reception. Even though in special cases it may worsen the reception as it spreads short errors over a long time span. See the top drawings of figure 7.8 for details; • if indeed a system works with higher link margins (difference between the received and the required signal power strength for error-free reception is higher than 7 dB), interleaving over a long time span (e.g. 10 s) can strongly improve the reception as it is able to recover many bad reception conditions. See the bottom drawings of figure 7.8 for details. Interleaving can be seen as time-averaging of the received time variant C/N. It averages out good and bad reception conditions over the duration of the time interleaver., In practice the C/N does not follow an on-off characteristics, but has a more complex time dependency related to the environment, user speed and associated channel state, shadowing and fading parameters. These channel first and second order statistics are reflected by the channel models described in the Clause A.7. As shown in the previous clause, the FEC is designed to cope with a certain erasure percentage, dependent on the code rate selected for the physical layer FEC. Figure 7.8: Interleaving with different margins; top: not sufficient margin; bottom: sufficient margin ETSI 136 ETSI TS 102 584 V1.1.2 (2010-XY) Physical layer interleaving (which is addressed here) makes use of the fact that one turbo code word is cut into various pieces (up to 48, identical to the number of interleaver taps) and spread over many distinct time instants. In the receiver, these pieces are again collected, reordered into the original sequence and forwarded to the turbo decoder. Dependent on the capabilities of the receiver (refer to the receiver Class definition), a decision has to be taken whether the receiver is able or not to successfully deinterleave the selected SH signal. Details on typical interleaver profiles are given in clause 7.2.3.3.5, deinterleaving strategies are given in clause 7.2.3.4.3, and details on the compatibility of the receiver with the transmitted signal is given in clause 7.2.3.4. 7.2.3.2 Selection of receiver classes 7.2.3.2.1 Overview Two different types of receivers are addressed in TS 102 585 [2]: • Class 1 receivers can work with a number of interleaver units up to 6 528, each of 126 bits, which is the equivalent to 1/2 (half) of the SH-frame capacity. Two types of interleaver configurations in the transmit signal are compatible with this type of receiver: - short uniform interleaver; - the late burst of a long uniform-late interleaver; - the late burst of an early-late interleaver. • Class 2 receivers can work with a number of interleaver units up to 417 792, each of 126 bits, which is the equivalent to the capacity of 32 SH-frames. All types of interleaver configurations in the transmit signal (which are within the range of the memory storage capabilities of the receiver) are compatible with this type of receiver: - short uniform interleaver; - long uniform-late interleaver; - long uniform interleaver; - long early-late interleaver. 7.2.3.2.2 Class 1 / Class 2 compatible interleaver profile selection As discussed within clause 7.2.2.6.4, Class 1 receivers may still operate with interleaver profiles intended primarily for Class 2 receivers. Therefore, it was introduced as a mandatory requirement that any receiver not only derives the full list of interleaver taps, but cross-checks this with its own memory and decoding capabilities. The expected behaviour of any terminal (class 1 / class 2) is the following: • calculate all interleaver taps as signalled in either TPS (OFDM) or signalling field (TDM); • drop k taps which exceed the memory capabilities of this receiver type and keep only L[0] to L[47-k]; • check whether the remaining code rate is still decodable (remaining code rate < 1,0); • decode the remaining fragment of turbo code word if both the memory and the remaining code rate allow this. One example for the performance penalty of such a configuration has been given in table 7.8. Assigning e.g. a late burst contribution of 50 % of all taps (nof_late_taps = 24), which can all be decoded by a class 1 receiver, the performance loss would be 5,3 dB (see clause 7.2.2.6.4) for an effective code rate of 2/3 instead of a total code rate of 1/3. These values may differ for the selection of other code rates and/or assignment of taps to the late part, and depend on the channel conditions. The benefit of such a network configuration is that both class 1 and class 2 receivers can co-exist in a network primarily intended for class 2 receivers, allowing also a soft-transition between the classes. ETSI 137 ETSI TS 102 584 V1.1.2 (2010-XY) 7.2.3.3 Interleaver profile description using the waveform parameters 7.2.3.3.1 Overview For all transmitted signals in a DVB-SH system, the interleaver parameters can be chosen and signalled independently. Certain rules apply when setting the interleaver parameters for signals that are intended to be combined in the receiver prior to FEC decoding, see clause 7.2.2.3.3 for details. The difference in the signalling also results from the different SH frame lengths in OFDM and TDM mode, which are defined in multiple of CU (capacity units) and IU (interleaver units) and not in units of time. Defining identical interleaver profiles for OFDM and TDM will typically lead to slightly different interleaver length (in time). However, the maximum memory capabilities of the receiver classes have to be respected. The interleaver parameters are always defined from the receivers perspective, giving clear guidance on the selection of parameters for the de-interleaving process. Long physical layer interleaver contradicts with variable burst sizes and distances of DVB-H when power saving is addressed. However, mapping techniques of DVB-H over DVB-SH can partially compensate the contradiction. In general, time slicing (as introduced by DVB-H and preserved in DVB-SH) and long time interleaving (as introduced within DVB-SH) seem to be contradictory: • time slicing tries to concentrate the data to reduce the on-time of a receiver, thus lowering significantly the power consumption; • time interleaving tries to spread the data over a large amount of time to benefit from time-variant channels. The tools provided in DVB-SH provide the necessary flexibility in the choice of time slicing and time interleaving jointly. Therefore, the concept of SH-service was introduced, signalled by the service_synchronization_function. Each SH-service has a constant repetition period RPSH and may cover a number of time-sliced services. These SH-services are then fed into the time interleaver who's repetition period RPTI should be identical to the repetition period of the SH-service. From the transmitter's perspective, this categorization of services looks like n virtual transmission channels with reduced throughput. By selecting a reasonable number of SH-services, this SH-service category aggregates in the transmitter: • a small number (2 to 3) of DVB-H services within one SH-service. By choosing 1 out of n present SH-services, this SH-service category allows the receiver: • to decode only parts of the whole multiplex (reducing significantly the throughput requirements and deinterleaver memory necessary) • to benefit from time slicing on an SH-service by SH-service basis (less but still considerable power savings when compared to DVB-H time slicing mode) Figure 7.9 displays the relationship between time slicing services and SH-services, between SH-services and SH frames, and the output of the time interleaver. The mapping and parameter selection is the following: • n1 DVB-H services map onto one SH-service; n1 may be different for each SH-service; • One SH-service maps onto n2 EEFRAMEs; n2 may be different for each SH-service; • The repetition period RPSH for each SH-service is identical in the number n3 of SH frames; • The time interleaver works with the repetition period RPTI. The repetition period is given in SH frames with the following condition: RPTI = RPSH = slice_distance (with slice_distance being defined in the next clause). The usage of the parameter set as described above ensures that features like VBR support (at the transmitter) and time slicing (at the receiver) can be, at least partially, exploited jointly with (long) physical layer interleaving. ETSI 138 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.9: Timing relationships between Time-Slice, SH-Service and SH-frame, prior to and after interleaver (short interleaver) Figure 7.10 displays the relationship between time slicing services and SH-services, between SH-services and SH frames, and the output of the time interleaver for the case where the time interleaver is not a short interleaver. In the example an early-late interleaver is shown. This interleaver spreads a burst into two bursts with distance RPSH. ETSI 139 ETSI TS 102 584 V1.1.2 (2010-XY)                                                                                                                                                  Figure 7.10: Timing relationships between Time-Slice, SH-Service and SH-frame, prior to and after interleaver (e.g. early-late interleaver) 7.2.3.3.2 Parameters and description The waveform document [1] specifies the following table of parameters for the description of the interleaver. • Common_multiplier: this parameter is a multiplier of the number of interleaver units for all interleaver taps within one slice. The minimum value is 1, the maximum value is 63. • Nof_late_taps: this parameter partitions the interleaver into a late and a non-late part. The border between the two different parts is somewhere between 0 (no late part available) and 48 (only late part available). • Nof_slices: this parameter defines the number of interleaver slices over which the interleaver distributes the data. This parameter has to be set together with the next parameter slice_distance to determine the interleaver length in time, and has to be aligned to time slicing as described in clause 7.2.3.3.1. In case of selection of a late part only, this parameter is set to 1, and all subsequent parameters can be set to 0. Also in case of selection of no late part this parameter is set to 1, otherwise this parameter is set > 1 (refer to Figure 5.10 and Table 5.39 of [1]). • Slice_distance: this parameter defines the distance between two interleaver slices in SH-frames. The capacity (in CU) of an SH-frame depends on the modulation type, modulation order (TDM), and roll-off (TDM) selection. The TDM parameters itself of course depend from modulation order (OFDM), guard interval length ETSI 140 ETSI TS 102 584 V1.1.2 (2010-XY) (OFDM).To transform this value in an increment in IU, the parameter nof_Data_CU has to be calculated according the following clause. • Non-late increment: this parameter is an additional multiplier for all interleaver taps within all non-late slices. It has to be multiplied with the common_multiplier to determine the increment in all slices which are not the late part of the interleaver. In order to ease modulator design, it is recommended to have the following constrains applied to the selection of the common_multiplier: • using OFDM with 16QAM modulation, the common_multiplier must be a multiple of 2 (in 4 k modes) or 4 (in 8 k modes); for all other modes, no constraints apply; • using OFDM with QPSK modulation, the common_multiplier must be a multiple of 2 (in 8 k modes); for all other modes, no constraints apply. For alignment of Table 5.9 and Table 5.39 of [1] it shall be noted that there is no real reason to set parameter non_late_increment = 0 in case where the receiver configuration only applies one slice. Configurations where this is allowed are shown in Table 5.39 of [1] and in other available interleaver configurations (see clause A10). For better understanding the example in the subsequent section a dependent parameter is introduced. • Nof_nonlate slices: this parameter defines the number of interleaver slices over which the non-late part interleaver distributes the data. It is calculated as follows: o Nof_late_taps = 48: then Nof_nonlate_slices = 0 o Nof_late_taps = 0: then Nof_nonlate_slices = Nof_slices o Nof_late_taps = 1..47: then Nof_nonlate_slices = Nof_slices-1 7.2.3.3.3 Calculation examples Figure 7.11 gives an overview on how to derive exactly the tap delays for the time interleaver. ETSI 141 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.11: Process description for deriving the interleaver parameters Figure 7.12 explains the derivation of the list of the 48 interleaver delays L[0] to L[47] from the parameters as derived above. ETSI 142 ETSI TS 102 584 V1.1.2 (2010-XY)                                                                                                         Figure: 7.12: Process description for calculation of interleaver parameters 7.2.3.3.4 Calculating the interleaver transit delay The interleaver transit delay in time can be calculated from the interleaver parameters and the waveform parameters. The interleaver transit delay can be considered as length of the interleaver, or as spreading, as it is the difference between shortest (=0) and longest tap delay. The interleaver transit delay is calculated from the delay of the latest tap. It is of no matter whether it is calculated from transmitter or receiver side. The different views lead to different formulas, but provide the same results. As it is common practice, in this document the calculation is done from the receiver point of view. In a first section the interleaver transit delay is calculated in terms of IUs, later the delay is calculated in terms of time. 7.2.3.3.4.1 Calculation in terms of integer number of IU For detailed calculation different scenarios have to be considered, the detailed formulas are given here: 1) Only late taps (or spreading inside late taps): largest_IU_per_tap_value = (nof _ late _ taps − 1) * common _ multiplier 2) No late taps largest_IU_per_tap_value = (nof _ slices − 1) * slice _ dist * DataIUperSHframePerTap + (48 / nof _ slices − 1)* common _ multiplier * non _ late _ incr 3) Late taps and Non-late taps largest_IU_per_tap_value = (nof _ slices − 1) * slice _ dist * DataIUperSHframePerTap + ((48 − nof _ late _ taps ) / (nof _ slices − 1) − 1)* common _ multiplier * non _ late _ incr With ETSI 143 ETSI TS 102 584 V1.1.2 (2010-XY) IUperSHframePerTap = 816 *16 , for OFDM, where PAD IUs are also interleaved, and IUperSHframePerTap = DataIUperSHframePerTap , for TDM, where PAD IUs are not interleaved. Note: The formula for only late taps also gives the transit delay of the late part in case of a uniform-late profile. Note: According to section 7.2.3.3.2 parameter nof_nonlate_slices can be used. In 2) nof_nonlate_slices = nof_slices, in 3) nof_nonlate_slices = nof_slices – 1. 7.2.3.3.4.2 Calculation in terms of time The interleaver transit delay in time can be calculated from the interleaver length in terms of integer number of IU and the waveform parameters. For a given configuration in OFDM, this delay is always constant for all IUs, because all IUs are interleaved. Therefore the interleaver transit delay in time given below is unique. However in TDM, since some parts of the SH frame are not interleaved (SF and PADDING), the interleaver delay may vary depending on the position of the IUs in the frame. Therefore, in the following, the transit time is calculated for the IU located at the beginning of the SH frame since this is the one used for synchronization purposes (see clause 7.5). a) OFDM case The OFDM time interleaver transit delay can be expressed as a function of the SH frame duration (noted TSHF): largest_IU_per_tap_value * 48 ofdm_interleaver_transit_delay = * TSHF , 816 *16 In this formula, TSHF/(816*16) is the IU duration at the interleaver output. This transit delay can be expressed as: • A multiple of the SH frame duration, plus • A fractional SH frame duration as follows: ofdm_interleaver_transit_delay = N OFDM * TSHF + ofdm_frac * TSHF with  largest_IU_per_tap_value * 48  N OFDM = Floor   816 * 16  and largest_IU_per_tap_value * 48 - NOFDM * 816 * 16 ofdm_frac = 816 * 16 Assuming that TSHF is expressed in units o 100 ns, i.e.,  2176 * 896 * (1 + GI)  TSHF = int   Bps_OFDM * BW_OFDM *10    , where BW_OFDM is expressed in Hz (recall: for the 1.7 MHz bandwidth, the value 1.6 MHz must be used in this formula). To express the interleaver transit delay in units of 100 ns the following rule applies: ETSI 144 ETSI TS 102 584 V1.1.2 (2010-XY) ofdm_interleaver_transit_delay = NOFDM*TSHF + T_frac_ofdm (1) where T_frac_ofdm = Floor (ofdm_frac*TSHF) is a fraction of the SH frame duration expressed in 100 ns. b) TDM case The transit delay of the TDM time interleaver can be also be expressed as • A multiple of the SH frame duration, plus • A fractional SH frame duration But the computation of this transit delay is different from the OFDM case, as only DATA IUs are passed through the interleaver: signalling IUs and PADDING IUs are inserted in the SH frame after the interleaver. tdm_interleaver_transit_delay = NTDM * TSHF + tdm_frac * TSHF with  largest_IU_per_tap_value * 48  N TDM = Floor     IU_DATA_SHF_TDM  Where IU_DATA_SHF_TDM is the number of DATA IUs in the TDM SH frame, and is given by: Num_Codewords_per_SHfr IU_DATA_SHF_TDM = * 16 tdm_ code_rate/6 where Num_Codewords_per_SHfr = floor[(Frame_Width_CUs – 3)* code_rate / 6] Exact expression of the fractional part The exact expression of the fractional part is given by largest_IU_per_tap_value * 48 - NTDM * IU_DATA_SHF_TDM tdm_frac = Frame_With_CUs * 16 This formula takes into account the insertion of the signalling IUs and PADDING IUs after the time interleaver. Again, assuming that TSHF is expressed in units of 100 ns, to express the interleaver transit delay in units of 100 ns the following rule applies: tdm_interleaver_transit_delay = NTDM*TSHF + T_frac_tdm (2) where T_frac_tdm = Floor (tdm_frac*TSHF) is a fraction of the SH frame duration expressed in 100 ns. Figure: 7.13: TDM interleaver delay ETSI 145 ETSI TS 102 584 V1.1.2 (2010-XY) Approximation of the fractional part For special interleaver configurations that have been used for simulation, lab measurements and/or field trials later in this document an average calculation formula for tdm_frac has been used: largest_IU_per_tap_value * 48  largest_IU_per_tap_value * 48  tdm_frac_approx = - Floor    IU_DATA_SHF_TDM  IU_DATA_SHF_TDM  c) Numerical applications The examples here are for information. Example 1: SH-B network OFDM parameters: BW = 5 MHz, GI = 1/8, 16-QAM (Bps_OFDM = 4), time interleaver 8/48/1/0/0. TDM parameters: BW = 5 MHz, roll-off= 0,15, QPSK (Bps_TDM =2), time interleaver 16/0/1/0/40 The largest_IU_per_tap_value of the OFDM interleaver is equal to 376, and we obtain: TSHF (x100 ns) NOFDM ofdm_frac * TSHF ofdm_interleaver_transit_delay (x100 ns) (x100ns) 1096704 1 419328 1516032 The largest_IU_per_tap_value of the TDM interleaver is equal to 30080, and we obtain: TDM Frame_Width_Cus IU_DATA_SH_TDM NTDM tdm_frac * tdm_interleaver_transit_delay code rate TSHF (x100ns) (x100ns) 1/5 434 6720 214 909708 235604364 2/9 434 6480 222 833899 244302187 1/4 434 6528 221 181941 242553525 2/7 434 6720 214 909708 235604364 1/3 434 6624 217 1015841 239000609 2/5 434 6720 214 909708 235604364 1/2 434 6720 214 909708 235604364 2/3 434 6768 213 356302 233954254 Example 2: SH-B network OFDM parameters: BW = 5MHz, GI = 1/4, 16-QAM (Bps_OFDM = 4), time interleaver 8/0/1/0/40 TDM parameters: BW = 5 MHz, roll-off = 0,15, 16-APSK (Bps_TDM =4), time interleaver 8/0/1/0/40 The largest_IU_per_tap_value of the OFDM interleaver is equal to 15040, and we obtain TSHF (x100 ns) NOFDM ofdm_frac * TSHF ofdm_interleaver_transit_delay (x100 ns) (x100ns) 1218560 55 358400 67379200 ETSI 146 ETSI TS 102 584 V1.1.2 (2010-XY) The largest_IU_per_tap_value of the TDM interleaver is equal to 15040, and we obtain TDM code rate Frame_Width_Cus IU_DATA_SH_TDM NTDM tdm_frac * TSHF tdm_interleaver_transit_delay (x100ns) (x100ns) 1/5 952 14880 48 614400 59105280 2/9 952 15120 47 902400 58174720 1/4 952 14976 48 245760 58736640 2/7 952 15120 47 902400 58174720 1/3 952 14976 48 245760 58736640 2/5 952 15120 47 902400 58174720 1/2 952 15168 47 721920 57994240 2/3 952 15120 47 902400 58174720 7.2.3.3.5 Typical interleaver profiles In this clause, typical interleaver profiles are presented. These interleaver profiles have been evaluated to be compliant to the receiver classes as mentioned above and taken as reference for simulation and evaluation. In Clause A.12 the performance characteristics for these profiles are given over the reference channels selected. The support of time slicing is incorporated into the interleaver profiles, which can be parameterized according to the rules applied for the generation of time slicing on the link layer. To preserve the power-saving effect of time slicing, the parameters for a long interleaver and time slicing need to be selected jointly, otherwise reduction in the gain of power-saving through time-slicing may occur. 7.2.3.3.5.1 Short uniform interleaver profile (reference Terr48) The short uniform interleaver profile is intended for Class 1 receivers capable of saving the equivalent of 1/2 (half) SH-frame in OFDM mode. The resulting maximum interleaver length for this short uniform interleaver profile is therefore 1 (one) SH-frame in OFDM mode. Depending on modulation order, channel bandwidth and guard interval selection, this is equivalent to time spans between 100 ms (bandwidth 5 MHz, guard interval 1/32, modulation 16QAM) and 240 ms (bandwidth 5 MHz, guard interval 1/4, modulation QPSK). For other bandwidths the SH-frame length (with it the interleaver length) scales accordingly. The short uniform interleaver profile used to derive simulation results is defined in Clause A.10. A view of this interleaver path is given in figure 7.14. ETSI 147 ETSI TS 102 584 V1.1.2 (2010-XY) QPSK R=1/2 Short 46 43 40 37 34 31 28 25 22 19 16 13 10 7 4 1 0 50 100 150 200 250 Figure: 7.14: Interleaver delays for taps L[0] to L[47] (short uniform, 5/48/1/0/01) 7.2.3.3.5.2 Long uniform interleaver profile (reference Uni10) The long uniform interleaver profile is intended for Class 2 receivers capable of saving the equivalent of 32 SH-frames in OFDM mode. The resulting maximum interleaver length for this long uniform interleaver profile is therefore in the range of 64 SH-Frames in OFDM mode. Depending on modulation order, channel bandwidth and guard interval selection, this is equivalent to time spans between 6,4 s (bandwidth 5 MHz, guard interval 1/32, modulation 16QAM) and 15,6 s (bandwidth 5 MHz, guard interval 1/4, modulation QPSK). The long uniform interleaver profile used to derive simulation results is defined in Clause A.10. A view of this interleaver path is given in figure 7.15. The use of MPE-IFEC "on top" of the long uniform interleaver profile is not recommended, but not strictly forbidden as both classes of terminals (Class 1 and Class 2) have to support MPE-IFEC. 1 Interleaver parameters: common_multiplier/nof_late_taps/nof_slices/slice_distance/non_late_increment ETSI 148 ETSI TS 102 584 V1.1.2 (2010-XY) QPSK R=1/3 Uniform 46 43 40 37 34 31 28 25 22 19 16 13 10 7 4 1 0 2000 4000 6000 8000 10000 12000 14000 Figure 7.15: Interleaver delays for taps L[0] to L[47] (long uniform, 40/0/12/4/2) 7.2.3.3.5.3 Long uniform-late interleaver profile (reference ULate10) The long uniform-late interleaver profile is intended for Class 2 receivers capable of saving the equivalent of 32 SH-frames in OFDM mode. The resulting maximum interleaver length for this long uniform-late interleaver profile is typically larger than for the long uniform interleaver profile, and can be in the range of up to 128 SH-Frames in OFDM mode. One special case occurs when considering Class 1 terminals in networks transmitting signals intended primarily for Class 2 terminals. Depending on the detailed parameter selection, a Class 1 terminal may still - with penalty in mobile performance - be able to successfully decode the received signal, even when the full diversity is not exploited. The recommended Class 1 receiver behaviour is explained in clause 7.2.3.2.2. A proper parameter selection of this long uniform-late interleaver profile is mandatory for this coexistence to work. Two examples of such a long optimized- uniform-late interleaver profile are given below. Depending on modulation order, channel bandwidth and guard interval selection, this is equivalent to time spans between 12,8 s (bandwidth 5 MHz, guard interval 1/32, modulation 16QAM) and 31,2 s (bandwidth 5 MHz, guard interval 1/4, modulation QPSK). The long uniform-late interleaver profile has been used to derive simulation results is defined in Clause A.10. A view of this interleaver path is given in figure 7.16. The use of MPE-IFEC "on top" of the long uniform-late interleaver profile is not recommended, but not strictly forbidden as both classes of terminals (Class 1 and Class 2) have to support MPE-IFEC. ETSI 149 ETSI TS 102 584 V1.1.2 (2010-XY) QPSK R=1/3 Uniform-late 46 43 40 37 34 31 28 25 22 19 16 13 10 7 4 1 0 2000 4000 6000 8000 10000 12000 Figure 7.16: Interleaver delays for taps L[0] to L[47] (long uniform-late, 10/24/9/5/12) 7.2.3.3.5.4 Long early-late interleaver profile The long early-late interleaver profile is intended for Class 2 receivers capable of saving the equivalent of 32 SH-frames in OFDM mode. The resulting maximum interleaver length for this long early-late interleaver profile can be set to the same value as for the long uniform interleaver profile, and can be in the range of up to 128 SH-Frames in OFDM mode. Given the same interleaver length the long early-late interleaver profile requires the same memory than the long uniform interleaver profile. One special case occurs when considering Class 1 terminals in networks transmitting signals intended primarily for Class 2. Depending on the detailed parameter selection, a Class 1 terminal may still - with performance penalty in mobile condition - be able to successfully decode the received signal, even when the full diversity is not exploited. The recommended Class 1 receiver behaviour is explained in clause 7.2.3.2.2. A proper parameter selection of this long early-late interleaver profile is mandatory to ensure that this Class 1/Class 2 coexistence can work. Depending on modulation order, channel bandwidth and guard interval selection, this is equivalent to time spans between 12,8 s (bandwidth 5 MHz, guard interval 1/32, modulation 16QAM) and 31,2 s (bandwidth 5 MHz, guard interval 1/4, modulation QPSK). An example of the long early-late interleaver is shown in figure 7.17. The use of MPE-IFEC "on top" of the long early- late interleaver profile is not recommended, but also not forbidden as both classes of terminals (Class 1 and Class 2) have to support MPE-IFEC. ETSI 150 ETSI TS 102 584 V1.1.2 (2010-XY) QPSK R=1/3 Early-Late 46 43 40 37 34 31 28 25 22 19 16 13 10 7 4 1 0 2000 4000 6000 8000 10000 12000 14000 Figure 7.17: Interleaver delays for taps L[0] to L[47] (long early-late, 10/24/2/x/1) 7.2.3.3.5.5 Long optimized-uniform-late interleaver profile The long optimized-uniform-late interleaver profile is intended for Class 2 receivers capable of saving the equivalent of 32 SH-frames in OFDM mode. The resulting maximum interleaver length for this long optimized-uniform-late interleaver profile can be set to the same value as for the long uniform interleaver profile, and can be in the range of up to 128 SH-Frames in OFDM mode. Given the same interleaver length the long optimized-uniform-late interleaver profile requires the same memory than the long uniform interleaver profile. One special case occurs when considering Class 1 terminals in networks transmitting signals intended primarily for Class 2. Depending on the detailed parameter selection, a Class 1 terminal may still - with performance penalty in mobile conditions - be able to successfully decode the received signal, even when the full diversity is not exploited. The recommended Class 1 receiver behaviour is explained in clause 7.2.3.2.2. A proper parameter selection of this long optimized-uniform-late interleaver profile is mandatory to ensure that this Class 1/Class 2 coexistence can work. In the same manner as Class 1 terminals can exploit the late part of the interleaver profile for successful decoding also Class 2 terminals can exploit the late part at startup for “fast access”. To enable this Class1 decoding or Class 2 “fast access” the coding rate included in the late interleaver taps has to be properly selected. For the two examples of the long optimized-uniform-late interleaver profile given in figure 7.18 and figure 7.19 the code-rate of the late part is set to 4/5. To achieve this code-rate with the late part it is clear that the available capacity to be put on the uniform part depends on the overall code rate. For a low code-rate (e.g. 1/5) there is a larger amount of capacity available for the uniform part than there is available in case of a high overall code-rate. In the example (with code-rate 4/5 in the late part) for code- rate 1/5 there is free capacity to fill 36 uniform taps, whereas for code-rate ½ only 18 uniform taps can be used. Thus for the long optimized-uniform-late interleaver profile the interleaver performance for the uniform part increases with lower code-rate. Selecting a target code-rate 4/5 in the late part is a rule-of-thumb to achieve successful decoding / fast access valid for high C/N. For accurate parameterization the target code-rate for the late part depends on the C/N operating point, the C/N threshold for the waveform configuration (the delta is the C/N margin) and the overall code-rate. According to this input and the acceptable erasure rate for this configuration (refer to Table 7.8 7.2.2.6.4) can be calculated, thus assuming the erasures are the part included in the uniform part of the interleaver the target code-rate for the late part of the interleaver can be calculated for the scenario. Depending on modulation order, channel bandwidth and guard interval selection, for R=1/5 this is equivalent to time spans between 7,3 s (bandwidth 5 MHz, guard interval 1/32, modulation 16QAM) and 17,7 s (bandwidth 5 MHz, guard interval 1/4, modulation QPSK), for R=1/2 this is equivalent to time spans between 15,4 s (bandwidth 5 MHz, guard interval 1/32, modulation 16QAM) and 37,3 s (bandwidth 5 MHz, guard interval 1/4, modulation QPSK). The use of MPE-IFEC "on top" of the long optimized-uniform-late interleaver profile is not recommended, but also not forbidden as both classes of terminals (Class 1 and Class 2) have to support MPE-IFEC. ETSI 151 ETSI TS 102 584 V1.1.2 (2010-XY) QPSK R=1/5 Optimized-Uniform-Late 45 41 37 33 29 Branch-Nr. 25 21 17 13 9 5 1 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 Delay in Interleaver Units (ILUs) Figure 7.18: Interleaver delays for taps L[0] to L[47] (Long optimized-uniform-late, 5/12/10/5/4) QPSK R=1/2 Optimized-Uniform-Late 45 41 37 33 29 Branch-Nr. 25 21 17 13 9 5 1 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 Delay in Interleaver Units (ILUs) Figure 7.19: Interleaver delays for taps L[0] to L[47] (Long optimized-uniform-late, 5/30/10/5/4) ETSI 152 ETSI TS 102 584 V1.1.2 (2010-XY) 7.2.3.4 Interleaver profile alignment between satellite and terrestrial / different carriers in non-SFN configurations This clause is dedicated to give the background on the alignment of interleaver profiles between satellite and terrestrial transmission, if different carrier frequencies are used. This clause is not applicable to SH-A in SFN mode. The details on transmitter delay alignment (in terms of interleaver delay compensation) are given in clause 7.5. The synchronization aspects which are not interleaver-related are also given in clause 7.5 for all types of DVB-SH networks. There, the following items are addressed: • distribution network delays and the use of the SHIP packet for synchronization; • compensation of the satellite round-trip-delay using the SHIP mechanisms. 7.2.3.4.1 Overview Certain rules apply when the interleaver on the satellite and the terrestrial transmission are different, this can be the case if the interleaving parameters are selected differently, but can also be the case if the same interleaver parameters are selected for different waveform parameters. This is possible in the following network configurations: • SH-A in MFN mode; • SH-B. Thus, the interleaver profile selected will in most cases not be identical, due to the possibility to choose the parameters individually for both transmission paths in terms of: • modulation and modulation order (OFDM/TDM, QPSK/8PSK/16QAM/16APSK); • code rates (between 1/5 and 2/3); • interleaver length (between 100 ms and 30 s); • throughput. To be able to combine both demodulated signals, and to minimize the penalty on the receiver in terms of de-interleaver memory consumption, the transmit signals are constructed such that the difference in delay is compensated already in the transmitter. Figures 7.20 and 7.21 introduce the underlying concept. ETSI 153 ETSI TS 102 584 V1.1.2 (2010-XY)                                                        Figure 7.20: Reference system block diagram (physical layer) The blocks marked in grey (framing and interleaving) have to be considered for each input stream individually, but with knowledge of the other path if combining is intended. This is explained below. To minimize the memory impact on the receiver, the transmitters are requested to align the transmission latency between all the different time interleaver transit delays used in the DVB-SH system. Figure 7.21 shows the concept. ETSI 154 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.21: Conceptual Deployment of interleaver profiles with different length In this example, the "effective" time interleaving is shorter for the path 2 part than for the path 1 part (see both drawings on the right hand side, which displays the de-interleaver required on the receiver). One transmitter (in case of SFN the complete set of transmitters in the network) therefore has to compensate the different end-to-end delays on the shorter transmission path, in this case the terrestrial one. In general DVB-SH is flexible on which interleaver profile is used for terrestrial or satellite transmission, to reflect this in the figure the transmission is split into part 1 and 2, rather than terrestrial and satellite. Note: By definition the delay of the shortest receiver tap (L[0]) is always 0, for both TDM and OFDM. Note: The figure only provides a conceptual view on the interleaver transition delay compensation that needs to be done at the transmitter. The detailed procedure and how to achieve interleaver delay compensation (in front of the interleaver, inside the interleaver, or after the interleaver) is left to open implementors, some options are presented in clause 7.5. 7.2.3.4.2 void The content of this section is moved to clause 7.5. 7.2.3.4.3 Interleaver synchronization on the receiver side Due to interleaver delay compensation in the transmitter, the receiver can assume to receive synchronized OFDM and TDM SH-frames (in SH-B) and synchronized OFDM1 and OFDM2 SH-frames (in SH-A-MFN). The delay between the two signals is limited on air as given in clause 7.5.2. According to the different interleaver profiles for TDM and OFDM as given in the TDM signalling field, or the OFDM TPS bits respectively, the receiver aligns the SH-frames received in parallel, performs de-interleaving and combining/decoding without further consistency checks. The basic fact is that the delay of interleaver tap L[0]= 0 at the receiver side, therefore the output of the interleaver for TDM and OFDM is directly aligned. ETSI 155 ETSI TS 102 584 V1.1.2 (2010-XY) Once TDM and OFDM framing is detected the receiver directly is able to start de-interleaving and decoding task on the synchronously received SH-frames. The de-interleaving / decoding task needs to be done according to clause 5.4.3 of [1]. 7.2.3.5 Fast interleaver synchronization strategies This section is informative. 7.2.3.5.1 Motivation When long interleaver profiles are used, minimization of the access time to successfully decode a service is crucial. Independently whether it is the first access to a service ("zapping time") or the access to a service after a signal dropout ("recovery time"), it is preferable that the user gets fast feedback on whether the current location is suited for reception. The choice of a proper decoding strategy is essential for streaming services which are protected by long interleavers. Independently whether the long interleaving is done on the physical or the link layer, different decoding strategies as presented in the following are possible. 7.2.3.5.2 Definition of jitter and delay The major impact of the choice of decoding strategy is related with jitter and delay. The following definitions apply: • jitter: variation of decoding instant or variation of decoder output rate; • delay: time span between injection (transmitter side) and decoding (receiver side) of payload; • reduced diversity decoding: makes use of the fact that the full diversity (redundancy) is not needed in all reception scenarios. As soon as the receiver has successfully gathered sufficient number of packets to decode, it starts the decoding process immediately. The decision is typically made on the effective code rate: - choose the net code rate (e.g. 1/3); - derive the ratio of packets received for one code word (e.g. 40 %); - calculate the effective code rate (60 % erased bits from code rate 1/3 correspond to effective code rate of 5/6); - check whether the receiver is capable of decoding (effective code rate < 1,0 or any other limit < 1,0). 7.2.3.5.3 Early decoding Early decoding tries to minimize the end-to-end delay of a time-interleaved signal. Figure 7.22 shows the typical delays which are in many cases smaller than the full time span of packets. Figure 7.22: Principle of early decoding ETSI 156 ETSI TS 102 584 V1.1.2 (2010-XY) However, assuming that not all injected packets are necessary for decoding purposes, the jitter at the output of the physical layer may increase as soon as the receiver runs into a channel situation that is worse than before: it needs now the full redundancy transmitted, such that for a certain amount of time, no or only very few packets will be processed. The summary is the following: • early decoding does not always need the full end-to-end delay; • early decoding can provide short zapping times in good reception conditions; • video rebuffering is necessary in fading environments. 7.2.3.5.4 Late decoding Late decoding tries to minimize the jitter at the output of the physical layer decoder. It always waits for the full redundancy transmitted; the decoding instant is almost jitter-free. Figure 7.23 shows the constant delay which is the full time span of redundancy transmitted. For regular physical layer interleaving late-decoding is assumed to be baseline, as data can immediately be fed to the decoder. The decoding capability at startup depends on the margin and increases over time (with the interleaver length as maximum duration). Figure 7.23: Principle of late decoding The interleaver profiles defined in the waveform document is designed for the late decoding principle: it is possible to transmit a "late" burst which is delayed very long in the transmitter but has only a very short delay in the receiver. When access is requested to a service which is interleaved over a long time span, this late burst enables the receiver to reconstruct the data transmitted without having received all parts before. Then, as time advances, more "uniform" parts are available, making the reception more robust, until the full interleaver length is reached and full diversity is achieved. This principle of increased redundancy as time advances in presence of a “late” part in the interleaver profile can be understood as “fast access”. In case of late decoding the “delay” of the transmitted data is located the transmitter, as it is calculated from the late burst, which is not further delayed inside the receiver. The summary is the following: • late decoding allows jitter-free output of the physical layer; • short zapping times can be achieved in good reception conditions; • the waveform is designed to support late decoding; • the full end-to-end delay is always experienced even in good reception conditions; • the delay driving the end-to-end delay is located in the transmitter. ETSI 157 ETSI TS 102 584 V1.1.2 (2010-XY) 7.2.3.5.5 Zapping time in case of long physical layer interleaver The use of low code rates provides a high coding gain, a high performance in case of fading and offers short zapping time. In theory only R (code rate as ratio) of the data has to be received to allow a decoding, if the received data are error free. If not yet all data of one code word are received this is equivalent to a higher code rate. If for example for a code rate of R=1/3 40 % of the interleaved data are received this can be considered as a code rate of R=5/6. For this code rate a higher C/N is required for a correct FEC frame decoding. If the available C/N is low a lower code rate may be required for error free decoding. In this case the receiver has to wait for the next burst to progressively lower the required C/N. The interleaver profiles are configurable. The Uniform-Late profile ensures that a big part of the encoded code word is available already with the first burst. In case of good reception conditions this part is already sufficient for successful decoding. Thus the Uniform-Late interleaver profile is recommended, if a short zapping time is important but some C/N performance loss is acceptable compared to an uniform long interleaver. The late decoding principle ensures a jitter free decoding (jitter introduced by the multiplexing, variable bit rate coding and the time slicing are not considered here). The principle of the zapping behaviour of the physical layer is summarized in figure 7.24: • an input burst is subdivided in several small bursts by the time interleaver. To simplify the drawing one burst is split in 6 short bursts; • in case of uniform-late profile 50 % of the data are send as "late burst". In the example this is equivalent to 3 bursts are spread over a long time, whereas the remaining 3 bursts are send a one bigger burst; • from a FEC coding point of view the partitioning of the data in short burst is equivalent to sending complementary punctured subsets of the Turbo encoder output. Assuming a code rate of R=1/3 the subset send as late burst is equivalent to a code word encoded with a code rate of R=2/3; • if the receiver is tuned to the channel at the time t0 for the burst n the late burst is available at the time t1. This defines the zapping time for good reception conditions. The remaining data are considered as "erasures". According to table 7.8, 50 % erasure are allowed if the C/N is 5,3 dB above the threshold. This is equivalent to a nearly instantaneous zapping. In this case the latency caused by time slicing and the AVC encoder is dominating; • in bad reception conditions a lower code rate is required to decode the data. If for example 4 (3 are transmitted as late burst, 1 out of 3 from the uniform part is received) are received the equivalent code rate is R=1/2 or the not yet received data are considered as "33 % of the data are erasure". According to table 7.8 the required additional C/N is 2,7 dB; • for very bad reception condition when all of the transmitted redundancy is required to receive the data the zapping time may be up to the full interleaver length. Please note: If all redundancy is assigned to the Turbo code the physical layer uses a lower code rate resulting in a higher receiver sensitivity compared to a configuration where the parts of the redundancy are assigned to the IFEC. ETSI 158 ETSI TS 102 584 V1.1.2 (2010-XY)                                            Figure 7.24: Principle of fast zapping The numbers used for the example should be compared to the link budget. Typically a link margin (difference between available C/N for LOS reception and required C/N for AWGN channel) in the range of 6 dB to 10 dB is recommended (see Clause A.12). With the given numbers the following conclusions can be derived • for LOS reception (very good reception condition) the available link margin is typically sufficient to support instantaneous zapping; • if the receiver is in a bad reception condition when the channel is selected a longer zapping time may result; • receiver offering a lower link margin (e.g. receivers with low gain antenna) main have a higher zapping time; • for terrestrial reception the reception area can be split in three parts: o area with high C/N: short zapping time is provided; o area with medium/low C/N: a longer zapping time has to be accepted; o area with very low C/N: if all redundancy is assigned to the physical layer (no IFEC is used) the receiver offers a higher sensitivity. 7.2.4 SH framing This clause is dedicated to give advice on the alignment of the start of an SH-frame after “MAPPING ON CU and SH- FRAMING” (according to Figures 5.11 and 5.12 of [1]) in OFDM/TDM mode with respect to the framing in front of the convolutional time interleaver. Without time interleaver the alignment is straight forward as given in [1]. In this section it is described how the time interleaver affects the SH-frame generation in OFDM/TDM configuration. This section in detail describes the process of mapping the sequence of CUs at the output of the time interleaver onto OFDM symbols, TDM PL slots respectively, according to Figure.7.20 ETSI 159 ETSI TS 102 584 V1.1.2 (2010-XY) 7.2.4.1 OFDM Definition: For OFDM the “start of an SH-frame” shall be understood as the first sample of the guard interval of the OFDM symbol that carries the first IU of the SH-frame, the first IU being considered after the rate adaptation and the padding, but before the convolutional interleaver according to [1] Figure 5.11. This first OFDM symbol is built when this same first IU of the first CW of the first SH-frame exits from the interleaver after the convolutional interleaver, but before the mapping on CU and SH framing according to [1] Figure 5.11 (see Figure 7.25 for a schematic view of this process). Considering the start of SH frame in OFDM the following facts shall be noted:  There is always an integer nof_CW_per_SH_frame  The nof_IU_per_CW is a multiple of nof_taps,  therefore the first IU of a codeword is always introduced in tap 0 of the interleaver,  The last IU of an interelaver cycle introduced in the last tap 47 stays at the same location within the interleaver cycle also at the output of the interleaver.  The first IU of an SH-frame is introduced in the first tap 0 is also always at the same position in an OFDM symbol Note: Figure 7.25 shows a schematic view of how the start-of-SH frame is identified for OFDM. For visualization purposes some parameters are set to special values that are not available in DVB-SH, these modified settings are: nof_IU_per_CW 96 nof_CW_per_SH_frame 6 nof_IUs_per_OFDM_symbol 50 nof_taps 48 largest_tap_length 47 nof_interleaver_cycles_before_IU_0 47 nof_IUs_before_IU_0 2256 nof_symbols_before_IU_0 45 ETSI 160 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.25: Start of SH frame / first OFDM symbol of SH frame for OFDM (for visualization some parameter values differ from standard DVB-SH, see assumptions box) ETSI 161 ETSI TS 102 584 V1.1.2 (2010-XY) For information: With the 2 given preceding constraints, the first IU of an SH-frame inserted in tap 0, is always at the same position within an OFDM symbol. In the example this is IU(0,0). The demonstration Figure 7.25 is the following: 1) we demonstrate, at cold start, that if we introduce IU(0,0) in tap 0, it will always exit at same position within the symbol (at start of symbol 0) 2) we demonstrate that, on a continuous basis, the next IU(0,0) will always be introduced in tap 0 and will also exit at same position within the symbol To 1) Demonstration of the unique position of IU(0,0) at cold start: - let us assume the synchronization on the time is the only valid (no synchronization of the first IU inside the symbol), this means that the IU introduced at tap 47 of delay 0 is immediately introduced in the symbol at position 47 out of [0;49] possible; in the example, this is the IU(0,47); - the number of IU to be inserted between the IU(0,0) in the symbols is computed at the exit of the interleaver and equal to nof_interleaver_cycles_before_IU_0*nof_taps; since there are 47 interleaver_cycles_before_IU_0, this lead to 47 x 48 = 2256 IUs. - given the size of the symbol equal to 50 IUs, it is easy to compute the number of symbols transmitted before the one carrying IU(0,0) and the position of the IU(0,0) inside his symbol (0):  nof_symbols_before_IU_0 = floor(nof_IUs_before_IU_0/IUs_per_OFDM_symbol) = 45  nof_IUs_before_IU_0_in_symbol = nof_IUs_before_IU_0 - IUs_per_OFDM_symbol * nof_symb_before_IU_0 = 6  Therefore the IU(0,0), introduced on tap 0 with a symbol build on the exit of the tap 47 will always be positioned on the 6th IU in the symbol (using the given parameters). - now the second constraint of building a symbol on IU(0,0) start implies a constant shift of 6 IU before  Therefore, the position of the IU(0,0) shall always be the same within the symbol if inserted at the tap 0. To 2) Demonstration of maintenance of that position for following SH frames IU(0,0) after cold start: For this we must demonstrate: 2a) that following IU(0,0) are also introduced in tap 0 2b) that an SH frame always requires an integer number of symbols to carry the SH frame To 2a) Following IU(0,0) are always introduced in tap 0: In [1] §5.4.3, it is said that "for each cycle of the interleaver, 48 non interleaved IUs are read sequentially (starting on a coded word) and fed into the branches. The output of the interleaver is the 48 interleaved IU. Output is read synchronously with the input." Therefore the IU(0,0) shall always be inserted at tap 0 because it is the first IU of the first CW that starts the cycle at the input. The SH frame being equal to 816 CU, the number of interleaver cycles to feed a complete SH frame to the interleaver is equal to the integer 272 and each IU(0,0) finds naturally the same tap 0: 816 CU * (2016 bits_per_CU / 126 bits_per_IU ) / 48 taps = 272 cycles To 2b) There are always an integer number of symbols required to carry an SH frame: The number of IU per symbol actually depends on modulation parameters: Nof_IU_per_sybol = FFT_size * 756 * mod_order / 126 = FFT_size * modulation_order * 6 where FFT_size belongs to {1;2;4;8} and modulation_order belongs to {2;4} Nof_IU_per_symbol belongs to {12; 24; 48; 96; 192} ETSI 162 ETSI TS 102 584 V1.1.2 (2010-XY) The total number of IU per SH frame is equal to 13056 IUs that makes 1088, 544, 272, 136 or 68 symbols depending on the nof_IU_per_symbol. Therefore there is always an integer number of symbols required to carry the SH frame. From the receivers point of view it is obvious that the first IU of an SH frame IU(0,0) must be located at start of the first OFDM symbol of that SH frame. As the delay of tap 0 on the receiver is 0, the first IU that is fed into the receiver interleaver is also the first IU that exits from the interleaver, on both sides this IU corresponds to the start of an SH- frame. From this it is clear that IU(0,0) needs to be at the start of the first OFDM symbol of an SH frame. 7.2.4.2 TDM Definition: For TDM the “start of an SH-frame” shall be understood as the first sample of the PL slot that carries the SF right in front the first data IU of the SH-frame, the first data IU being considered after the rate adaptation, but before the convolutional interleaver according to [1] Figure 5.12. This first PL slot is built with the SF when this same first data IU of the first CW of the first SH-frame exits from the interleaver after the convolutional interleaver, but before the mapping on CU and SH framing according to [1] Figure 5.12 (see Figure 7.26 for a schematic view of this process). Taking the first sample of the first PL slot as start of SH- frame is directly related to the first sample of the pilot field at start of this first PL slot. Considering the start of SH frame in TDM (same as for OFDM) the following facts shall be noted:  There is always an integer nof_CW_per_SH_frame,  if a SH frame has more capacity than required for nof_CW_per_SH_frame and SF, then PAD CUs are inserted, PAD CUs are not interleaved,  The nof_IU_per_CW is a multiple of nof_taps,  therefore the first IU of the SH frame is always introduced in tap 0 of the interleaver,  The last IU of an interelaver cycle introduced in the last tap 47 stays at the same location within the interleaver cycle also at the output of the interleaver.  The first IU of an SH-frame introduced in the first tap 0 is also always at the same position in a PL slot. Note: Figure 7.26 shows a schematic view of how the start-of-SH frame is identified for TDM. For visualization purposes some parameters are set to special values that are not available in DVB-SH, these modified settings are: nof_IU_per_CW 96 nof_CW_per_SH_frame 6 nof_taps 48 largest_tap_length 47 nof_interleaver_cycles_before_IU_0 47 nof_IUs_before_IU_0 2256 CUs per_PL_slot (=modulation_order): QPSK 2 nof_PAD_CUs 1 nof_PL_slots_before_IU_0 (at start) 73 ETSI 163 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.26: Start of SH frame / first PL slot of SH frame for TDM (for visualization some parameter values differ from standard DVB-SH, see assumptions box) From the receivers point of view it is obvious that the first IU of an SH frame IU(0,0) must be located as first data IU after the SF in the according PL slot that SH frame. As delay of tap 0 on the receiver is 0, the first IU that is fed into the receiver interleaver is also the first IU that exits from the interleaver, on both sides this IU corresponds to the start of an SH frame. From this it is clear that IU(0,0) needs to be the first data IU of an SH frame. ETSI 164 ETSI TS 102 584 V1.1.2 (2010-XY) 7.3 OFDM and TDM elements 7.3.1 OFDM Elements This clause is informative. 7.3.1.1 Overview The OFDM modulation part is a full reuse of the functions of DVB-H with reasonable extensions necessary for the application "DVB-SH". The main new element is the introduction of a 1 k-FFT mode suitable for all channelizations. Furthermore, the TPS signalling has been extended to support the full signalling necessary in DVB-SH. All design guidelines given in the Implementation Guidelines for DVB-H [3] are still valid and are explicitly referred here. The main difference in the use of the OFDM part with respect to DVB-H is its use also at lower spectral efficiencies, down to QPSK modulation with code rate 1/5. This imposes higher requirements on the demodulator which should be designed and tested to comply with low C/N, satellite large-scale fading channels and frequency-selective reception scenarios. In the next clause, one example for the performance of OFDM in a mobile channel is given to demonstrate the interaction between demodulator and decoder. Further OFDM specific simulation results are given in Clause A.11 and A12, measurement results are given in Clause A13. 7.3.1.2 Performance in mobile TU6 Channel 7.3.1.2.1 Simulated capacity performance The following figures 7.27 and 7.28 summarize the simulated TU6 performance for various configurations assuming ideal channel estimation. The simulation conditions and the results are taken from Clause A.12. . The quality of service criterion is FER=5 %, to be in line with the measured results. The results are represented in the plot as capacity versus C/N for different waveform configurations: a) 16 QAM modulation with FEC coding rates R={1/5, 2/9, 1/4, 2/7, 1/3, 2/5, 1/2, 2/3} and two different physical layer interleaving; b) QPSK modulation with FEC coding rates R={1/5, 2/9, 1/4, 2/7, 1/3, 2/5, 1/2, 2/3} and two different physical layer interleaving profiles. The physical layer interleaving profiles are: short =200 ms class 1 interleaver, long=10 sec class 2 interleaver. Figures 7.27 and 7.28 highlight the additional interleaving gain of long physical interleaving. For 50 km/h already a short interleaving is sufficient and a longer interleaving provides only a small additional gain for 16QAM almost zero for QPSK, whereas for lower speed the gain is significant i.e. 1 dB for 16QAM and 0,5 dB for QPSK. ETSI 165 ETSI TS 102 584 V1.1.2 (2010-XY) Simulated Capacity at 3 km/h TU6 10 9 8 Capacity (Mb/s) 7 6 5 QPSK_Short_TU6_3km/h 4 QPSK_UL_TU6_3km/h 3 16QAM_Short_TU6_3km/ 2 h 16QAM_UL_TU6_3km/h 1 -3 -1 1 3 5 7 9 11 13 15 C/N (dB) Figure 7.27: Simulated Capacity vs. C/N results for OFDM, TU6 channels at 3 km/h Simulated Capacity at 50 km/h TU6 10 9 8 7 Capacity (Mb/s) 6 5 QPSK_Short_TU6_50km/h 4 QPSK_UL_TU6_50km/h 3 16QAM_Short_TU6_50km/h 2 16QAM_UL_TU6_50km/h 1 -3 -1 1 3 5 7 9 11 13 15 C/N (dB) Figure 7.28: Simulated Capacity vs. C/N results for OFDM, TU6 channel at 50 km/h 7.3.1.2.2 Measured capacity performance The following figures 7.29, 7.30, 7.31, 7.32 summarize the TU6 performance measured in laboratory using a real receiver for various configurations. The tests conditions and the results are taken from Clause A.12. The quality of ETSI 166 ETSI TS 102 584 V1.1.2 (2010-XY) service criterion is FER=5 %. The results are represented in the plot as capacity versus C/N for different waveform configurations: a) 16 QAM modulation with FEC coding rates R={1/5, 2/9, 1/4, 2/7, 1/3, 2/5, 1/2, 2/3}, two GI values and two different physical layer interleaving profiles; b) QPSK modulation with FEC coding rates R={1/5, 2/9, 1/4, 2/7, 1/3, 2/5, 1/2, 2/3}, two GI values and two different physical layer interleaving profiles. The physical layer interleaving profiles are: short =200 ms class 1 interleaver, long=10 sec class 2 interleaver. Two cases are provided: with GI of 1/8 and with GI of 1/4. Note: The simulated performance is using a uniform-late interleaver profile, for the measurements a uniform interleaver profile is used. Note: Some of the previous curves have been computed using GI = 1/8, which is not the baseline in A12. Results reported in figure 7.33 show that for required C/N at low and medium speed, GI has no impact on the required C/N values Measured Capacity at 5 km/h TU6 10 9 QPSK_Short_TU6_5km/h 8 QPSK_L_TU6_5km/h Capacity (Mb/s) 7 16QAM_Short_TU6_5km/h 6 16QAM_L_TU6_5km/h 5 4 3 2 1 -3 -1 1 3 5 7 9 11 13 15 17 19 C/N (dB) Figure 7.29: Measured Capacity vs. C/N results for OFDM, TU6 channels at 5 km/h (GI = 1/8) ETSI 167 ETSI TS 102 584 V1.1.2 (2010-XY) Measured Capacity at 50 km/h 10 9 QPSK_Short_TU6_50km/h 8 QPSK_L_TU6_50km/h Capacity (Mb/s) 7 16QAM_Short_TU6_50km/h 6 16QAM_L_TU6_50km/h 5 4 3 2 1 -3 -1 1 3 5 7 9 11 13 15 17 19 C/N (dB) Figure 7.30: Measured Capacity vs. C/N results for OFDM, TU6 channels at 50 km/h (GI = 1/8) Measured Capacity at 5 km/h TU6 10 9 QPSK_Short_TU6_5km/h 8 QPSK_L_TU6_5km/h Capacity (Mb/s) 7 16QAM_Short_TU6_5km/h 6 16QAM_L_TU6_5km/h 5 4 3 2 1 -3 -1 1 3 5 7 9 11 13 15 17 19 C/N (dB) Figure 7.31: Measured Capacity vs. C/N results for OFDM, TU6 channels at 5 km/h (GI = 1/4) ETSI 168 ETSI TS 102 584 V1.1.2 (2010-XY) Measured Capacity at 50 km/h TU6 10 9 QPSK_Short_TU6_50km/h 8 QPSK_L_TU6_50km/h 16QAM_Short_TU6_50km/h Capacity (Mb/s) 7 6 16QAM_L_TU6_50km/h 5 4 3 2 1 -3 -1 1 3 5 7 9 11 13 15 17 19 C/N (dB) Figure 7.32: Measured Capacity vs. C/N results for OFDM, TU6 channels at 50 km/h (GI = 1/4) The previous curves confirm that long interleaving does not yield much improvement at medium speed in TU6 channel condition. At lower speed the performance improvement is significant. (1 dB for QPSK and 2 dB for 16QAM). This is due to the fact that for measurement a uniform late interleaver has been used instead of a uniform long interleaver exploited in the simulations. AWGN and TU6 Characteristics for QPSK TC1/2 class1 Time Interleaver Doppler Frequency (Speed) vs C/N @ FER5 in TU6 for various GI 26dB 24dB QPSK TC 1/2 GI 1/8 22dB QPSK TC 1/2 GI 1/4 20dB QPSK TC 1/2 GI 1/16 18dB QPSK TC 1/2 GI 1/32 16dB C/N @ FER5 14dB 12dB 10dB 8dB 6dB 4dB 2dB 0dB -2dB -4dB 1 10 100 1000 Doppler Frequency (Hz) Figure 7.33 : Impact of GI on the required C/N for QPSK Some conclusions:  There is a good matching between the simulated and the measured capacity in TU6 at various speeds;  In most of the cases, 1 dB implementation loss can be achieved for both QPSK and 16QAM. ETSI 169 ETSI TS 102 584 V1.1.2 (2010-XY)  Long physical layer interleaving shows its benefit on low speed and it is higher for 16QAM than QPSK and also more marked for a real demodulator than an ideal one. This may be due to the long interleaver mitigation effect of channel estimation errors. 7.3.1.2.3 Simulated required C/N as a function of the mobile speed The used reference receiver model characterizes a DVB-SH receiver performance in an ideal way using two numbers, C/Nain and Fd3dB. The C/Nain gives the minimum required C/N for BER = 10-5. As shown in figure The C/N curve is flat up to high Doppler frequencies. Fd3dB gives the Doppler frequency, where the C/N requirement has raised by 3 dB from the C/Nain value. Simulation results are given for QPSK and different code rates in figure 7.34. This clause gives a first indication on the performance of the DVB-SH waveform using the reference demodulator of Clause A.11.2.1) instead of ideal channel estimation. The values for ideal and real channel estimation are summarized in table 7.9. The interleaver configuration used for this simulation is the short uniform interleaver. Figure 7.34: Simulated DVB-SH OFDM reference receiver C/N behaviour in mobile TU6 channel From figure 7.34, the relevant parameters C/Nain and Fd3dB can be extracted for this case. Table 7.9: Simulated DVB-SH OFDM reference receiver C/N behaviour in mobile TU6 channel Code rate Ideal estimation: Real estimation: Fd3dB C/Nain C/Nain 1/2 4,0 dB 5,5 dB 230 Hz 1/3 1,0 dB 2,4 dB 370 Hz Further results on the Doppler performance with DVB-SH are given by measurement results in the following clause. 7.3.1.2.4 Measured required C/N as a function of the mobile speed Different measurements have been made with all the different modulation/coding cases, and at different speeds up to the maximum sustainable Doppler spread. The results in figures 7.35 and 7.36 represent the variation of required C/N ETSI 170 ETSI TS 102 584 V1.1.2 (2010-XY) versus the applied maximum Doppler frequency corresponding to a certain speed, depending on the used central frequency. Different C/N values are represented in the following plots: a) AWGN required C/N : “isolated” points on the left of the graph; b) Fdmax, corresponding to the vertical part of the curves, and maximum Doppler frequency; c) Required C/N at Fdmax/2, corresponding to half of the maximum speed. One can notice that the required C/N value is quasi constant from 100 Hz to Fdmax/2; d) Required C/N at Fdmax/2 + 3 dB corresponding at the previous C/N plus 3 dB and with a corresponding speed. AWGN & TU6 channel profiles : Code rates for QPSK class1 C/N@FER5% 30dB 5MHz 2K 1/8 QPSK 1/5 25dB 5MHz 2K 1/8 QPSK 2/9 5MHz 2K 1/8 QPSK 1/4 20dB 5MHz 2K 1/8 QPSK 2/7 5MHz 2K 1/8 QPSK 1/3 15dB 5MHz 2K 1/8 QPSK 2/5 C/N 5MHz 2K 1/8 QPSK 1/2 10dB 5MHz 2K 1/8 QPSK 2/3 5dB 0dB -5dB 1Hz 10Hz Doppler Frequency 100Hz 1000Hz Figure 7.35: QPSK terrestrial performances AWGN & TU6 channel profiles : Code rates for 16QAM Class1 C/N@FER5% 30dB 5MHz 2K 1/8 16QAM 1/5 5MHz 2K 1/8 16QAM 2/9 25dB 5MHz 2K 1/8 16QAM 1/4 5MHz 2K 1/8 16QAM 2/7 20dB 5MHz 2K 1/8 16QAM 1/3 5MHz 2K 1/8 16QAM 2/5 5MHz 2K 1/8 16QAM 1/2 C/N 15dB 5MHz 2K 1/8 16QAM 2/3 10dB AWGN 5dB 0dB 1Hz 10Hz Doppler Frequency 100Hz 1000Hz Figure 7.36 : 16QAM terrestrial performances Figure 7.37 provides a summary of the maximum Doppler frequency (Fdmax) in S band versus the bit rate. ETSI 171 ETSI TS 102 584 V1.1.2 (2010-XY) Max doppler frequency versus bit rate for TU6 5MHz, 2K, GI=1/8 700Hz 1/5 650Hz Max doppler for16QAM 2/9 Doppler frequency for FER5% 1/4 Max Doppler for QPSK 600Hz 2/7 550Hz 1/31/5 2/9 2/5 1/4 500Hz 1/2 2/7 1/3 450Hz 2/5 2/3 1/2 400Hz 2/3 350Hz 300Hz 0Mbps 2Mbps 4Mbps 6Mbps 8Mbps 10Mbps 12Mbps Bit rate Figure 7.37: Maximum Doppler frequency with respect to bit rate (Class1) Finally, figure 7.38 shows the expected impact of FFT size on the maximum Doppler frequency.  Figure 7.38: Influence of FFT size on maximum Doppler Frequency Overall these results can be summarised as follows: Fdmax,2K = 2 x Fdmax,4K = 4 x Fdmax,8K, i.e. Fdmax is inversely proportional to the FFT length. 7.3.1.3 Hierarchical terrestrial modulation performance 7.3.1.3.1 Hierarchical modulation implementation In this clause we provide results based on a simulation campaign for the following simulated radio scenarios for the AWGN environment (hierarchical 16QAM – coding rate = 1/5, 2/9, 1/4, 2/7, 1/3, 1/2, 2/3). ETSI 172 ETSI TS 102 584 V1.1.2 (2010-XY) The DVB-SH modulator achieving HM implements two parallel TS data stream of different priority, coded, interleaved and framed (at bit level) independently (Figure 7.39): - High-Priority data stream (HP) with a coding rate RHP - Low-Priority data stream (LP) with a coding rate RLP. The hierarchical mapping consists in mapping each data stream (HP and LP) hierarchically onto the bits word of the 16 constellation points (or symbols) of a standard 16QAM. For each constellation symbol, the two Most Significant Bits (MSB) code the HP data stream and identify the quadrant of the complex signal plane. The two Low Significant Bits (LSB) code the LP data stream and identify the position of the symbol in the quadrant.   LP HP LP HP LP Coding HP Coding LP Decoding HP Decoding with RLP with RHP LP Framing& HP Framing& LP Deframing & HP Deframing & Interleaving Interleaving Deinterleaving Deinterleaving   Inner Deinterleaver Inner Interleaver 16-QAMhierarchical Mapper 16-QAMhierarchical Mapper   α>1 α >1     Figure 7.39: architecture scheme of SH-A receiver implementing hierarchical modulation At the receiver side, the demapper module receives both TS data streams HP and LP completely de-correlated as inputs. The receiver is basic in the sense that it does not use HP to help of LP decoding, which gives worst-case results. The two parallel independent TS data stream transport independent contents. 7.3.1.3.2 Raw performance results The performance of HM in AWGN is characterized by SNR operating points leading to BER = 10-5 after decoding of the turbo-code and to a maximal MFER 5%. This targeted performance at physical layer level is assumed to guarantee good video and audio or data decoding at application layer level. The Figure 7.40 gives SNR operating points in the hierarchical mode for α = 2 and α =4. DVB-SH AWGN alpha=2 BER=10-5 DVB-SH AWGN alpha=4 BER=10-5 15 20 13,7 17,8 1 ,6 1 15,6 10,2 15 14,3 13,2 10 9,1 12,4 8,3 1 ,7 1 1 ,1 1 7,6 10,6 7,1 6,5 10 5,2 5 2,7 4,2 5 1,1 1,9 -0,1 0 0,5 -1 -0,6 -1,7 0 -1,5 -2,3 -2,9 -2,2 -3,3 -2,8 -5 -5 9 7 5 3 9 7 5 3 9 7 5 3 9 7 5 3 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ 2/ P P P P P P P P LP LP LP LP LP LP LP LP H H H H H H H H Figure 7.40: hierarchical reception in DVB-SH versus coding rates - α = 2 and α = 4 ETSI 173 ETSI TS 102 584 V1.1.2 (2010-XY) Comparing the results obtained in the non-hierarchical mode (baseline QPSK or 16QAM) we note SNR gaps with the SNR operating points obtained in the hierarchical mode for the both values of α. They are given in the Table 7.10, Table 7.11 and Table 7.12. Table 7.10: SNR gap for HP hierarchical reception in SH (α = 2 and α = 4) vs. QPSK HP alpha=2 Penalty HP alpha=4 Penalty Coding rate Label QPSK C/N / QPSK Coding rate Label QPSK C/N / QPSK 0,2 1/5 HP 1/5 -3,6 -2,9 0,7 0,2 1/5 HP 1/5 -3,6 -3,3 0,3 0,22 2/9 HP 2/9 -3,1 -2,3 0,8 0,22 2/9 HP 2/9 -3,1 -2,8 0,3 0,25 1/4 HP 1/4 -2,5 -1,7 0,8 0,25 1/4 HP 1/4 -2,5 -2,2 0,3 0,29 2/7 HP 2/7 -1,8 -1,0 0,8 0,29 2/7 HP 2/7 -1,8 -1,5 0,3 0,33 1/3 HP 1/3 -0,9 -0,1 0,8 0,33 1/3 HP 1/3 -0,9 -0,6 0,3 0,4 2/5 HP 2/5 0,1 1,1 1,0 0,4 2/5 HP 2/5 0,1 0,5 0,4 0,5 1/2 HP 1/2 1,4 2,7 1,3 0,5 1/2 HP 1/2 1,4 1,9 0,5 0,67 2/3 HP 2/3 3,5 5,2 1,7 0,67 2/3 HP 2/3 3,5 4,2 0,7 Table 7.11: SNR gap for HP hierarchical reception in SH (α = 2 and α = 4) vs. 16QAM Coding rate 16QAM α=2 α=4 C/N required SNR gap vs. C/N required SNR gap vs. 16QAM 16QAM 0,20 1/5 0,7 -2,9 -3,6 -3,3 -4,0 0,22 2/9 1,3 -2,3 -3,6 -2,8 -4,1 0,25 1/4 1,9 -1,7 -3,6 -2,2 -4,1 0,29 2/7 2,8 -1,0 -3,8 -1,5 -4,3 0,33 1/3 3,7 -0,1 -3,8 -0,6 -4,3 0,40 2/5 5,0 1,1 -3,9 0,5 -4,5 0,50 1/2 6,8 2,7 -4,1 1,9 -4,9 0,67 2/3 9,7 5,2 -4,5 4,2 -5,5 HP decoding at same coding rate: - maintains almost same performance as QPSK (loss limited to 0,3 to 1,7 dB) - increases hugely performance with regards to non-hierarchical 16QAM (up to 5,5dB). When α increases and/or the coding rate decreases: - C/N penalty with regards to QPSK decreases (at the same coding rate between QPSK and HP), - C/N gap with regards to 16QAM increases, (at the same coding rate between QPSK and LP). Table 7.12: SNR gap for LP hierarchical reception in SH (α = 2 and α = 4) vs. QPSK & 16QAM LP alpha 2 Penalty LP alpha 4 Penalty Coding rate Label 16QAM QPSK C/N / QPSK / 16QAM Coding rate Label 16QAM QPSK C/N / QPSK / 16QAM 0,2 1/5 HP 1/5 0,7 -3,6 6,5 10,1 5,8 0,2 1/5 HP 1/5 0,7 -3,6 10,6 14,2 9,9 0,22 2/9 HP 2/9 1,3 -3,1 7,1 10,2 5,8 0,22 2/9 HP 2/9 1,3 -3,1 11,1 14,2 9,8 0,25 1/4 HP 1/4 1,9 -2,5 7,6 10,1 5,7 0,25 1/4 HP 1/4 1,9 -2,5 11,7 14,2 9,8 0,29 2/7 HP 2/7 2,8 -1,8 8,3 10,1 5,5 0,29 2/7 HP 2/7 2,8 -1,8 12,4 14,2 9,6 0,33 1/3 HP 1/3 3,7 -0,9 9,1 10,0 5,4 0,33 1/3 HP 1/3 3,7 -0,9 13,2 14,1 9,5 0,4 2/5 HP 2/5 5,0 0,1 10,2 10,1 5,2 0,4 2/5 HP 2/5 5,0 0,1 14,3 14,2 9,3 0,5 1/2 HP 1/2 6,8 1,4 11,6 10,2 4,8 0,5 1/2 HP 1/2 6,8 1,4 15,6 14,2 8,8 0,67 2/3 HP 2/3 9,7 3,5 13,7 10,2 4,0 0,67 2/3 HP 2/3 9,7 3,5 17,8 14,3 8,1 LP decoding at the same coding rate is affected by an important performance degradation compared to non-HM: - Loss versus non-HM 16QAM: 4 –5,8 dB penalty for alpha = 2 and between 8 to 9,9 dB for α = 4 - Loss versus non-HM QPSK: 10 dB penalty for α = 2 and 14 dB penalty for α = 2. This penalty increases by 4 dB with alpha=4 (alpha = 2 is better than alpha = 4). Decreasing the coding rate has little effect on QPSK but degrades (up to 1,8dB) 16QAM performance. ETSI 174 ETSI TS 102 584 V1.1.2 (2010-XY) 7.3.1.3.3 Summary: Based on AWGN simulation results, the following conclusions can be drawn: For HP stream an α value of 4 represents the best choice. However this value would cause a major (4 dB compared to α=2) performance loss increase for the LP stream. Therefore α=2 seems to be a good trade-off. Using this α value, in an AWGN channel we can see that a “basic encoding & decoding” can provide two types of usages: - Starting from a non-HM QPSK reference point: - HM can provide an increased throughput on the LP :  At marginal cost on HP: 0,7 to 1,7dB penalty with regard to QPSK at the same coding rate;  The penalty of LP with regard to non-HM QPSK is 10 dB at same code rate, which may require to use lower code rates (and then different bit rates) on this TS, reducing the penalty to as little as 3 dB; - Starting from a non-HM 16QAM reference point: - HM can provide an important SNR gain on the HP:  This gain of HP with regards to 16QAM is between 3,6 to 4,5 dB for the same code rate; Using lower code rates on HP stream (and thus different bit rates regarding to 16QAM code rate) is increasing the gap up to 12,6 dB. The reception area of both LP and HP would be impacted by a loss of 4,0 to 5,8 dB for the same code rates. Different code rates could be used to reduce this penalty down to 1,5d B for an identical bit rate by increasing HP and decreasing LP. In such condition, the HP gain is also lowered. Some trade off can be found at 1/3 and ½ code rates where there are gains and penalties of the same order as illustrated below in Figure 7.41. ( C / N ) pe na lt y ( C / N ) ga in ( H P - ( 16 Q A M - HM comparison to 16QAM at same bitrate 16 Q A M ) m a x( H P ;LP ) ) 7 7 6 6 5 5 4 4 3 3 2 2 1 1 0 0 -1 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 -1 penalty (different code rates) Code rates penalty (same code rates) Gain (different code rate) Gain (same code rate) Figure 7.41: C/N penalty and gain of HM versus 16QAM as a function of code rate 7.3.2 TDM Overview The TDM modulation of DVB-SH is derived from DVB-S2 with reasonable adaptations necessary. An important new element is the constant distance of the pilot fields, independent from code rate choice, plus an addition of a signalling field incorporating all relevant parameters for FEC and interleaving. The guidelines concerning modem algorithms design are given in the Clause A.11. TDM-specific simulation results are presented in Clause A.11. ETSI 175 ETSI TS 102 584 V1.1.2 (2010-XY) 7.3.2.1 TDM Bandwidth Selection The DVB-SH waveform specification makes no assumption on the alignment of TDM and OFDM bandwidth in the SH-B case although the examples given are all considering equal bandwidth. Therefore, although unequal bandwidth configurations for TDM and OFDM are made possible by the waveform specification, the unequal bandwidth configurations are not considered for this release of the implementation guidelines that consider equal bandwidth in the TDM and OFDM modulations. 7.4 Combining techniques Only a short overview of combining techniques is given below. 7.4.1 Overview of diversity combining techniques In general the diversity combining techniques can be grouped in: a) antenna diversity; b) satellite-to-terrestrial selective combining (incl. hand-over); c) diversity combining between different reception path (including hand-over to other frequency bands). Different methods of combining techniques are defined: • selective combining: the signal, which provides better quality, is chosen; • maximum ratio combining: the signals are combined, weighted accordingly to its specific reception quality; • complementary code combining: different code bits are combined, chosen accordingly to its occurrence in the puncturing pattern. For the satellite-to-terrestrial combining and hand-over different scenarios have to be considered. • SH-A, SFN In this case no special considerations have to be made for the combining inside the demodulator as combining occurs directly at the receiver antenna1. The satellite signal is considered as additional repeater in a SFN. But significantly higher delay spread may result and new characteristics of the channel impulse response have to be taken into account. • SH-A, MFN In this case the same content may be available over different carrier frequencies. As a consequence, it may be worthwhile that the receiver scans in the background for alternative (better) sources for the program to support a seamless hand-over. • SH-B In case of SH-B architecture the satellite and terrestrial signal is demodulated by separate demodulators. Three combining solutions are possible in this case: - the signal is selected after the FEC decoding (selective combining). This method is not recommended, as it provides the poorest performance; - the combining is done after de-interleaving and before FEC decoding. This allows that different interleaver profiles and even different code rates are used for the satellite and terrestrial signal. This method is called code combining. Code combining is very similar to maximum ratio combining. Further details on code combining and maximum ratio combining are given in clause 7.2.2.3.3; - combining before the de-interleaving (maximum ratio combining or selective combining): This method will only work if the terrestrial branch and the satellite branch use the same code rate and 1 It should be remarked that in case of SFN operations for both satellite and terrestrial repeaters the combining takes place at the antenna thus the signals are summing up with their own phase. This may create temporary constructive or destructive signal combining. ETSI 176 ETSI TS 102 584 V1.1.2 (2010-XY) interleaver parameter. For a fully SH-B compliant receiver this method is therefore not recommended. This combining point is attractive for antenna diversity, where the TDM and OFDM streams are combined separately. Combining before de-interleaving and code combining can be applied together at the same time, enabling for a very powerful diversity reception. In case of SH-B other strategies may be applicable also. For example it may be possible to completely switch off a receiver branch completely to reduce the power consumption. The evaluation of these techniques requires joint channel models covering the satellite and terrestrial propagation characteristics in one model. These models have been developed under the framework of the DVB-SH validation task force and are presented in [34] 7.4.2 void The content of this section has been integrated into clause 7.4.1. 7.4.3 void The content of this section has been moved to clause 7.7. 7.4.4 Experimental results for hybrid networks This clause is informative. 7.4.4.1 Introduction As explained above, there are many combining techniques. Amongst these combining techniques , there are those related to the Network Architecture, typically the Hybrid Network architecture, where satellite and terrestrial signals use the same frequency band and use the same modulation namely SHA-SFN mode, where signal combining is performed naturally at the user terminal antenna. The Hybrid Network description is included in Chapter 11 and this section gives the summary of following classes of results: • Simulations results; • Laboratory measurements results; • Field trial results. The Hybrid Channel tests and simulations aim to address hybrid situations when the on-air signal is a combination of terrestrial and satellite signals. This situation is particularly interesting when the satellite and the terrestrial signals are both using OFDM at the same frequency. This mode is called DVB-SHA in hybrid SFN in the standard. More generally, hybrid channel is defined by the combination at the receiver of: • One satellite link under ITS or SU conditions; • One or several terrestrial links under TU6 conditions. In the case of Hybrid SFN, the two signals are at the same frequency, using the same modulation/coding/interleaving schemes, and carrying exactly the same data. In this section, only SHA-SFN configurations have been retained from simulations and laboratory tests: • OFDM over satellite and terrestrial at same frequency bands; • A single terrestrial repeater is emulated. The objectives of the measurements are: • Assess the improvement of system coverage beyond the edge of the terrestrial cell coverage where the received C/N is below the required threshold; • Assess the possible existence of interference when satellite and terrestrial signals are not in the GI window and the impact of these interferences; • Assess the influence of the GI on the interference reduction when the interferences are existing. Taking into account the above objectives it will not be necessary to test the following “weak” hybrid situations: a) A very high terrestrial C/N (for instance 20 dB): this terrestrial signal will be very “dominant” and no satellite component impact will be observed. b) A very low terrestrial C/N as also no terrestrial component impact will be observed being the satellite signal dominant. ETSI 177 ETSI TS 102 584 V1.1.2 (2010-XY) The different tests conditions are the following:  Waveform baseline configurations:  5 MHz/2k/GI 1/8 or 1/4  Long Interleaving : Class 2 and Class 1 plus IFEC  Satellite channels configurations:  ITS 50 km/h  SU 50 km/h  Terrestrial network configurations:  2 kW EIRP single repeater  Different repeater heights:30, 45 and 60 m  TU 6 channel  Use of Costa Hata model in SU and Rural conditions  The Receiver is a car receiver, single antenna with satellite oriented receiver (circular polarized antenna) as most of the satellite reception area will be on roads and highways, and is located at the point corresponding to the terrestrial C/N received value The operating mode for the simulations is the following: • Based on the terminal location and repeater characteristics, terrestrial C/N is settled at the wanted value, which corresponds to a certain differential delay • The satellite received C/N is settled at the required value to comply with the ESR5(20) fulfilment ratio of 90 %. • Then compute the ESR5 with the CGC repeater on. • Observe if there is a performance improvement or degradation. The basic principles of the laboratory tests are the following • Take the reference curve in LMS (ITS or SU) satellite reception with Class 2 and/or IFEC. • Set the terrestrial C/N to the wanted value, which corresponds to a certain differential delay. • Perform the hybrid configuration. tests for different satellite received C/N • Compare the ESR5 fulfilment curve with the LMS reference curve mentioned before. • Observe if there is a performance improvement or degradation. As said above, the tests will be performed with class 2 and class 1 plus IFEC receiver configurations. In class 2, only QPSK 1/3 and 16QAM 1/3 configurations have been tested. For Class 1 plus IFEC, the following configuration has been tested: 16QAM ½ at physical layer plus MPE-IFEC 70 % with GI = 1/8. In a hybrid channel, there are three main parameters to consider: • The received terrestrial C/N, • The received satellite C/N, • The difference of time of arrival of satellite and terrestrial signals, called here Delta T (T). In the test, the satellite signal C/N is varying for each couple: C/N terrestrial/Delta T corresponding to mobile position. The different couple of parameters are provided in the Clause A.13.1.3 for each repeater configuration. To limit the number of cases, terrestrial, C/N is comprised between -10 dB and + 10 dB (5 values). The detailed configurations are described in the Clause A.13.1.3, but the correspondence between C/N and Delta T are summarized in table 7.13. Table 7.13: Summary of Terrestrial C/N values and timing differences Repeater at 30 m Repeater at 45 m Repeater at 60 m C/N (dB) Delta T (µs) C/N (dB) Delta T (µs) C/N (dB) Delta T (µs) terrestrial terrestrial terrestrial Sub Urban Case -10 57,07 -10,2 73,60 -7,6 73,60 -5,1 40,82 -4,7 48,24 -2,2 48,24 0 28,35 0,6 34,49 3 34,49 4,3 19,79 4,6 25,38 6,8 25,38 10,4 12,3 9,7 17,23 10,8 17,23 ETSI 178 ETSI TS 102 584 V1.1.2 (2010-XY) Rural Channel -9,6 107 -9,8 140,57 -6,9 140,57 -5,2 80,26 -4,8 100,30 -2 100,30 0 57,07 -0,2 73,60 2,4 73,60 4,9 40,82 5,3 48,24 7,8 48,24 10 6,99 9,4 25,38 -6,9 140,57 Two classes of hybrid channels can be defined : • The Sub Urban channel, which is the combination of SU LMS channel and Sub Urban Cost Hata channel • The Rural Channel, which is the combination of ITS LMS channel and Sub Urban Cost Hata channel 7.4.4.2 Simulation results The simulations shown here have been made in SHA/SFN configuration with Class 2 terminal only. In the simulator, with channel estimator, the reference C/N for ESR5 for 90 % fulfilment are the following: • In ITS 50 km/h, C/N =8.6 dB; • In SU 50 km/h, C/N =6.0 dB. In the laboratory tests: • In ITS 50 km/h, C/N =9.5 dB; • In SU 50 km/h, C/N=6.0 dB. The following table gives some simulations results and a comparison with the equivalent laboratory tests results. Table 7.14: Examples of simulations results in hybrid channel Modulation/ Channel / C/N values (dB) for the Delta T ( µs) Simulated Measured coding site height simulations ESR5 Laboratory ESR5 Fulfilment % fulfilment % With C/Nsat = 9,5 dB GI = 1/8 QPSK 1/3 Sub-Urban  9,7 dB terr. 17 100 100 45 m  6 dB sat QPSK 1/3 Sub urban 45 m  -10,2 terr. 73 85 80  6 dB sat QPSK 1/3 ITS  -9,8 dB terr. 140 78 72 45 m  8,6 dB sat QPSK 1/3 ITS  -9,6 terr 107 82 89 30 m  9 dB (+0,4 dB) [C/N sat = 10,0 dB] GI = 1/4 QPSK 1/3 ITS  -9,8 dB terr. 140 91 91 % 45 m  9 dB sat [C/N sat = 9,7 dB] The previous table show some degradation when  T is larger than the GT (here 44 µs with GI = 1/8). The different simulations results are quite in line with laboratory tests summarized in the following clause. ETSI 179 ETSI TS 102 584 V1.1.2 (2010-XY) 7.4.4.3 Laboratory results As detailed results are provided in the Clause A.13.1.3, a synthesis of the different tests results is provided, showing the impacts of hybrid channel. First an example of results is provided in figure 7.42: it represents the different results in sub-urban channels with 16QAM 1/3 at 40 km/h with Class 2 terminal, and repeater at 45 m height. Figure 7.42: Examples of performances in Sub Urban area at 45 m with 16QAM 1/3 From this example we can derive the following conclusions: • Most of the hybrid reception cases provide improvement compared to the satellite only reception as the required C/N is decreasing even when the differential delay is out of the GI; (see Table 7. 15) • When the terrestrial signal is under the C/N threshold (for instance C/N = 0 dB in QPSK while 2 dB are required), the resulting quality of service with satellite on (and even with very low satellite C/N values) is above 99 % ESR5; • In QPSK, only the yellow curve (with terrestrial received C/N equal to -10 dB,, and with differential delay equal to 164 % of the GI_ shows some slight degradation in the satellite reception (< 0.5 dB). Figure 7.43 represent the required C/N in satellite channel when in presence of terrestrial channel to get 90 % ESR5 fulfilment rate. All possible repeater heights have been considered. The first case is the QPSK 1/3 case, in two variants: GI = 1/8 and GI = 1/4. Abscise represents the terrestrial C/N, and different curves are representing for different repeater heights. The blue horizontal line in figure 7.43 represents the C/N required in LMS channel only. The following results have been found: • In SU channel the required C/N is 6 dB for ESR5 fulfilment of 90 %; • In ITS channel, requited C/N is 9,5 dB for ESR5 fulfilment of 90 %. ETSI 180 ETSI TS 102 584 V1.1.2 (2010-XY) QPSK : Hybrid Channel Suburbain 50 km/h Performances Overview 8 dB 7 dB 6 dB C/N for 90% availability 5 dB 30m high tower 4 dB 45m high tower 3 dB 60m high tower Satellite Only 2 dB 1 dB 0 dB -10 dB -5 dB 0 dB 5 dB 10 dB C/N (ter) Figure 7.43: QPSK Synthesis in Sub Urban channel with GI = 1/8 In fact the points below the blue horizontal line correspond to the hybrid mode improvement region while the points above correspond to the degradation region, In the case of the 30 m height and terrestrial C/N= -5 dB, the required availability is obtained with satellite channel C/N of only 4 dB. It is remarked that the maximum loss of 1 dB in SU channel for GI = 1/8 occurs when the terrestrial C/N=-10 dB. At this terrestrial C/N the differential satellite/terrestrial delay is outside the guard interval. as it corresponds to Delta T of 140 µs, which represents 314 % of GI = 1/8,and 157 % at GI = 1/4. For the case of GI = 1/4 reported in figure 7.44, the critical differential cases have disappeared, and there is improvement for the full terrestrial C/N range. QPSK : Hybrid Channel Suburban 50 km/h Performances Overview 8 dB 7 dB 6 dB C/N for 90% availability 5 dB 30m high tower 4 dB 45m high tower 3 dB 60m high tower Satellite Only 2 dB 1 dB 0 dB -10 dB -5 dB 0 dB 5 dB 10 dB C/N (Ter) Figure 7.44: QPSK Synthesis in Sub Urban channel with GI = 1/4 For the rural channel case, the experimental findings are reported in figure 7.45 for GI = 1/8 and in figure 7.46 for GI = 1/4. ETSI 181 ETSI TS 102 584 V1.1.2 (2010-XY) QPSK : Hybrid Channel Rural 50km/h Performances Overview 14 dB 12 dB 10 dB C/N for 90% availability 8 dB 30m high tower 45m high tower 6 dB 60m high tower Satellite Only 4 dB 2 dB 0 dB -10 dB -5 dB 0 dB 5 dB 10 dB C/N (Ter) Figure 7.45: QPSK Synthesis in Rural Channel with GI = 1/8 The figure 7.45 results with GI = 1/8 shows some degradation (up to 3.3 dB) for 45 and 60 m repeater height. Looking at figure 7.46 it is noted that the larger GI makes losses negligible i.e. less than 0,5 dB. QPSK : Hybrid Channel Rural 50km/h Performances Overview 12 dB 10 dB C/N for 90% availability 8 dB 30m high tower 6 dB 45m high tower 60m high tower 4 dB Satellite Only 2 dB 0 dB -10 dB -5 dB 0 dB 5 dB 10 dB C/N (Ter) Figure 7.46: QPSK Synthesis in Rural Channel with GI = 1/4 Figure 7.47 provides the results for 16QAM modulation with GI = 1/8. No results are available for 16QAM modulation with GI = 1/4. It is expected that GI = 1/4 will give the same improvements as for QPSK 1/3. ETSI 182 ETSI TS 102 584 V1.1.2 (2010-XY) 16QAM : Hybrid Channel Suburbain 50 km/h Performances Overview 12 dB 10 dB C/N for 90% availability 8 dB 6 dB 30m high tower 45m high tower 4 dB 60m high tower Satellite Only 2 dB 0 dB -10 dB -5 dB 0 dB 5 dB 10 dB C/N (Ter) Figure 7.47: 16QAM Synthesis in Sub Urban channel with GI = 1/8 In all cases, there is an improvement compared with Satellite Channel Only or terrestrial channel only, except for 45 m at C/N = -10 dB, with 0,5 dB degradation. Figure 7.48 provides results for the case of 16QAM modulation with rural channel and GI = 1/8. This graph shows some degradation in particular for a tower at 60 m height but improvement have been measured for terrestrial C/N above 0 dB. As said above, the case of 1/4 with 16QAM 1/3 was not performed, but as degradation with 1/8 was lower for 16QAM than for QPSK, with GI = 1/4, it is expected that degradation will be lower. 16QAM : Hybrid Channel Rural 50km/h Performances Overview 18 dB 16 dB 14 dB C/N for 90% availability 12 dB 10 dB 30m high tower 8 dB 45m high tower 6 dB 60m high tower Satellite Only 4 dB 2 dB 0 dB -10 dB -5 dB 0 dB 5 dB 10 dB C/N (ter) Figure 7.48: 16QAM Synthesis in Rural with GI = 1/8 For the case of MPE-IFEC, not all the cases have been measured and only GI = 1/8 has been considered. The most ETSI 183 ETSI TS 102 584 V1.1.2 (2010-XY) representative cases are summarized in table 7.16, which represents the ESR5 fulfilment ratio (three bottom rows) for various terrestrial C/N values and Delta T/GI ratio. The satellite C/N is set to 12 dB which provides 90 % ESR 5 fulfilment for SU LMS channel at 50 km/h. Table 7.16: Summary of results for MPE-IFEC MPE-IFEC Delta T/GI ratio 0% 50% 100% 150% 200% C/N terrestrial ESR 5 fulfilment ratio -6 dB 90% 90% 87% 68% 62% 0 dB 92% 92% 91% 51% 8% 6 dB 100% 100% 100% 99% 94% From the table above we can derive the following conclusions: • When the terrestrial signal is strong (C/N = 6 dB for instance) and whatever the differential delay, the improvement is always excellent: 100 % ESR5 when differential delay is up to 100 % of the GI. When differential delay exceeds 100 % of the GI, some improvement is also noticeable . • When terrestrial signal C/N is around 0 dB and with a differential delay below 100 % of the GI, the results are good. The performance is degraded when the differential delay increases, but we observe ESR5 fulfilment reduction to 8% corresponding to only 2 dB C/N degradation due to the sharp roll-off of the ESR5 curve. • When the terrestrial signal is weak, (C/N around – 6 dB), there is only a slight degradation when differential delay is equal to 100 % of the GI (around 0.1 dB), but some degradation for larger delays, quite low in terms of C/N, even if the ESR (5) value varies quite a lot. 7.4.4.4 Conclusions on laboratory measurements These previous clauses provided an extensive analysis of the hybrid SH-A-SFN architecture. The following conclusions can be derived: • The hybrid channel architecture gives an evident gain of quality of services, as it increases in most of the cases the ESR5 fulfilment ratio. • In the northern direction the differential delay is below 100 % of the GI even at long distance from the repeater. In this case even for GI = 1/8, and there is no risk of terrestrial interference to the satellite component, as the terrestrial signal becomes very weak at the edges. • On the southern direction (in sight of the satellite), the hybrid combining gain is noticeable, even at more that 100 % of the GI. When the delay is more than 150 % of the GI, there can be some interference at the far edges, but they can be mitigated by using a large GI or by careful network planning, focusing power towards the northern part corresponding to sub urban areas. 7.4.4.5 Field trial results 7.4.4.5.1 Results from Field Trials – Italy, January 2010 Within the J-ORTIGIA Project [i.33], field trials have been carried out in the nearby of Pisa (Italy) during January 2010. In the field test area, a complementary ground component (CGC) with two terrestrial repeaters has been deployed. The position of the two CGC repeaters (named MBI-1 and MBI-2) with respect to the test area is indicated in Figure 7.49. The two repeaters, both located at a rough height of 15 meters, with the aim to cover the city centre of Pisa, equipped with a 50 W amplifier over a unique sector, providing an EIRP of ~47dBm+18 dB (antenna gain) – 3 dB (losses) = 62 dBm. Both the repeaters were directly pointing the center of Pisa. The CGC intends to complement the satellite coverage provided by the W2A satellite. ETSI 184 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.49: Position of the CGC repeaters over the test area (above). The route used for the measurements is tracked (red-urban, blue-suburban, green-tree shadowing, below). Pisa, Italy. These pictures have been obtained via GoogleMaps (maps.google.com). The route used for the tests is depicted in Figure 7.49. The route was selected to assess the performance in urban and sub-urban scenarios. Most of the route is located in the city of Pisa where urban environment is the predominant one. The first and the last part of the route are characterized by an open sub-urban environment over a fast route on the north-south axis. Two SH-B configurations have been tested on the field: • A class-1 configuration, with satellite component configured with modulation QPSK and coding rate 1/2, while the CGC part is configured with 16-QAM and code rate 1/3. The scheme is complemented by an MPE- iFEC with B=17 and S=9, with nominal coding rate 2/3. The MPE-iFEC duration is 26 seconds ((B+S)*T0, with B=17, S=9 and T0 = 1 sec (burst periodicity)). The actual measured code rate for MPE-iFEC is actually lower, due to padding in the ADT tables, and it is close to ½. Hence, the overall efficiency of the class-1 configuration for the TDM part is approximately 2*(1/2)*(1/2)=0.5 b/s/Hz. • A class-2 configuration, with an interleaver of 5.4 seconds. The satellite component is protected by a rate 1/3 code, with 8-PSK modulation, while the CGC is still based on a 16-QAM signalling and on coding rate 1/3. Hence, the overall efficiency of the class-2 configuration for the TDM part is approximately 3*(1/3)=1 b/s/Hz. For the tests, two receivers provided by two distinct manufacturers have been simultaneously used. The measurement setup has been organized in a way that the two receivers had access to the same antenna signal. The receivers were operated with a G/T of -21,5 ÷ -22 dB / ºK. The LOS C/N measured on the satellite component is around 9 dB. Tests have been carried out with 3 different network topologies: satellite-only (with CGC switched on) trials, terrestrial- only trials, and code-combining trials. A summary of test results in terms of service availability (ESR5(20) criterion) is provided in Table 7.17. The Class-1 configuration is clearly advantaged by its lower spectral efficiency with respect to the Class-2 configuration. However, in both cases, the code combining feature has ensured a seamless hand over between satellite and terrestrial signals that is visible in the good service availability percentages achieved in hybrid reception. For the Class-1 configuration, a service availability close to 95% is achieved in hybrid reception, while a terrestrial-only reception would provide roughly 73%. Similarly, for the Class-2 configuration, a service availability of 90% can be achieved thanks to code combining, while in the terrestrial-only case it would not exceed 70%. In the table provided next, the performance of the two receivers in different sessions is depicted as well, confirming the stability of the obtained results Table 7.17: Summary of the performance achieved during the field trials. Pisa, Italy, January 2010. ETSI 185 ETSI TS 102 584 V1.1.2 (2010-XY) Average values Class-1 configuration service Class-2 configuration service availability (ESR5(20) criterion) availability (ESR5(20) criterion) Spectral Efficiency ~0.5 b/s/Hz ~1 b/s/Hz Terrestrial-only 71% 66.5% Satellite-only, CGC on 52% 40% Code combining 94% 90% Complete results RX 1 RX 2 RX 1 RX 2 Terrestrial-only 72% 70% 68% 65% Satellite-only, CGC on 55% / 48% 54% / 52% 39% 40% Code combining 94% / 95% / 95% 89% / 91% / 90% 93% 90% More insights on the performance of the system can be obtained by evaluating the ESR5(20) criterion over location. The results for the class-1 case are provided in Figure 7.50, while those for the class-2 configuration are presented in Figure 7.51. Figure 7.50: Service availability vs. location, Class 1 configuration. Terrestrial only (left), Satellite only (mid), code combining (right). Red spots identify areas in which the ESR5(20) criterion is not fulfilled. Receiver 1 (top) and 2 (bottom). In both cases, the satellite component seems to suffer mostly in the parts of the route which are developed along the East-West axis. Interestingly, in areas where neither the satellite component nor the CGC alone can provide coverage, their combination is successful. This is especially evident in Figure 7.51, on the western and in the northern parts of the route. ETSI 186 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.51: Service availability vs. location, Class 2 configuration. Terrestrial only (left), Satellite only (mid), code combining (right). Red spots identify areas in which the ESR5(20) criterion is not fulfilled. Receiver 1 (top) and 2 (bottom). Conclusions These trials confirm the advantage brought by the hybrid coverage of DVB-SH systems. Indeed, satellite and terrestrial coverage not only prove to be quite complementary, but moreover the hybrid code combining mechanism allows receiving the signal in areas where neither the satellite nor the terrestrial signal was correctly received. 7.5 Synchronization This clause explains how synchronization is achieved in a DVB-SH network. Distinction is made on the different hybrid radio frequency configurations (SFN and non SFN). The synchronization aspects which are not interleaver-related are also given in this clause for all types of DVB-SH networks. The following items are addressed: • distribution network delays and the use of the SHIP packet for synchronization; • compensation of the satellite round-trip-delay using the SHIP mechanisms. • SFN synchronization for OFDM transmitters • interleaver delay alignment for SH-A-MFN and SH-B networks 7.5.0 Introduction to time alignment of satellite / terrestrial signals under consideration of SHIP signaling This clause is related to the alignment of the satellite and terrestrial signals on air under consideration of the interleaver transit delays and the signaling via SHIP. For an SH-B network it is the target to ensure the time requirements given in clause 7.5.2 allowing for code combining in the receiver. As an OFDM transmitter in an SH-A-SFN network from the beginning does not know about the network condition the first processing steps are also applicable it. Whether time alignment under consideration of the interleaver delay is applicable to an SH-A-MFN network is not answered in this document, as no guidance on code-combining for SH-A-MFN is given in [1]. In particular the different SH framing imposes special treatment. Thus this document concentrates on SH-A-SFN and SH-B networks only. This section explains how the alignment of the TDM and OFDM SH frames can be performed at the transmission side using the SHIP signalling, i.e., in an automatic way without requiring any manual configuration of the transmitters. Both TDM and OFDM transmitters have to • Check if the tdm_function() is present in the SHIP packet [1]. ETSI 187 ETSI TS 102 584 V1.1.2 (2010-XY) • If tdm_function() is present: Calculate the difference of latency (in time) between both time interleavers. • If tdm_function() is present: Adjust the On-Air start of SH frame. This adjustment consists in the compensation of the difference of latency (in time) between the two time interleavers on the path with the smaller latency; • transmit the signal on-air under further consideration of the synchronization_time_stamp in the SHIP;. The complete procedure starting from parsing the SHIP packet is shown in the figure below:                                                      Figure 7.52: Rule to compensate the time interleaver latency based on the SHIP Interleaver delay compensation is only relevant for transmitters in an SH-B network. In case no TDM component is available in the network then the OFDM transmitters do not need to compensate any interleaver delay, this is covered in section 7.5.1. In an SH-B network, if the transmitter has the smallest transmit delay, it needs to update the transit delay with the largest value so that on-air start of SH-frame is common to all transmitters and given by the following formula: on_air_SH_frame_startcomp = on_air_SH_frame_start + largest_interleaver_transit_delay where largest_transit_delay = max (OFDM_interleaver_transit_delay; TDM_interleaver_transit_delay), and on_air_SH_frame_start is defined in 7.5.1 for the OFDM transmitters and in 7.5.2 for the TDM transmitters. where OFDM_interleaver_transit_delay and TDM_interleaver_transit_delay have to be calculated according to (1) and (2) in clause Erreur ! Source du renvoi introuvable.. In practice, this implies a form of time-compensation on the transmitter having the smallest transit time. However the time-compensation technique is only one possible implementation and cannot be mandated. The standard specifies only a time requirement, which is compatible with different implementations, at MPEG2, interleaver, time… and is easily ETSI 188 ETSI TS 102 584 V1.1.2 (2010-XY) understood by existing terrestrial modulators that are used to such time synchronization in traditional SFN networks. In clause 7.5.2 more details are provided. 7.5.1 Transmitter configuration in satellite-terrestrial SFN 7.5.1.1 Parameter selection In the case of satellite-terrestrial SFN operation (SH-A and SFN), full synchronization between the satellite component and the terrestrial network is necessary. The terrestrial network includes all terrestrial repeaters belonging to one SFN cell. Throughout the whole DVB-SH network, different non-overlapping SFN cells may exist which have to be locally synchronized to the satellite but not between themselves. Fine synchronization of terrestrial repeaters can be achieved, even when inclined GEO satellites are used as space component. This is achieved via: • coarse time and frequency synchronization performed on the satellite signal inside the hub; with this coarse synchronization, the satellite signal is perfectly synchronized with the hub location; • fine time and frequency synchronization performed at the terrestrial repeaters to compensate for variations within the satellite coverage, so between hub and SFN terrestrial re-transmitters positions. This synchronization can be performed on a per-repeater granularity. With these two steps of synchronizations, it is regarded feasible that the satellite signal can be synchronized with a remaining time error in the range between -5 % and +5 % of the guard interval length at each repeater cell centre. The difference of propagation time between the centre and the border of the repeater cell depends mainly on the configuration of the terrestrial transmitters, but also on the size of the terrestrial cell. Frequency synchronization inside the network shall guarantee for a small degradation with respect to the Doppler influence. For this reason the frequency accuracy shall be within +/-5 % of the OFDM subcarrier spacing. The information given below shows the order of magnitude for worst-case conditions (given a certain cell size). The principle behind the calculation of the delay called δd is given in figure 7.53. The following formula can be applied: δd = distance(satellite, repeater) + radius(repeater) – distance(satellite, receiver) Figure 7.53: Delay calculation principle for SFN operation with 1 repeater The distance is a function of longitudinal difference between repeater centre and satellite, latitude of the repeater centre and radius of the repeater. A parametric analysis shows the impact of the longitudinal difference (see figure 7.54). ETSI 189 ETSI TS 102 584 V1.1.2 (2010-XY) Delay as a as a function of latitudelongitude Delay function of latitude and and longitude 70 Latitude=70° 65 Latitude=60° Latitude=50° 60 Latitude=40° 55 delau (usec) Latitude=30° 50 Latitude=20° 45 Latitude=10° 40 35 Latitude=0° 30 0 10 20 30 40 50 60 70 80 delta_longitude (° ) Figure 7.54: Delay as a function of latitude and longitude difference between satellite and terrestrial repeater cell, for a cell radius of 10 km Taking the asymptotic values given with a delta_longitude at around 70°, the influence of the repeater radius can be displayed. Delay as a function of repeater radius function of repeater radius max_delta_longitude Max dellay as a and latitude for a given 200 180 160 140 120 Delay (usec) 100 80 GI=1/4 60 40 GI=1/8 20 GI=1/16 0 0 5 10 15 20 25 30 Repeater radius (km) Figure 7.55: Delay as a function of the repeater radius for 70° delta_longitude It must be recalled that these are minimum distances in the worst direction. In the opposite direction, the distance would be larger so that the "iso-δ zone" will not be a circle. The way this delay must be actually taken into account depends on network planning options: ETSI 190 ETSI TS 102 584 V1.1.2 (2010-XY) • guard interval margins: reduction of the GI actual time due to lower constraints on repeaters synchronization; a typical value is 10 % requiring terrestrial transmitter precision of ±5 % of the guard interval value (e.g. for 2 k mode 5 MHz, a precision of 1,12 µs is required for the shortest guard interval 1/32) leading to a decrease of 10 % of the GI actual value. Room for delay spread on the terrestrial transmission component must also be considered: this value is function of the propagation environment; • propagation losses: the effect of the delay reduces when the relative power of the components increases; so, propagations losses have to be considered. Practical values for OFDM 2k mode, 5 MHz bandwidth are given in table 7.18. Table 7.18: Max cell radius to ensure SFN between the satellite and terrestrial network at edge of one repeater Max Cell radius Max delay in µs GI = 1/4 GI = 1/8 GI = 1/16 GI = 1/32 12 km 79,8 80,64 40,32 20,16 10,08 6 km 39,9 80,64 40,32 20,16 10,08 3 km 19,65 80,64 40,32 20,16 10,08 1 km 6,55 80,64 40,32 20,16 10,08 For networks having several repeaters, the same principles apply, with additional contributions from the other receivers. See figure 7.56 for details. Figure 7.56: Delay calculation principle for SFN operation with 1 repeater For the worst case assumption, the difference in distance is now: δd = 2* radius(repeater) Provided the repeaters are synchronized with the satellite, they can be placed with distance equal to twice the maximum cell radius. Distances between repeaters could of course be made greater but then their reception would not be continuous, leading to a "spotted" map. Depending on the type of network, the typical distances between repeaters would be in 2k FFT mode, 5 MHz: • for 3G-based networks: 2 km allowing guard intervals of 1/32 or higher; • for broadcast-centric networks: 12 km allowing guard intervals of 1/8 or higher. ETSI 191 ETSI TS 102 584 V1.1.2 (2010-XY) 7.5.1.2 Recommendation on network equipment 7.5.1.2.1 Principles of SFN architecture In order to ensure exact synchronization between all transmitters, the transmitters are time synchronized by SHIP, providing functionality similar to MIP. • In traditional DVB-H deployment, the MIP is inserted by dedicated equipment called MIP inserter located downstream of the IP encapsulator. • In the context of DVB-SH, the SHIP insertion is highly recommended to be placed inside the IP encapsulator as the SHIP is also in charge of conveying SH services signalling that may be related to the DVB-H layers. The SHIP enables SFN synchronization in a similar way as the MIP for DVB-T and most of the recommendations found in [DVB-T-IG] also apply as presented in figure 7.57. The SHIP includes two key parameters to enable synchronization: synchronization_time_stamp: The synchronization_time_stamp of SHIP contains the time difference, expressed as a number of 100 ns steps, between the latest pulse of the "one pulse per second" reference (derived from GPS) that precedes the start of the SH frame M+1 and the actual start (i.e. beginning of first bit of first packet) of this SH frame M+1 at the output of the SHIP inserter. The granularity is 100 ns, the maximum value is 0x98 967F, equivalent to 1 s. maximum_delay: The maximum_delay contains the time difference between the time of emission of the start of SH frame M+1 of the DVB SH signal from the transmitting antenna and the start of SH frame M+1 at the SFN adapter, as expressed by the value of its synchronization_time_stamp in the SHIP. The value of maximum_delay must be larger than the sum of the longest delay in the primary distribution network and the delays in modulators, power transmitters and antenna feeds. The granularity is 100 ns, the maximum value is 0x98 967F, equivalent to 1 s. GPS DVB-SH modulator IPE IP flows Distribution + network SH-IP insertion 1pps GPS DVB-SH modulator MPEG2 TS MPEG2 TS 10MHz GPS 1pps Figure 7.57: principle of SFN synchronization in a DVB-SH network There exist particularities that are explained below. 7.5.1.2.2 DVB-SH particularities Repetition period: the first particularity concerns the signalling period of the SHIP. There is no more any mega-frame since this has been replaced by the SH-frame. SH-frame has a repetition interval that does not depend on the mode (1k, 2k, 4k or 8k) but only on the selection of guard interval and modulation. This repetition interval is always smaller than 1 s. In general, the repetition frequency of the SH frame (and the SHIP) is larger than for the MIP. ETSI 192 ETSI TS 102 584 V1.1.2 (2010-XY) As more than 1 SH frame per OFDM super-frame may occur (QPSK 8k, 16 QAM 4k and 8k), it is recommended to check the frame-number (bit s23 and s24 of SHIP) to detect the first SH frame (number 0) of super frame, by the way, all exciters will transmit the same super-frame at the same time. Interleaver control: the second particularity is the impact of the physical layer time interleaver. It spreads the data in time, with the full spread typically larger than the GPS 1pps impulse. This "latency" must not be confused with the delays introduced by the content delivery and the transmitter play-out, which are covered by the mechanisms above. The following rules apply for SFN operation: • modulators synchronize to the SFN network by evaluating the synchronization_time_stamp + max_delay and the GPS 1 pps and 10 MHz signalling, identical to DVB-H; this will take into account all transit delays brought by the digital and analogue processes except the interleaver transit delay; • compensation is made for the interleaver latency. This is done on the transmitter side taking care of the start of an SH-frame on the OFDM signal according to clause 7.2.4.1. No compensation is made on the receiver side, as the structure of the time interleaver is inherently self-synchronizing. The shortest tap L[0] (as seen by the receiver) of the time interleaver is (by nature) always set to 0, forcing the very first OFDM symbol of an SH frame to carry already some useful information; • "The instant for emitting the associated following SH-frame” shall be understood as the instant of the first sample of the guard interval of the OFDM symbol that carries the first IU of the SH-frame, the first IU being considered after the rate adaptation and the padding, but before the convolutional interleaver according to [1] Figure 5.11. This first OFDM symbol is built when this same first IU of the first CW of the first SH-frame exits from the interleaver after the convolutional interleaver, but before the mapping on CU and SH framing according to [1] Figure 5.11. This definition is directly related to the definition of the “start of the SH-frame” as given in clause 7.2.4.1. It is up to each modulator to take into account the transit delays brought by the digital and analogue processes, including processing time but not the delays for the time interleaver. The time chronogram is the following:                             Figure 7.58: on-air start of SH-frame in OFDM SFN case Where the interleaver_transit_delay has to be calculated according to (1) in clause Erreur ! Source du renvoi introuvable.. The on-air start of SH-frame is therefore equal to the time-stamp signalled by SHIP STS within current second signalled by GPS PPS plus the interleaver_transit_delay given by above formula. This time is assumed to be computed with sufficient precision to ensure SFN The SFN behaviour is guaranteed because all transmitters will start transmitting the first bit of the SH frame at the same time. ( )[ ] T_trans = STSSHIPn + max_delaySHIPn 10 7 + interleaver_transit_delay .’ ETSI 193 ETSI TS 102 584 V1.1.2 (2010-XY) 7.5.1.2.3 Possible implementations • To sign the emission date of the following SH-frame M + 1, the SHIP inserter will copy into the 3-byte valid synchronization_time_stamp field of the SHIP packet contained in the SH-frame of index M the value that will be reached by its own counter when it will emit the first bit of the first packet of the SH-frame M + 1. • An offset value, entered by the human operator of the distribution network, is then inserted by the SHIP inserter in the 3-byte field maximum_delay. This "offset" field must be greater than the maximum delay spread introduced by the network and is expressed as a certain number of 100 ns long (1/10 MHz) periods. • Due to its a priori knowledge of the PID of the SHIP, each channel modulator will extract the SHIP from the SH-frame M of the incoming MPEG-TS and add both time field synchronization_time_stamp and maximum_delay. Then it waits for its local counter to reach the resulting value before inserting the associated following SH-frame M + 1 into the time interleaver. 7.5.2 Transmitter configuration in non-SFN hybrid networks 7.5.2.1 Overview For configurations without SFN between satellite and terrestrial component (e.g. SH-A non-SFN or SH-B), the receiver has the possibility to combine the signals from different reception paths. Although not on-frequency, this can be considered as a kind of SFN, as the “same” content is transmitted. The same rules for the use of the (terrestrial) SFN configuration apply as in SFN operation; however the satellite and terrestrial ground component have to meet less stringent timing and frequency accuracy requirements. To allow combining between satellite and terrestrial signals, the alignment between the two components has to be ensured on the base of an SH frame. Therefore, the following requirements apply: • at the terrestrial transmitter location, the time difference between the satellite and terrestrial component must be in the range between -5 ms and +5 ms (see Note), • no specific requirements are made for the frequency synchronization between the two components; • the latencies of the different interleavers on satellite and terrestrial transmission have to be compensated to allow for proper alignment with minimum buffer requirements in the receiver. Note: The +/- 5 ms requirement is below +/-5 % of an SH frame duration for most of the possible DVB-SH waveform configurations, therefore the +/- 5% have been discarded. 7.5.2.2 DVB-SH particularities The requirements to compensate the latencies of the different interleavers on satellite and terrestrial transmission is a function of the time interleaver, the topic was introduced in clause 7.5.0. Like the SH-A-SFN case it is important to clearly define the on-air timing for the satellite and terrestrial components (OFDM and TDM or OFDM)  For OFDM the instant for "emitting the associated following SH-frame" shall be understood as the instant of the first sample of the guard interval of the OFDM symbol that carries the first IU of the SH-frame before the interleaver. This definition is directly related to the definition of the “start of the SH-frame” as given in clause 7.2.4.1. It is up to each modulator to take into account the transit delays brought by the digital and analogue processes, including processing time but not the delays for the time interleaver (same definition as in SH-A-SFN)  For TDM the instant for “emitting the associated following SH-frame” shall be understood as the instant of the first sample of the PL slot that carries the SF right in front the first data IU of the SH-frame, the first data IU being considered after the rate adaptation, but before the convolutional interleaver according to [1] Figure 5.12. This first PL slot is built with the SF when this same first data IU of the first CW of the first SH-frame exits from the interleaver after the convolutional interleaver, but before the mapping on CU and SH framing according to [1] Figure 5.12. ETSI 194 ETSI TS 102 584 V1.1.2 (2010-XY) This definition is directly related to the definition of the “start of the SH-frame” as given in clause 7.2.4.2. It is up to each modulator to take into account the transit delays brought by the digital and analogue processes, including processing time but not the delays for the time interleaver. For TDM also the duration of the SF has to be considered as the SH-frame start is corresponding to the SF and not to a data IU. Additionally the pilot field in front of the according PL slot has to be considered. Even though the formulas for interleaver delay compensation have an accuracy of 100 ns it has to be kept in mind that clause 7.5.2.1 imposes a less stringent requirement for OFDM / OFDM in SH-A-MFN or OFDM / TDM alignment in SH-B, thus the accuracy of the calculation can be reduced. The duration of the TDM SF can be calculated according to the following formula: 6528 . TDM_SF_dur ation = (TDM_Modulation * TDM_SymbolRate ) Accordingly the duration of the pilot field in front of a PL slot is: 80 . TDM_pilot_ duration = TDM_SymbolRate 7.5.2.3 Possible Implementations of Interleaver Delay Compensation There are multiple solutions to achieve interleaver delay compensation inside the transmitter, which one is used is left open to implementers. Note: It is important to keep in mind that even the descriptions below are given for the case where the satellite interleaver transit delay exceeds the one for the terrestrial path this can not be assumed to be always true. The same rule applies to the satellite modulator if the terrestrial interleaver transit delay is the longer one. At least the following implementations are possible and listed below for information: 7.5.2.3.1 Solution 1 The solution consists in using the existing capability of terrestrial modulators to synchronize at 100ns in order to comply with any time reference objective. The principle is then to know which time value is sought at the output of the modulator. This solution makes usage of a large physical buffer after the interleaver output. The estimation of the buffering time is derived from difference in time between satellite and terrestrial interleaver transit times. It can be seen that the setting of the modulator will resort to adding a delay of: largest_interleaver_transit_delay - terr_interleaver_transit_delay . But how this delay is fixed is completely left for implementation to the transmitter manufacturer. This solution is depicted below in Figure 7.59. The same solution (by exchanging satellite with terrestrial modulators) applies to the satellite modulator if the terrestrial interleaver transit delay is the longer one. 7.5.2.3.1 Solution 2 This solution is similar to the one before, but puts the additional delay into the interleaver memory. Thus, this solution corresponds to the conceptual view as given in Figure 7.21.. Regarding the depiction in Figure 7.59 the only difference of this solution is to replace the sum of the two arrows ‘terr_interleaver_transit_delay’ and ‘Additional buffer after interleaver’ simply be the new ‘terr_interleaver_transit_delay’, which in fact then is the same length . The same solution (with exchanging satellite by terrestrial modulators) applies to the satellite modulator if the terrestrial interleaver transit delay is the longer one. ETSI 195 ETSI TS 102 584 V1.1.2 (2010-XY) 7.5.2.3.1 Solution 3 The solution consists in using the existing capability of terrestrial modulators to synchronize with the GPS PPS. The principle is that if we need to synchronize an additional delay of x seconds (x>1), we can simply delay at MPEG2 level the required number of seconds and use an (internal) updated value of the SHIP time stamp and max_delay. The main advantage is that the buffer is now mainly at MPEG2 level instead of physical (coded) level, inducing large gains of memory. The figure below tries to capture the principle. Followed steps by MPEG2 pre-processor: 1) after reception of the SH frame ‘n’, store the MPEG2 stream for a duration equal to: sat_interleaver_transit_delay – terr_interleaver_transit_delay+maximum_delay-MPEG2_processing_max_delay Where MPEG2_processing_max_delay is the maximum delay to process the MPEG2 SH frame inside the transmitter 2) computes an updated STS’ and max_delay’ as follows: max_delay’ = MPEG2_processing_max_delay  (STSSHIPn + max_delaySHIPn )    STS' =  + sat_interleaver_transit_delay - terr_interleaver_transit_delay [10 7 ]  - MPEG2_processing_max_delay    This {STS’; max_delay’} will be used in replacement of existing {STS; max_delay}, but the STS and max_delay value will not be modified within the SHIP to keep code combining possible. 3) Then the MPEG2 pre-processor injects the stream at maximum_delay’ before expiration of STS’. This solution is depicted below in Figure 7.60. The same solution (by exchanging satellite with terrestrial modulators) applies to the satellite modulator if the terrestrial interleaver transit delay is the longer one. ETSI 196 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.59: Description of Solution 1 (e.g. satellite path with longer interleaver transit delay) Figure 7.60: Description of Solution 3 (e.g. satellite path with longer interleaver transit delay) ETSI 197 ETSI TS 102 584 V1.1.2 (2010-XY) 7.5.3 Receiver synchronization and re-synchronization This clause is for information. To successfully decode one FEC packet, various modules in the receiver have to be synchronized to the block transmission structure of the turbo code. This clause tries to summarize the different steps for a successful decoding and presents different possible strategies. The use of time slicing introduces some constraints on the synchronization, as receivers will have to either completely re-synchronize at the beginning of each burst, or use prediction from internal oscillators. This clause describes various strategies to synchronize both the OFDM and the TDM part with the SH frame. The first two methods do not reuse any existing knowledge from previous time-sliced bursts; the third one reuses information derived from previous time-sliced bursts. 7.5.3.1 SH frame synchronization strategy 1 (without a priori information) If no prediction of the SH frame structure is provided by the receiver, the synchronization to the SH framing has to be recovered at each switch-on. For OFDM, the TPS bits are used to recover the framing. By reading the TPS of the OFDM super frame, the receiver discovers the following information: • s23 and s24: OFDM frame number; • s35: OFDM super frame number in SH frame. However, the bit s35 acquisition is only required in modes where the number of super frames per SH frame is larger than 1, i.e. in QPSK 1k, QPSK 2k, and 16QAM 1k modes. For all other modes, the bit s35 is not used. After a maximum of 2 OFDM super frames (for the combinations as mentioned above) and 1 OFDM super frame for all other modes, the receiver is able to determine the SH framing and, thanks to the coding parameters already acquired, derive exact boundaries of the code words. Actually, the maximum duration can be 2 (respectively 1) OFDM super frame(s) if the receiver uses this strategy. For TDM, the SOF preamble is used to derive the boundaries of the SH frame. The maximum duration is identical to OFDM as the SH frame duration has been chosen identically for OFDM and TDM. 7.5.3.2 SH frame synchronization strategy 2 (without a priori information, for OFDM only) The second strategy is only applicable to OFDM and is tied to the absence of TDM. It is based on the alignment of interleaver and capacity units with the OFDM symbols and founds on the pattern of the scattered pilots. Without knowing the SH framing, the de-interleaver can immediately start the de-interleaving process, as soon as the first group of 4 OFDM symbols has been acquired. This is possible due to the self-synchronizing structure of the convolutional interleaver. Each 48 interleaver units (48*126 bit = 6 048 bit) can be handled as a group and fed into the de-interleaver. For any 4 OFDM symbols, the number of bits to be processed is an integer multiple of 6 048 bit: • 1k, QPSK, 4 symbols equivalent to 756*4*2 bit = 6 048 bit; • 8k, 16QAM, 4 symbols equivalent to 6048*4*4 bit = 96 768 bit = 16*6 048 bit. After a sufficient de-interleaver duration (to receive enough parity data for decoding), it is possible to perform a limited number of "blind decoding trials". 7.5.3.2.1 Option 1 The number of tests is limited because the code words are aligned with the pattern of scattered pilots and, as a consequence, are aligned with the de-interleaver cycles. In practice, the number of cycles is equal to: ETSI 198 ETSI TS 102 584 V1.1.2 (2010-XY) N BIL 6048 (With NBIL being the turbo block size after interleaver and puncturing), and is completely independent of the modulation. The procedure is then the following: • wait for 2 code words during 2 * N BIL ; bit_rate • test for successful decoding; • if the test is not successful, shift by one cycle by waiting for 6 048 bits during a duration of 6048 and retry bit_rate the decoding after a waiting time of 2 * N BIL . bit_rate The maximum wait duration is N BIL *  2 * N BIL + 6048  , the average is half of this value.   6048   bit_rate   7.5.3.2.2 Option 2 As soon as the receiver has done the scattered pilots acquisition, it is able to number the symbols in a row of 4 and to position the start of a code word with regard to the output of the smallest duration tap L[0]. The maximum number of tests depends on the modulation, FFT size and denominator of the fractional code rate. They are given in table 7.19 for code rate k/n where denominator n value is below or equal to 5. The duration of the tests is implementation dependent and depends on the turbo-decoding strategy. Typical implementations will require reception of an equivalent of 2 code words for each try, implying reception duration of 2 code words multiplied by the number of tries. The procedure is then the following: • wait for 2 code words • test for successful decoding • if the test is not successful, move to next position without waiting and restart the test In case of successful test decoding, the receiver knows the code word boundaries and also the CBCOUNTER_FB which gives the position index of the code word inside the current SH frame. So the receiver knows exactly the SH-framing. These two strategies (frame duration and decoding tries) can be applied in parallel and the first successful one will be used to minimize the acquisition time. As can be seen in table 7.19, code word acquisition is generally faster than OFDM frame or OFDM super frame acquisition except for the 1k mode. ETSI 199 ETSI TS 102 584 V1.1.2 (2010-XY) Table 7.19: Typical synchronization times for strategy 1 and strategy 2 MOD carriers 1/code_rate Maximum Option 1 Option 2 SH frame acquisition time number_of_tries Nof_tries*(2 CW Nof_tries * CW (ms) + 6 048) (ms) (ms) 16QAM 8k 2 1 16 3,58 243,712 16QAM 4k 2 1 3,58 121,856 16QAM 2k 2 1 3,58 60,928 16QAM 1k 2 2 7,17 45,696 QPSK 8k 2 1 32 7,17 243,712 QPSK 4k 2 1 7,17 121,856 QPSK 2k 2 2 14,34 91,392 QPSK 1k 2 4 28,67 45,696 16QAM 8k 3 8 35 43,01 243,712 16QAM 4k 3 4 21,50 121,856 16QAM 2k 3 3 16,13 60,928 16QAM 1k 3 3 16,13 45,696 QPSK 8k 3 4 70 43,01 243,712 QPSK 4k 3 3 32,26 121,856 QPSK 2k 3 3 32,26 91,392 QPSK 1k 3 6 64,51 45,696 16QAM 8k 4 1 61 7,17 243,712 16QAM 4k 4 1 7,17 121,856 16QAM 2k 4 2 14,34 60,928 16QAM 1k 4 4 28,67 45,696 QPSK 8k 4 1 122 14,34 243,712 QPSK 4k 4 2 28,67 121,856 QPSK 2k 4 4 57,34 91,392 QPSK 1k 4 8 114,69 45,696 16QAM 8k 5 8 94 71,68 243,712 16QAM 4k 5 7 62,72 121,856 16QAM 2k 5 5 44,80 60,928 16QAM 1k 5 5 44,80 45,696 QPSK 8k 5 7 188 125,44 243,712 QPSK 4k 5 5 89,60 121,856 QPSK 2k 5 5 89,60 91,392 QPSK 1k 5 10 179,20 45,696 7.5.3.3 SH frame synchronization strategy 3 (with prediction) This strategy is applicable to both OFDM and TDM. The receiver may provide frame prediction according to the following schemes. The prediction times are considered to be feasible with consumer product oscillators, assuming a frequency accuracy of (±) 50 ppm. For OFDM, the use of 2k mode, 5 MHz bandwidth with QPSK modulation and guard interval 1/8 has been selected for this example: • prediction of the OFDM symbol number is possible over one second within an interval of (±) 12,4 % (50 µs at a total symbol length of 403,2 µs per OFDM symbol); • prediction of the OFDM regular pilot structure (4 OFDM symbols) is possible over one second within an interval of (±) 3,1 % (50 µs at a total length of 4 OFDM symbols of 1 613 µs); • prediction of the OFDM frame structure (68 OFDM symbols) is possible over ten second within an interval of (±) 1,8 % (500 µs at a total length of 68 OFDM symbols of 27,4 ms); The length of the SH frame in OFDM frames depends on the selection of the modulation and guard interval and the selection of the FFT mode. The following extreme cases are possible: • "very short SH frame": For 8k FFT and 16QAM modulation, one SH frame maps to 1 OFDM frame; • "very long SH frame": For 1k FFT and QPSK modulation, one SH frame maps to 16 OFDM frames; ETSI 200 ETSI TS 102 584 V1.1.2 (2010-XY) Combined with the highest (9,14 MHz) symbol rate available, this maps to the following prediction accuracies: • "very short SH frame": Prediction of the SH frame structure is possible over ten seconds within an interval of (±) 0,73 % (500 µs at a total length of 68 OFDM symbols of 68,54 ms); • "very long SH frame": Prediction of the SH frame structure is possible over ten seconds within an interval of (±) 0,36 % (500 µs at a total length of 16*68 OFDM symbols of 137,09 ms). It can be concluded that, with low requirements on the receiver oscillators used, the receiver is capable of predicting the SH frame alignment for both OFDM and TDM over a time span of more than 10 s and even with oscillators worse than those used for the calculation exercise. Using this proposed scheme, re-acquisition time is only dependent on the re- acquisition time of the demodulators, but no longer on the SH framing, allowing a good prediction of code word boundaries. 7.5.4 HP/LP Synchronization in Hierarchical mode In an SFN with hierarchy, the relative alignment of HP and LP streams must be the same at all modulators, in order to make them emit identical signals. HP and LP streams may originate at different locations, and so it is likely that HP and LP SFN adapters may not emit suitably synchronised streams. Network Adapter Tx HP MPEG2-TS HP Network Adapter Transmitter Network Adapter Rx MPEG-TS Tx MPEG2-TS LP LP Figure 7.61: distribution of 2 MPEG-2 transport streams in a hierarchical modulation context A deterministic method of aligning the two streams is therefore required. As explained in [i.22], it is undesirable to align the SH-frame starts of the HP and LP streams, as this may involve large delays and result in a significant offset to the time_of_emission of one of the streams. The minimum requirement for alignment is that a sync byte of one stream (the slave stream) is coincident with the sync byte at the SH-frame start of the other stream (the master stream). Subsequently, only SH-frame information from the master stream may be used. This will ensure proper operation of various synchronised processes within the modulator. The requirement alignment is illustrated in Figure 7.62. 1 pps HP SH frame start HP 47 … 47 … 47 … 47 … 47 tSHF-HP=STSHP+maximum_delayHP LP SH frame start tSHF-LP=STSLP+maximum_delayLP tSHF-HP- tSHF-LP LP 47 … 47 … 47 … 47 … 47 … 47 … Packet n Aligned here Packet 0 Figure 7.62: principle of HP and LP synchronization The HP stream is designated as the master and the LP as the slave. The alignment reference point is the SH-frame start of the HP stream. The LP stream is delayed with respect to the HP, to ensure an MPEG-TS packet start is aligned to the ETSI 201 ETSI TS 102 584 V1.1.2 (2010-XY) reference point. The amount of this delay is less than one complete MPEG-TS packet as follows: the packet n (counting from the LP mega-frame start) is aligned to the HP mega-frame start, where n is given by: n = [(tSHF-HP - tSHF-LP) * PLP / dMF] where - PLP is the number of bytes in one LP SH frame and dMF is the duration of one SH frame; - n must be rounded down to the nearest integer, i.e. 2,7 would become 2; -2,3 would become -3. This expression caters for the slave LP SH frame start either lagging (n is positive) or leading (n is negative) the master HP SH frame. The time is in 100 ns units.” 7.6 System throughput calculations DVB-SH capacity at MPEG TS interface level may be calculated from the parameters defined in the waveform definition [1]. This is performed in calculating first the number of MPEG TS packets in an SH Frame, and then the duration of the SH frame. The ratio of these two numbers give the bit rate capacity of the system. Please note that the throughput calculation is independent of the selection of FFT mode in OFDM. To support seamless hand-over between OFDM and TDM, the SH frame length of the TDM part has been aligned to the SH frame length in OFDM. Therefore, the TDM parameters can not be calculated independently but imply a selection of OFDM parameters which is represented in the following tables. The Calculation results are for information. 7.6.1 Calculation of the number of MPEG TS packets per SH Frame Each turbo code word contains exactly 8 MPEG TS Packets. For OFDM, table 5.11 of the waveform definition [1] gives the number of turbo code words in an SH frame as a function of the FEC code rate. For TDM, table 5.12 of the waveform definition [1] gives the number of Capacity Units (CU) per SH Frame. A CU is a set of 2 016 bits. From table 5.8 of the waveform definition [1], the number of CU per turbo code word may be calculated for each code rate. The total number of turbo code words in an SH frame is then the integer part of the division of the number of Capacity Units (CU) per SH Frame by the number of CU per turbo code word. 7.6.2 Calculation of the SH Frame duration The SH frame length is defined as a function of the length of OFDM frames. Therefore, the frame length of both an SH-frame in OFDM and TDM mode are identical. The number of OFDM Frames per SH frame is defined in the table 5.10 of the waveform definition [1]. Each OFDM Frame is composed of 68 OFDM Symbols. The duration of an OFDM symbol is given in tables 5.22, 5.24 and 5.26 of the Waveform Definition document [1]. 7.6.3 Typical MPEG-TS bit rates for OFDM The following tables give the MPEG TS bit rates in Mbps of the OFDM waveform for various DVB-SH configurations. ETSI 202 ETSI TS 102 584 V1.1.2 (2010-XY) 7.6.3.1 Channel Bandwidth 5 MHz, OFDM Table 7.20 GI 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 QPSK 1/4 1,333 1,481 1,679 1,876 2,222 2,666 3,357 4,443 1/8 1,481 1,646 1,865 2,085 2,468 2,962 3,730 4,937 1/16 1,568 1,742 1,975 2,207 2,614 3,136 3,950 5,227 1/32 1,616 1,795 2,035 2,274 2,693 3,231 4,069 5,386 16QAM 1/4 2,666 2,962 3,357 3,752 4,443 5,332 6,714 8,887 1/8 2,962 3,291 3,730 4,169 4,937 5,924 7,460 9,874 1/16 3,136 3,485 3,950 4,414 5,227 6,273 7,899 10,455 1/32 3,231 3,591 4,069 4,548 5,386 6,463 8,139 10,772 7.6.3.2 Channel Bandwidth 1,7 MHz, OFDM Table 7.21 GI 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 QPSK 1/4 0,427 0,474 0,537 0,600 0,711 0,853 1,074 1,422 1/8 0,474 0,527 0,597 0,667 0,790 0,948 1,194 1,580 1/16 0,502 0,558 0,632 0,706 0,836 1,004 1,264 1,673 1/32 0,517 0,574 0,651 0,728 0,862 1,034 1,302 1,723 16QAM 1/4 0,853 0,948 1,074 1,201 1,422 1,706 2,149 2,844 1/8 0,948 1,053 1,194 1,334 1,580 1,896 2,387 3,160 1/16 1,004 1,115 1,264 1,413 1,673 2,007 2,528 3,346 1/32 1,034 1,149 1,302 1,455 1,723 2,068 2,604 3,447 7.6.4 Typical MPEG-TS bit rates for TDM The following tables give the TDM MPEG TS bit rates in Mbps of the TDM waveform for various DVB-SH configurations. 7.6.4.1 Channel Bandwidth 5 MHz, TDM, with 15 % roll-off and QPSK in OFDM Table 7.22 TDM Modulation QPSK Roll Off : 15 % OFDM modulation : QPSK OFDM GI 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 1/4 1,530 1,728 1,925 2,222 2,567 3,110 3,900 5,184 1/8 1,536 1,755 1,975 2,249 2,633 3,127 3,950 5,266 1/16 1,510 1,684 1,917 2,207 2,556 3,078 3,891 5,169 1/32 1,556 1,676 1,915 2,214 2,573 3,112 3,890 5,146 Table 7.23 TDM Modulation 8PSK Roll Off : 15 % OFDM modulation : QPSK OFDM GI 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 1/4 2,320 2,567 2,913 3,308 3,900 4,690 5,826 7,800 1/8 2,359 2,633 2,962 3,346 3,950 4,718 5,924 7,899 1/16 2,323 2,614 2,904 3,311 3,891 4,705 5,866 7,841 1/32 2,334 2,573 2,872 3,291 3,890 4,668 5,805 7,779 ETSI 203 ETSI TS 102 584 V1.1.2 (2010-XY) Table 7.24 TDM Modulation 16APSK Roll Off : 15 % OFDM modulation : QPSK OFDM GI 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 1/4 3,110 3,456 3,900 4,443 5,184 6,221 7,800 10,417 1/8 3,127 3,511 3,950 4,498 5,266 6,308 7,899 10,532 1/16 3,136 3,485 3,891 4,472 5,227 6,273 7,841 10,455 1/32 3,112 3,411 3,890 4,428 5,146 6,224 7,779 10,353 7.6.4.2 Channel Bandwidth 5 MHz, TDM, with 15 % roll-off and 16QAM in OFDM Table 7.25 TDM Modulation QPSK Roll Off : 15 % OFDM modulation : 16QAM 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 1/4 1,481 1,679 1,876 2,172 2,567 3,061 3,851 5,134 1/8 1,536 1,646 1,865 2,194 2,523 3,072 3,840 5,156 1/16 1,510 1,626 1,859 2,207 2,556 3,020 3,833 5,111 1/32 1,436 1,676 1,915 2,154 2,513 2,992 3,830 5,146 7.6.4.3 Channel Bandwidth 5 MHz, TDM, with 25 % roll-off and QPSK in OFDM Table 7.26 TDM Modulation QPSK Roll Off : 25 % OFDM modulation : QPSK OFDM GI 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 1/4 1,432 1,629 1,827 2,074 2,419 2,913 3,653 4,888 1/8 1,426 1,536 1,755 2,030 2,359 2,852 3,566 4,718 1/16 1,452 1,568 1,801 2,033 2,381 2,904 3,601 4,821 1/32 1,436 1,556 1,795 2,035 2,394 2,872 3,591 4,787 7.6.4.4 Channel Bandwidth 1,7 MHz, TDM, with 15 % roll-off and QPSK in OFDM Table 7.27 TDM Modulation QPSK Roll Off : 15 % OFDM modulation : QPSK OFDM GI 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 1/4 0,490 0,553 0,616 0,711 0,822 0,995 1,248 1,659 1/8 0,492 0,562 0,632 0,720 0,843 1,001 1,264 1,685 1/16 0,483 0,539 0,613 0,706 0,818 0,985 1,245 1,654 1/32 0,498 0,536 0,613 0,709 0,823 0,996 1,245 1,647 Table 7.28 TDM Modulation 8PSK Roll Off : 15 % OFDM modulation : QPSK OFDM GI 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 1/4 0,743 0,822 0,932 1,058 1,248 1,501 1,864 2,496 1/8 0,755 0,843 0,948 1,071 1,264 1,510 1,896 2,528 1/16 0,743 0,836 0,929 1,059 1,245 1,505 1,877 2,509 1/32 0,747 0,823 0,919 1,053 1,245 1,494 1,857 2,489 7.7 Coexistence with other Physical layer elements This clause is for information. ETSI 204 ETSI TS 102 584 V1.1.2 (2010-XY) 7.7.1 Introduction The deployment of DVB-SH CGC raises the problem of coexistence of the DVB-SH system with other systems operating in adjacent S band, like 3G, WiFI, Bluetooth... Of particular interest is the case of coexistence with 3G networks as they are operating in the 2110-2170 MHz on the down link and 1920-1980 MHz on the uplink. The coexistence problem is in fact a two fold problem: • The first aspect is the possible impact of co-location, especially in the case of coexistence with 3G or other Mobile systems. Something that we could call short distance “interference” of course this concerns the reciprocal interference scenarios from DVB-SH to 3G and from 3G to DVB-SH. • The second aspect is the long distance interference that reflects the mutual impact of the two systems (DVB- SH and the other one), mainly at the receiver level. This second aspect is also to be studied in both directions: from DVB-SH to 3G and vice-versa. Due to the lack of measurements up to now, only the 3G case is covered, and partially. No measurements have been done in WiFi or Bluetooth environment. Most of the simulations and measurements were made on the impact of DVB-SH on 3G systems. The 3G case Concerning the 3G coexistence four aspects are covered in this clause. • The first one is specific to some deployment strategies that include co-location of 3G and DVB-SH sites with cable and/or antenna sharing; this can have an impact on the link budget • The second one concerns the impact of the DVB-SH signals on the 3G service, based on real field tests • The third one is the result of simulations on the impact of a high power CGC on a UMTS network operating in the adjacent band, which is of course the worst case. • The fourth one is the results of simulations concerning the impact of UMTS signals on DVB-SH signals, especially on satellite signals when there is no CGC.. The two first item results are based mainly on the outcomes of the SFR/Vodafone/ALU (with the help of Thales Alenia Space) experiment in Pau and also on some early Toulouse trials. 7.7.2 Site sharing The general picture of the site co-sharing is the following: ETSI 205 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.63: UMTS Base Station and DVB-SH transmitter general co location architecture The impact on the 3G downlink link budget are the following • Considering the previous antenna sharing architecture, the impact on the 3G downlink budget is 0,7 dB, due to diplexer and jumper. • O the 3G uplink , there is the same attenuation of 0,7 dB that is compensated by the TMA (Tower Mounted Amplifier) • The ACLR specification for DVB-SH in S-UMTS bands are the same as for 3G.: 45 dB for ACLR 1 and 50 dB for ACLR 2. During the field trials in Toulouse, no interference case has been noticed. On the MobileTV link budget, the influence of the site sharing is the following: • The impact on the link budgets (downlink only) is in the range 0,6-0,8 dB • The 3G antennas are generally compliant with MTV signal transmission; their transmission pattern is similar to the UMTS one as the frequency band is adjacent Impact of the DVB-SH on 3G Quality of Service During the deployment of the Pau field trial, SFR has performed different tests on their Key Indicator parameters (KIP). 3G Quality of Service was measured before, during and after installation of DVB-SH transmitters, based on the following KIP: - Call set-up success rate (CSSR) - Call completion success rate (CCSR) - Interrupted call for various reasons (refer to label “Interrupt Rate” in the Graph”) As shown in figure 7.64, all quality indicators remained unchanged after installation of the DVB-SH transmitters sharing 3G sites and antennas. ETSI 206 ETSI TS 102 584 V1.1.2 (2010-XY)          Figure 7.64: Results of SFR tests on QoS 7.7.3 Results of simulations Different simulations have been performed concerning the impact of CGC and specifically high power CGC on the reception of UMTS inside the same area. As the case of collocated CGC and 3G Base Station is not critical and there is no interference, the simulations have been concentrating on the case of non co-located sites. The simulations have taken the city of Torino as deployment site. Thus, we consider a UMTS network with a classical cellular topology, and a High Power CGC with two possible powers: 100 and 300 W. The BS configuration parameters are the following, using a three-sector antenna and 20 W power amplifier and with 17 dBi antenna gain and 3 dB feeder losses. The configurations and parameters of the different sites are gathered in table 7.29. Table 7.29: BS configuration Antenna height Max Site name ASL Number of sector Azimuth spacing AGL EIRP/sector Configuration 1 All Sites 30 m 196 m 3 120° 57 dBm The UMTS traffic parameters are the following: • Channel overhead: 10% • Cell load: 100% • Number of 12,2 kbps services: 40 • Number of 64 kbps services: 15 • Number of 128 kbps services: 10 • Number of 384 kbps services: 4 The UM “Voice service” UE parameters are listed in table 7.30. Table 7.30: Voice service UE parameters Central Antenna Power Eb/No Eb/ No Power Antenna Ec/Io Noise frequency height control forward reverse 2167,5 MHz 0,125 W Omni 1,5 m 60 dB -14 dB 5 dB 6 dB -102 dBm ETSI 207 ETSI TS 102 584 V1.1.2 (2010-XY) The “Data service” UE parameters are listed in table 7.31 where Ec/I0 is the energy per chip over noise plus interference power spectral density required for the CPICH channel and Eb/N0 is required energy per bit over noise power spectral density for the traffic channel. Table 7.31: Data services UE parameters Central Antenna Power Eb/ No Eb/ No Power Antenna Ec/Io Noise frequency height control forward reverse 2167,5 MHz 0,125 W Omni 1,5 m 60 dB -14 dB 2 dB 5 dB -102 dBm The TS terminals parameters are listed in table 7.32. Table 7.32; CGC configurations ACLR Site name Antenna height AGL ASL Number of sector Azimuth spacing EIRP/sector 1st Channel Configu 248 m Cernaia 82 m 3 120° 69 dBm 49 dB ration 1 Configu 248 m Cernaia 82 m 3 120° 64 dBm 49 dB ration 2 The summary of the different results are contained in tables 7.33 and 7.34.. Table 7.33: Influence of the CGC on the UMTS covered area UMTS only UMTS + UMTS + (REF) config1 config2 Surface (km²) covered having Ec/I0 > -14 dB 764,3336 763,0632 763,6 CPICH % covered having Ec/I0 > -14 dB 100 99,83 99,90 Difference / REF (km²) -1,2704 -0,7336 Surface (km²) covered having Eb/N0 > 5 dB 866,9536 866,9536 866,9536 Voice % covered having Eb/N0 > 5 dB 100 100,00 100,00 Difference / REF (km²) 0 0 Surface (km²) covered having Eb/N0 > 2dB 865,8448 865,2012 865,666 Datas % covered having Eb/N0 > 2dB 100 99,93 99,98 Difference / REF (km²) -0,6436 -0,1788 Table 7.34 Influence of the CGC on the number of lost users ETSI 208 ETSI TS 102 584 V1.1.2 (2010-XY) REF UMTS + config3 UMTS + config4 Eb/No min (dB) 5,19 2,76 4,54 Eb/ No max (dB) 24,43 24,43 24,43 Eb/ No ave (dB) 22,02 21,96 21,99 Nb of connected sub 17 066 17 066 17 066 % of connected sub 100 100 100 Nb of sub lost 0 0 Ec/Io min (dB) -14 -24,93 -20,57 Ec/ Io max (dB) -10,01 -10,01 -10,01 Ec/ Io ave (dB) -11,55 -11,6 -11,57 Nb of connected sub 17 066 16 911 16 966 % of connected sub 100 99,09 99,41 Nb of lost subscribers 155 100 Conclusions • We see that a high power CGC transmitter introduces a very low degradation of the UMTS services. • In the worst case (EIRP=69 dBm and high antenna height) we have a loss of only 0,4% of the covered area for the data service and 0,05% for the voice. • If the same CGC is located at a lower altitude, the interferences become negligible since there is no degradation of the voice service and only 0,07% of the area is lost for the data. • The number of lost subscribers is less than 1% in the worst case without taking into account the possible reconnection to an other Node B. • The simulations show that a high power CGC transmitter located in the same geographical area than a UMTS network and using an adjacent frequency degrades very few the UMTS service (less than 0,5% of the covered area and less than 1% of the subscribers). • The degradation zones are clearly located around the high power CGC. The rest of the covered area is not affected by the CGC. • Using interference reciprocity, one can deduce that DVB-SH will be as much impacted by UMTS Base Station operation in adjacent frequency band. 7.7.4 Theoretical impact of UMTS over DVB-SH With the increasing number of DVB-SH systems, the coexistence of existing UMTS networks and of S-Band components is becoming an important issue to be addressed. For this reason, an investigation into the reciprocal interference between the two systems is needed. The aim of this section is to make an assessment of the interference between the Italian DVB-SH network and the UMTS network. The assessment has been carried out following a variety of procedures and a specific methodology that were defined in accordance with the constraints and targets related to this issue. During the first two phases of the J-Ortigia project, the forward link DVB-SH transmissions were located in the first 5 MHz of the [2170-2200] MHz band and were therefore adjacent to the band of an Italian UMTS service provider (Vodafone). This clause and the information it contains is the work of M.B.I.Srl and the J-Ortigia Consortium done under ESA Contract N.ESTEC 21517/08/NL/US. The use, copy, redistribution of this clause, in whole or in part, is authorized only upon written permission of M.B.I. Srl and the J-Ortigia Consortium, under following conditions:  The author and the text of the entire document must remain unchanged ETSI 209 ETSI TS 102 584 V1.1.2 (2010-XY)  A clear and complete citation must be supplied referencing the author, the full title and the version of the original document 7.7.4.1 Assessment of the Interference The aim of this section is to provide an assessment of the impact of UMTS interference on the pilot Italian DVB-SH network out in place around Pisa. This is done by quantifying the SNIR degradation and associating it to specific areas around the base stations. An example of how this can affect the quality of service is provided by addressing a typical use case and identifying the relative “exclusion zone”. Finally, a set of on-field measurement extracted from a measurement campaign is presented in order to provide an example of realistic values. 7.7.4.1.1 Methodology Due to the impossibility of controlling the UMTS network and to the low level of the interference signal, a theoretical assessment based on standard specification is carried out first. For the evaluation of the reciprocal interference of DVB- SH and UMTS networks, a symmetrical study can be performed. The analysis considers several useful values and figures defined in the harmonized standards, which can be adopted as reference system characteristics. There follows a brief description of these parameters. BS/CGC Maximum Output Power, Pmax, is the mean power level per carrier measured at the antenna connector in specified reference condition. Emission Mask specifies the maximum output power that a CGC or a BS shall not exceed. This is an absolute value, specified for various output power ranges and given in function of the frequency offset from the carrier frequency of the relevant system. Adjacent Channel Leakage power Ratio (ACLR) is the ratio of the filtered mean power centered on the assigned channel frequency to the filtered mean power centered on the adjacent channel frequency. The ACLR is therefore a relative value. Minimum receiver sensitivity is specified as the minimum input level required by the receiver; this value is taken as reference in order to characterize the interference with respect to the thermal noise power. Table 7.36 summarizes the main useful information which characterizes both the UMTS and the DVB-SH network. Table 7.35: RF Characteristics Emission Mask Minimum ACLR Integration receiver Maximum output power sensitivity Network type/Unit [dB] [dBm] [dBm] [dBm] Wide Area BS No upper limit -0,75 [i.36] -45 -106+(C/N)MIN UMTS Medium Range [i.36] [i.36] [i.35] BS < +38 [i.36] -45 0,77 -105+(C/N)MIN HPA 53 DVB-SH [i.34] [i.34] according to clause 10.4.4 LPA 47 The following analysis will focus on the interference caused by the UMTS network to the DVB-SH: due to the similarity of the two systems, similar results can be expected for the interference caused by DVB-SH, which will be ETSI 210 ETSI TS 102 584 V1.1.2 (2010-XY) addressed later on. The analysis will consider medium range BS for urban environment characterization and wide area BS for rural areas. The area around the base station (BS) is characterized based on the impact of interference on the overall SNIR. The following analysis provides therefore a quantification of out-of-band emissions on adjacent systems, but does not give specific information on the quality of service, which depends on the level of the received signal power. For example, in case of DVB-SH Complementary Ground Component (CGC) being co-located with a UMTS BS, the reception could still be possible even though significant SNIR degradation is present. As derived from Table 7.35, it is assumed there is a user terminal with a noise figure NF=2 dB and with a consequent noise power of -105 dBm/5 MHz. The scope of this assessment is to evaluate the impact of the interference level on the final SNIR and in order to this: - The previous value of noise power is fixed and taken as reference - The interference power is expressed as a percentage of this noise power - The corresponding impact on the SNIR in terms of degradation is calculated - The areas where the various amount of degradation can be expected are identified and classified The results can be applied to any reception scenario by simply considering the level of received signal power in order to calculate the specific SNIR. Following this methodology it is possible to identify three areas around the base station: 1) One area where the reception is limited by the interference level; this will be called Interference limited area. Taking into account the definition of minimum sensitivity for a DVB-SH receiver, it is possible to consider a value of -105 dBm/5MHz as a threshold for the identification of the area. This corresponds to a receiver with a noise figure NF=2 dB. Here, the impact of the UMTS out-of-band power on the SH user terminal C/N is higher than 3 dB (noise temperature increase higher than 100%) over the whole area and it is exactly 3 dB on the edge. 2) An intermediate region where the impact of the interference on the C/N goes from a maximum of 3 dB to a definitely negligible degradation of 0,04 dB (which corresponds to an increase in noise temperature due to UMTS interference less than 1%, i.e. -125 dBm/5MHz) 3) The interference free area, where the impact of the UMTS out-of-band signal is definitely negligible (less than C C = −ε 1%). This is the area where the N + I N and therefore no degradation due to UMTS can be expected. This analysis is carried out both for an urban UMTS BS with a medium range output power and for a rural wide range BS. In the first case, the ACLR specified in [i.36] shall be used in order to evaluate the power of the out-of-band UMTS signal based on the in-band one. On the other hand, as no upper limit to the output power is specified, in the second case it is recommended to consider the integration of Emission Mask as an upper limit for the interference power. Both values are included in Table 7.35. These considerations will be used in the RF coverage prediction tool. Interference Limited Area It identifies the area where the interference is predominant, i.e IUMTS _ MAX > N for a receiver with a NF=2 dB. In this area, the interference limits the C/(N+I). However, reception could still be possible in the case where a higher SH signal power is received (as long as the Csh/(N+I)>(C/N)min). The radius of the interference limited zone is therefore given by the distance from the source of the interference where the following condition is fulfilled: RIL = r |: IUMTS _ MAX ( r ) = −105 dBm/5 MHz ≈ −172 dBm/Hz In this area, the impact of the interference power on the overall C/(N+I) is higher than 3 dB and is exactly 3 dB on the edge of the area. A similar condition can be also applied when considering the DVB-SH as the interference source and the UMTS as the victim system. For a medium range BS, an ACLR of 45 dB, it is possible to obtain the interference RF prediction by subtracting the ACLR from the in-band power: ETSI 211 ETSI TS 102 584 V1.1.2 (2010-XY) IUMTS _ MAX = CUMTS − ACLR By applying this result to the predictions of an urban BS in Pisa broadcasting with 38 dBm and a typical UMTS sector antenna of 18dBi it is possible to evaluate theoretically the related interference limited zone: Figure 7.65: Interference limited area (in red) for an urban UMTS BS in Pisa. Max radius 825 m With regard to the study of a rural BS, a wide range output power is considered, in order to deal with the worst case reception condition for rural cells. As for this type of BS there is not specified limit to the output power, the only limitation expressed in the standards is the UMTS Emission Mask absolute power, which can be calculated by an integral of the emission mask. • This value, indicated in, Table 7.36 is equal to -0,75 dBm for the UMTS. • The relative coverage prediction, with the -105 dBm threshold, is shown in figure 7.66. ETSI 212 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.66: Interference limited area (in red) for a rural UMTS BS in Pisa. Max radius 1380 m From the previous figures, an interference limited zone of 825 m and 1380 m is expected for the urban and rural base station respectively. This result indicates that in those areas the interference power could significantly affect the reception quality, whenever a weak S-band signal is received. However, as already underlined, the limits imposed by the standards represent only a boundary, as the real filters can perform better. The final quality of service is then determined by CSH level. Intermediate region The distance where the impact of the interference power is exactly 3 dB identify the beginning of the intermediate region. In this region the increase in noise temperature due to UMTS interference goes from a maximum of 100% (on the first edge), to a minimum of 1% (on the second edge): RIR = r |: −125 < IUMTS _ MAX ( r ) < −105 dBm/5 MHz In this region, the interference does not dominate the noise power and this is why its impact on the final C/(N+I) is less than 3 dB. However, it is possible to identify other intermediate thresholds between the two mentioned values. In particular, table 7.37 summarizes the considered thresholds. ETSI 213 ETSI TS 102 584 V1.1.2 (2010-XY) Table 7.37: Overall SNIR degradation, interference percentage and relative thresholds  C    Degradation (D)  I  N+I   % I_umts Threshold N [dB] -3 100% -105 -2 58% -107 -1 25% -111 -0,5 12% -114 -0,04 1% -125 Based on this table, the evaluation of the extension of each zone can be obtained through RF coverage predictions. In particular, the ACLR value is still used to derive the interference power for an urban BS and the Emission mask for a rural one. Figure 7.67 :Various Interference thresholds for an urban UMTS BS in Pisa. Quasi Interference free area (in green) ETSI 214 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 7.68: Various Interference thresholds for a rural UMTS BS in Pisa. Quasi Interference free area (in green) Interference Free Area The interference created by the interfering system is considered negligible when the increase of noise temperature created by this interference is less than 1%, with a consequent C/(N+I) degradation of 0,04 dB. This gives a threshold of −125 dBm/5 MHz . It is therefore possible to conclude that the interference from the UMTS system to DVB-SH is negligible if the UMTS out-of-band power is lower than this value. The radius at which this condition is fulfilled represents the BEGINNING of the interference free area: RIF = r |: IUMTS _ MAX ( r ) < −125 dBm/5 MHz * on the basis of a UE with NF=2 dB In this area, the interference does not limit the C/(N+I). For a urban BS, the interference free area begins at 3830 m distance (see Figure 3) while, as shown in Figure 4, for the rural one it begins after about 6 km. Table 7.38 summarizes the radius from the UMTS BS which characterizes each area, from the interference limited zone to the interference free area, with the intermediate region in-between. ETSI 215 ETSI TS 102 584 V1.1.2 (2010-XY) Table 7.38: Maximum Radius of each area  C    Degradation (D)  I  Urban BS Rural BS N+I   % N Max Radius [m] Max Radius [m] [dB]  I   100%  -3   825 1340 N 58%   I  < 100% -2  d < -3   902 1580 N 25%   I  < 58% -1  d < -2   1200 2340 N 12%   I  < 25% -1  d < -0,5   1560 2940 N 1%   I  < 12% -0,5  d < -0,04   3580 6045 N  I   1% d  -0,04   > 3580 > 6045 N 7.7.4.2 Exclusion zone with Satellite only reception The term exclusion zone indicates the area where, in the absence of UMTS BS, DVB-SH reception is possible (i.e. the C/N is above the minimum required), however, when switching on the BS, the interference does not allow correct reception anymore. In other words, it is the area where the following two conditions are fulfilled: C C C C ≥  & <  N  N min N + I  N min In order to simplify the evaluation of the extension of the exclusion zone, a satellite only reception scenario is considered, where the signal power is therefore fixed and nearly constant over the whole area considered. The user equipment is assumed to have a noise figure of NF=2dB. The reference satellite signal power in the city of Pisa, based on a passive antenna with NF=0 and a receiver with NF=2, is around -99,84 dBm. It is then assumed to have a QPSK signal with coding rate 1/2. For this configuration, in a LOS AWGN channel, the minimum required C/N for a TDM signal is equal to 1,6 dB (including 0,5 dB of implementation losses). In this context, the radius of the exclusion zone can be defined as: REZ = r |: −1  C  −1  CSAT   −1  C  +   =  SAT I   UMTS ( r )   N    N  min   Consequently, all the points (x,y) where : −1  C  −1  CSAT   −1 C  SAT  +   <  I   UMTS ( x, y )   N    N min   ETSI 216 ETSI TS 102 584 V1.1.2 (2010-XY) belong to the exclusion zone and no correct reception can be expected. This corresponds to the following condition on the interference power:  C  −1  CSAT  −1  IUMTS ( x, y ) > CSAT ⋅   −    N  min   N    C Considering that over the 5 MHz CSAT = -99,84 dBm, N = −105 dBm and   = 1,6dB , the exclusion zone  N  min is then characterized by a UMTS out-of-band power of: IUMTS ( x, y ) > -104 dBm/5 MHz Considering a rural BS, broadcasting with very high power, a satellite only reception would be a very critical case. With the above assumption, the resulting radius of the exclusion zone is 1280 m, as shown in Figure 7.69. Figure 7.69: Rural wide area BS exclusion zone ETSI 217 ETSI TS 102 584 V1.1.2 (2010-XY) 7.7.5 Measurement results from On-Field Campaign The UMTS BS output power is highly time variable. Due to this and to the low range of powers which characterizes the out-of-band signals, many of the on-field tests carried out cannot give evidence of the expected behaviour of interference. Some urban measurements are presented in this clause in order to provide an example of the assessment procedure and of some realistic values. The main outcome of the on-field measurement campaign is that a different and higher ACLR value has been measured in relation to an urban UMTS cell, which confirms that filters can perform also better than what specified in regulations. In this framework the determination of the ACLR is the main purpose of the assessment. 7.7.5.1 Measurements setup The measurement setup is described in figure 7.70. The test setup is composed by a passive antenna +cascaded by an external low-noise amplifier and a spectrum analyser (Agilent MXA N9020A SA). Passive Antenna EXT AMP SA Gamp=25 NFSa=22,2 NFAmp=4,4 Figure 7.70: Measurement Setup ETSI 218 ETSI TS 102 584 V1.1.2 (2010-XY) Table 7.39: Measurement Setup Spectrum analyzer: Agilent MXA N9020A DANL -151,8 dBm/Hz Noise Figure 22,2 dB External Amplifier Spin Electronics WBP 18 - 007 Gain 25 dB NF @ 290 K 4,4 dB Passive Antenna Procom Polarization Vertical Gain 0 dB NF 0 dB A specific procedure described in the Agilent Tutorial [i.37] defines how it is possible to calculate the sensitivity and the new DANL* of this configuration of the measurement setup. The resulting noise figure is: NFsys ≈ 5,6 dB NF sys Where the total noise figure is defined based on the specification of a calculation procedure, described in [i.37]. With the above configuration it is possible to have the following sensitivity and DANL*: Sensitivity ≈ − 174 − 5,6 = −168,4 dBm / Hz = −101,4dBm / 5MHz DANL * ≈ Sensitivity + G pre = −143,4 dBm / Hz = −76,4dBm / 5MHz All the following results were obtained by measuring the signal power in three adjacent bands: the first channel [2165- 2170] MHz gives the UMTS in-band power while the eventual interference power must be measured in the second channel [2170-2175] MHz. However, since the interference level is often near the DANL* of the measurement chain, a third measurement in the upper adjacent channel [2175-2180] MHz is needed in order to validate it. This procedure is needed because of the impossibility of switching off the UMTS network. The DVB-SH network is in the meanwhile switched off. The P upper and lower values simply give the difference in dBc (dB to Carrier) of the right and left channels in comparison to the central channel ([2170-2175] MHz). 7.7.5.2 On-field Measurement Results Table 7.40 lists some measurement results related to an urban UMTS and a suburban base station in Pisa, taken at different distances from the BS. The noise power related to the measurement chain is equal to the sensitivity previously calculated: N = −101,4 dBm / 5MHz ETSI 219 ETSI TS 102 584 V1.1.2 (2010-XY) Table 7.40. UMTS field strength measurements Ch1 Ch 2 Distance Ch 3 from the 2165-2170 MHz Plower 2170-2175 MHz Pupper Meas. 2175-2180 MHz BS Code [CUMTS + N ]dBm + G pre [dBc] [ IUMTS + N ]dBm + G pre [dBc] [m] N + G pre [dBm] [dBm] [dBm] Urb_168 168 -10,65 60,89 -71,54 -4,5 -76,05 Urb_263 263 -11,52 61,42 -72,94 -3,2 -76,17 Urb_377 377 -17,01 57,31 -74,32 -2,1 -76,47 Sub_40 40 -9,25 62,15 -71,4 -4,7 -76,04 The test locations are showed in figure 7.72. Figure 7.71. Test locations for the urban UMTS BS The measurements reveal a very high BS output power. This can be due either to the location of the BS or to the number of users connected to that station, which is proportional to the traffic and therefore to the output power. In order to validate the measurement of very low signals, like interfering signals, it is recommendable to: - Check that the obtained values are above the DANL. - Check that the values obtained by cleaning measurements from the gain are above the sensitivity. - Compare the obtained interference trend with the predicted results of the network planning tool (for the urban BS only). - Compare the obtained values with the signal power of an adjacent channel where no signal is expected (Channel 3). This is because, being the interference level in that channel negligible, the displayed value can be taken as a reference lower bound (in fact it often corresponds to the DANL*). The value Pupper in Table 7.40 represents this difference. The measures here selected confirm the previous checks. Evidence of the last two points is given below. ETSI 220 ETSI TS 102 584 V1.1.2 (2010-XY) Table 7.41 summarizes the results predicted by the network planning tool relatively to the urban BS obtained by fixing the farthest measured value (377 m). Table 7.41. Prediction and measurement on the interference                     distance = 168 m distance = 263 m distance = 377 m Figure 7.72. Prediction on the interference trend Looking at figure 7.72 the values predicted by the tool for the 168m and 263m points are very close to the measured ones. It is therefore possible to conclude that the out of band measurements denote an interference trend coherent with what expected. The Pupper value can be affected by measurement artefacts and it is therefore advisable to consider mainly the measurements in which such Pupper is higher than 2 dB. This is once more the case of the quoted measurements: in this context, not only is the power significantly high, but also the Pupper denotes clearly that the power measured in channel 2 represents the interference power. Summarizing all the elements, it is possible to assert that:  the measured value denote an interference trend coherent with what predicted  all the values obtained were above the available sensitivity/DANL  the urban BS measurements present always ΔP upper > 2 dBc which decreases for increasing distance Then, the real value of the interference IUMTS can be calculated by linearly subtracting N obtaining the table 7.43 results. Table 7.42. UMTS interference calculations Distance CUMTS IUMTS ACLR Meas. from the Code BS [m] [dBm] [dBm] [dB] Urb_168 168 -35,65 -98,25 -35,65-98,25= -62,6 ETSI 221 ETSI TS 102 584 V1.1.2 (2010-XY) Urb_263 263 -36,52 -100,54 -36,52-100,54= -64 Urb_377 377 -42,01 -103,5 -42,01-103,5= -61,5 Sub_40 40 -34,25 -98,05 -38,25-98,05= -63,8 Note that the measurement results can be affected by a certain bias, linked to the stability of the used instruments. Furthermore, being interested in very low signals which are close to the thermal noise level accentuates this uncertainty, thus leading to consider few dB of approximation. Based on these observations, it can be concluded that: The urban UMTS Vodafone base station used an ACLR of approximately 60 dB. With such values some degradation of the C/(N+I) can be expected, as they are above the - 105 dBm threshold. This would mean that the measurement locations belong to the interference limited area and that the impact of the interference on the SNIR is higher than 3 dB degradation. However, the maximum distance where the measurement still give relevant results is 377 m, which is quite lower than the radius identified in Figure 1 (825 m) . The results of this on field measurement session denote that better conditions can be found (i.e ACLR of 60 dB) and that the output power of a BS can be significantly high due to the number of users and amount of traffic. 7.7.6 Conclusions and Recommendations The different trials, simulations and computations provided in the previous sections can be summarised as follows: • In the case of terrestrial deployment, the impact of DVB-SH repeaters on UMTS reception is very low, quite undetectable. • Even if no measurements have been performed, the reciprocal is also valid. In the different trails, no specific perturbations have been observed on DVB-SH reception. • The reciprocal impact of UMTS and DVB-SH networks is in the same order of magnitude that the reciprocal impact of two UMTS networks. • When DVB-SH reception is only performed through satellite (no CGC deployed), the possible presence of UMTS Base Stations can cause interference on the DVB-SH reception. In this case, the zone where the interference degrades the received satellite signal by a factor of 3 dB is called exclusion zone. • During measurement campaigns, the measured radius of the exclusion zone is around 400 m in urban areas, far lower than the predicted value of 800 m, and the probable presence of a CGC will reduce this zone drastically. • Though no measurements were reported on rural Base Station impact, one can extrapolate that the measured value will be also around half the predicted value, that is to say around 700 m instead of 1300 m. • Some results coming from J Ortigia confirm that the presence of S band signals in adjacent band degrades slightly only the DVB-SH signal received from the satellite, decreasing the ESR (5) fulfilment ratio by less than 10 %, which represents less than 0,5 dB degradation. • We recommend therefore performing careful Radio Network Planning when deploying the system, in order to minimise possible, but still limited, reciprocal interference between UMTS Networks and CGC Networks. ETSI 222 ETSI TS 102 584 V1.1.2 (2010-XY) 8 Services 8.1 General considerations From the service point of view, the following similarities between DVB-SH and its DVB precursor can be noted: • DVB-SH signal is intended to be received by a variety of mobile and portable devices; • DVB-SH offers high data rates, even in moving conditions and reuses technical elements of the DVB-H specification so that harsh conditions of terrestrial, urban mobile propagations can be addressed in the same manner; • DVB-SH is a broadcast-centric delivery system: the same content is delivered to an unlimited audience without the risk of network saturation; • DVB-SH has the flexibility of narrow-casting thanks to the support of multicast protocols; • DVB-SH is an IP-based system using MPEG2-TS as the baseline transport layer (and consequently, DVB MPE as the default encapsulation protocol). Existing DVB signalling such as PSI/SI, SFN, Time-Slicing, MPE-FEC are mostly preserved; • DVB-SH can carry DVB-IPDC in essentially the same way as DVB-H does. In this framework, it can support two-way interactive services when coupled with an appropriate interactive channel. There are however key differences from DVB-H that should be kept in mind, due to the hybrid-network nature of DVB-SH: • DVB-SH operates in frequency bands internationally allocated to Mobile Satellite Services. • DVB-SH does not require that the same MPEG2 multiplex be shared with DVB-T or DVB-S/S2 services, nor does DVB-SH requires the use of DVB scrambling at MPEG2-TS for Access Control; • although the first deployments of DVB-SH will be based on MPE, the DVB-SH framing specifications has the provisions to allow migration to the new GSE protocol; • when the 2 GHz S-band band is used, synergy with 3G telephony infrastructure would be exploited, especially in areas where such infrastructure already exists. Network planning for DVB-SH in urban areas could be similar to the 3G planning with the benefit that indoors coverage could be made essentially the same; • a DVB-SH coverage is always composed of a satellite coverage complemented by a terrestrial coverage. The services offered in these two coverages are strongly linked but not necessarily the same; • as a consequence of the above, DVB-SH services are a mix of Common services and Local services. Common services are usually those with very large audience while Local services have more fragmented audiences, possibly with geographical dependencies. A Local service package for one city/town may differ from the package for another city/town; • there are challenges for DVB-SH due to higher mobility, satellite specific propagation channels, and, in some cases, higher frequency bands. Some of these constraints are addressed somewhat differently by SH-A and SH-B architectures; • although the Common services are available in both the SC and the CGC, the service attributes may differ depending on the user location, more precisely between different reception modes: satellite-only, terrestrial- only and combined satellite-terrestrial receptions. For example, specific physical parameters may be selected so that higher user speed is possible with satellite-only reception; • DVB-SH interactive services may rely on a satellite return channel (e.g. in the 2 GHz MSS band) or on a terrestrial return channel (which could be independent from its CGC). The direct return path via satellite could ETSI 223 ETSI TS 102 584 V1.1.2 (2010-XY) be invaluable in areas without sufficient terrestrial coverage, or in case of catastrophic events leading to unavailability of terrestrial infrastructures. 8.2 Service classification DVB-SH services could be considered as just a supplement to traditional TV broadcasting. In such a narrow view, DVB-SH just extends the service consumption from the homes to other places. Taking a more forward-looking view, it is expected that new services will naturally develop, exploiting close cooperation between the broadcast infrastructure, the mobile infrastructure and the content provisioning infrastructure. The characteristics of the whole spectrum of services that can be offered with DVB-SH with the above assumption and their implications on the network, the terminal and the user behaviour are discussed below. 8.2.1 Service categories The services can be categorized using several criteria related to the way it is intended to be consumed. The following questions may be helpful in defining a particular focus for a particular DVB-SH implementation (see note). • is the service Common or Local?; • on which terminal category is the service most likely to be received? (Refer to clause 10 for terminal categories definitions and descriptions); • what are the likely use-cases? An important factor to take into account is the degree of "cooperation" that can be expected from the user. User cooperation means that the user can adapt his speed, can position/configure the terminal for maximum reception quality or accepts to refrain from certain actions when he is notified that the actual receive condition is not compatible with such actions. An example is a system that can warn the user that he is receiving under frequent signal blockages so that although the FEC is recovering the packet losses of the current service, the zapping time will be excessive due to the necessity to refill the FEC receive buffer for the new service; • another important fact to take into account is the expected speed of the terminals under use. A vehicle-mounted application will obviously be travelling at high speeds. However, hand-held devices such as mobile phones or portable media players will commonly be used by passengers in vehicles. Also, car manufacturers are increasingly supporting streaming audio from mobile handsets by Bluetooth into the car stereo system; again this must work "at speed" and the extension of this to video should also be considered; • does the service always require a display and what is the maximum size of such display? A radio service does not require the display to be switched on (or the display can be used for another application, e.g. games). If the display is switched off, the user may wish to put the terminal in his pocket (in the same way he is used to with his MP3 player). The video bit rate should be sufficient in order to avoid visible video artefacts.; • does the service require "multi-tasking"? For example, is it required to be able to watch a program while answering an incoming phone call with the same device? Another example is the ability to download a program while watching another one; • is the service time-critical? Can the content be delayed by several seconds compared to the actual time it is occurring or shown in traditional broadcasting? An example of time-critical service is a live TV program with participation of the selected audience by phone (mobile or fixed phone calls). Such services would not tolerate a latency of more than 600 ms; • are feedbacks of terminal information available? Terminal information could be: terminal Class, terminal location (at least coverage location, e.g. SC coverage or CGC cell-id), max speed, on-time periods, receive quality parameters, etc. These feedbacks could be used by the service provider or the network provider. These feedbacks could be in real-time during interactive sessions of an interactive service but they can also be periodically scheduled. NOTE: The guidance for the selection of key DVB-SH technologies and architectures according to the main service requirements is for further study. ETSI 224 ETSI TS 102 584 V1.1.2 (2010-XY) 8.3 Service attributes 8.3.1 Definition Service attributes describe the properties attached to a service as perceived by the end user in various reception conditions. This covers classical parameters found in DVB-H but also new parameters required by the DVB-SH specificities. DVB-H classical attributes: • bit rate profile (CBR, VBR, etc.); • quality of service (Bit Error Rate, Frame Error Rate, ESR5(20), etc.). DVB-SH specific attributes: • coverage (satellite, terrestrial); • zapping time; • end-to-end delay; • ease of use in various mobility situation (moving speed, static reception, various degrees of user cooperation, etc.). 8.3.2 Application The service attributes can be defined at several interfaces. Content Network chipset handset provider Provider I1 I2 I3 Figure 8.1: Application of service attributes concept • interface between content and network provider (I1): a service level agreement (SLA) is contracted between the content provider and the network operator so that the content operator requests specific quality, the level of channel and content protection, the coverage, the type of target terminals; • interface between network operator and chipset (interface I2): network operator transmits the "semantic" attached to the content using over-the-air ESG metadata in a non real time manner. The ESG from DVB-H needs to be updated with the added semantics required in DVB-SH; • interface the chipset and the handset (interface I3): only the chipset is able to know the real-time parameters such as if and when the user passes from one area covered by the satellite only to another covered by the CGC. This real time information is very "terminal-dependent" and is not be subject of the standardization process but left for implementation (the chipset may get this information by different means). By crossing the real time information with the semantics conveyed by the ESG, the handset is able to process the content and display it to the end user in an agreed manner. Different policies can be found: - the ESG display may differentiate Common from Local content. He could adopt his consumption behaviour based on this information (for a commuter between a rural and urban area, he would know where the rural content is aware, then he selects it); - for those contents that are not available everywhere, some hysteresis behaviour is possible at the boundary. It may be useful for the terminal to show the content inside the ESG as "selectable" only when its reception is stable enough; - zapping time may differ between those contents that are available via satellite (and which could necessitate a longer zapping time) and those that are available via terrestrial and should have a "terrestrial" zapping time. For the satellite contents, decision could be made depending on actual reception conditions to display immediately (typically under CGC) or to delay display (in case conditions may be less favourable, for instance under satellite-only coverage). But another policy could be to ETSI 225 ETSI TS 102 584 V1.1.2 (2010-XY) display immediately, whatever the actual quality of the stream is, in order to favour immediate perception of the user. So zapping time could be made variable in terms of content attributes and terminal conditions. Table 8.1: Service attributes at interfaces Service attributes I1 I2 I3 Bit rate profile X Quality of service X coverage (satellite, terrestrial) X X X zapping time requirements X X X end-to-end delay requirements X X X Target terminal (moving, fixed, cooperative…) X X X 8.4 Consequences of hybrid architecture on implementation of a DVB-SH service offering The hybrid nature of DVB-SH systems has consequences on the several service aspects: handover implementation, service discovery and access, Electonic Service Guide (ESG). 8.4.1 Handover issue For the Local services which are available only terrestrially, handover can be treated in the same way as in DVB-H. It should be noted that in the 2 GHz S-band, the CGC transmitters are likely to implement an SFN over large areas, at least at the level of a city coverage. In a SH-A configuration, it is most likely that SFN between SC and CGC is implemented. There is no handover in this case for the Common services. In a SH-B configuration, handover is required since the Common services must be transmitted on two different frequencies for the SC and the CGC, using (by definition) two different modulations and possibly different FEC and interleaver parameters (the same is true for SH-A whenever SFN is not used). It should be noted that there are three differences between DVB-H handover and the DVB-SH-B handover: • firstly, DVB-H handover assumes that there is only one front-end and one demodulator in the receiver. In SH- B, the TDM/OFDM configuration implies two demodulators working in parallel, one for the satellite signal and the other for the terrestrial signal (the RF front-end may be shared); • secondly, the satellite TS and the terrestrial TS are not necessarily signalled by DVB-SI as two different TS. When the soft combining property is configured, the two signals coming from the satellite and terrestrial origins are merged before the Turbo decoder and appear at this decoder output as a single TS (see clause 7.2.2.3.3 Maximal ratio combining and complementary code), despite the fact that the actual content of this TS will vary as a function of the presence or absence of the CGC signal [21]. When soft combining is not used, handover is merely switching between two TS outputs. It is recommended that even in this case the PSI/SI be kept identical between the two TS; • in DVB-H handover between terrestrial MFN cells, the receiver is usually able to estimate the power of the signal coming from different cells in order to optimize the moment when the handover is executed. In the DVB-SH-B hybrid environment, the satellite average signal power is almost constant, with occasional and almost unpredictable shadowing or fading due to terminal movement. Thus no handover prediction is possible. 8.4.2 Network and service discovery mechanisms Network discovery in DVB-SH is based on DVB PSI/SI with the introduction of new descriptors (to be detailed in a new revision of EN 301 192 [9]). Due to the possible use of SFN between the SC and its CGC, all the transport streams, whether transmitted via satellite or via terrestrial repeaters, belong the same network in the DVB sense. ETSI 226 ETSI TS 102 584 V1.1.2 (2010-XY) Independently of the use of SFN, a DVB-SH network may be divided into "regions", within each a different terrestrial frequency plan is used. A simple example of the partitioning of an SH-network into regions is given in figure 8.2. Figure 8.2: Illustration of the concept of partitioning an SH-network into regions As illustrated by the figure above, a Region should not be interpreted as a contiguous area, or as a cluster of CGC coverages in SFN mode. For example, in Region R2, the figure shows two disjoint clusters of CGC coverage in SFN mode. The Local content in these two clusters may be different while their frequency plan is identical, by definition of the term "Region". When the SC and the CGC operate in MFN for the Common content, the frequency of the terrestrial retransmission (the "hybrid frequency") may be different between Regions, as depicted in the left part of the examples. When the SC and CGC operate in SFN for the Common content, the physical parameters chosen for the two "terrestrial- only" frequencies may differ between Regions, as shown in the right part of the examples. This may be imposed by interference constraints at the border of the satellite beam (see discussion on "exclusion zones" in clause 11). This is illustrated in the figure by indicating that frequency f3 has more terrestrial capacity than frequency f2 in Region R1 (and vice versa for Region R2). The information on the validity of a frequency plan when the receiver is in a given region is signalled through a mechanism similar to the one used to signal the Cell_ID in DVB-T/H. Usually, the Common services of a particular network are managed by a single IP Platform, possibly separated from the Local services which can be managed by several different IP Platforms. The user may have to select several IP Platforms. It is desirable that, in normal operation, the need to monitor and parse SI information is limited to a minimum and that the discovery and selection of services are based on metadata signalled at the IP level. 8.5 Other considerations DVB-SH terminal is expected to enjoy a regime of free circulation and use all over Europe as it is the case with mobile phones and receivers of broadcast services. ETSI 227 ETSI TS 102 584 V1.1.2 (2010-XY) Terminals with transmit capability on satellite S-band (1 980 MHz to 2 010 MHz) may require compliance to the following ETSI standard ([i.32]). It is possible that, through roaming agreements, a Common service in one satellite coverage could be made available also as a Local service in another satellite coverage. It is highly desirable that such feature be offered in the way that looks transparent to the user. In cross-border situations it is desirable that such feature be treated as a special handover. 9 Network configurations 9.1 Considerations on network configurations Two main types of networks can be considered. On one side we have network that are designed to target portable devices (handsets, handheld devices) and on the other side, network that are designed to target vehicular reception. 9.1.1 Mobile TV network targeting portable devices The Mobile TV service to handsets are/will be mostly deployed, managed and commercialized with the support of mobile operators. The typical handset will be a cellular handsets with a 2,2" to 4" screen that will allow to display mobile TV on top of regular features such as voice telephony, SMS or Internet access. Depending on countries, these handsets will be subsidized by the mobile operator that will get revenues from the mobile TV (flat fee). Interactive services, possibly VoD and other service will transit using the existing 3G network and will provide additional revenues to the mobile operators. Another Mobile TV offer can also target non connected devices like Portable Multimedia Players. In this case the service, likely to be controlled by broadcasters, is an extension of broadcast digital TV paradigm to portable receivers essentially based on live channels reception combined with PVR functionalities. Both free to air or pay TV model would apply. In both cases, services can be provided directly from a satellite in rural areas, but a terrestrial repeater network should be deployed to provide indoor coverage in urban areas where the satellite signal is insufficient. The terrestrial repeaters being designed for a smooth integration in existing 2G or 3G cellular sites. Depending on the requested indoor coverage quality, the number of TV channels, the number of cellular sites to be equipped with a low power DVB-SH repeaters will range from 1/4 up to 1/1. The terrestrial repeaters can be fed either via satellite signal using a DVB-S2 radio interface (typically in Ku or Ka band) or via terrestrial IP network. In rural areas, satellite ensures nationwide direct reception. To improve indoor coverage, domestic gap-fillers can be used in rural areas to improve satellite reception. Repeaters in urban areas, mostly for indoor coverage, support satellite coverage. These repeaters re-transmit the nationwide channels and will therefore offer indoor coverage identical to a co-located UMTS system. To increase the system capacity in urban areas, adjacent carrier transmitters complement the satellite signal, allowing additional Local content. In each urban area, these complementary channels can be either national, regional or local ones. Finally, for handset based service, a cellular network, or a combination of broadcast type sites (high elevation towers),and cellular type repeaters, can be used to stream in unicast TV channel with limited audience TV and Video on Demand (VoD) as well as for interactivity return channel. 9.1.2 Mobile TV network targeting vehicle The Mobile TV service to vehicle are mostly deployed, managed and commercialized directly by the broadcasters or the satellite operators, possibly without the support of mobile operators. The typical terminal will be installed in a car with typically a 7" screen, hence higher data rate per channel is expected. This screen can be coupled with other services such as GPS, Video player, MP3 player, etc. Possibly, limited interactive services will be possible using a return channel directly to the satellite. The objective of the infrastructure is to optimize outdoor coverage across a territory. Services will be provided directly from a satellite in most areas but a terrestrial repeater network should be deployed to provide coverage in dense urban areas where the satellite signal could be insufficient (street canyon, underpass, etc.). As there is no need in this configuration for indoor coverage (except in tunnels), the repeater network will mainly use medium power DVB-SH repeaters that in most case will not be collocated with existing cellular infrastructure. Typically, only around 10 repeaters will be required to cover a typical urban metropolitan area. The terrestrial repeaters can be fed either via satellite signal using a DVB-S2 radio interface (typically in Ku or Ka band) or via terrestrial IP network. ETSI 228 ETSI TS 102 584 V1.1.2 (2010-XY) Satellite ensures nationwide direct reception. Suitable spatial technologies (large antennas, high-power platforms, etc.) are used to provide the required net capacity. Repeaters in dense urban areas support satellite coverage. These repeaters re-transmit at the frequency of the satellite carrier. According to frequency scheme selection, a similar complement of urban channels can be implemented. Finally, a return link from the vehicular terminal up to the satellite can be used to provide interactive service using for example the ETSI standardized GMR1and2 waveform already operational on Thuraya and AceS system. 9.2 Synchronization of Satellite and CGC for Common content 9.2.1 Introduction Basic mode of operation of DVB-SH networks rely on hybrid architecture when satellite broadcast content is retransmitted terrestrially in areas when satellite link availability is reduced. When those terrestrial repeaters are deployed, SFN operation is of great importance and is required in the following conditions: • when operating in SH-A mode, there is a need for dual SFN operations : between the different terrestrial repeaters, like in DVB-T or DVB-H networks, and between the satellite and the terrestrial repeaters, in a way called hybrid SFN operation; • when operating in SH-B mode, there is a need of synchronicity between the different terrestrial repeaters and the satellite but the requirements are not as stringent as for SH-A mode. 9.2.2 Terrestrial SFN The terrestrial SFN operation is described in TR 101 190 [i.22] "Implementation guidelines for DVB terrestrial services". Only basic principles are recalled here, and some differences between the two systems are described in more details. 9.2.2.1 Principle In a SFN, all transmitters are synchronously modulated with the same signal and radiate on the same frequency. Due to the multi-path capability of the multi-carrier transmission system (COFDM) signals from several transmitters arriving at a receiving antenna may contribute constructively to the total wanted signal. However, the limiting effect of the SFN technique is the so-called self-interference of the network. If signals from far distant transmitters are delayed more than allowed by the guard interval they behave as noise-like interfering signals rather than as wanted signals. The strength of such signals depends on the propagation conditions, which will vary with time. The self-interference of an SFN for a given transmitter spacing is reduced by selecting a large guard interval. It should be noted that the impact of delayed signals outside the guard interval may depend on receiver design. As an empirical rule, to successfully reduce self-interference to an acceptable value the guard interval time should allow a radio signal to propagate over the distance between two transmitters of the network. In order to keep the redundancy due to the guard interval down to a reasonably low value (25 %), the useful symbol length has also to be large given the transmitter spacing. On the other hand a smaller guard interval would lead to a higher number of transmitters. The SFN operation is spectral and power efficient as it uses the same frequency channel overall, and increases the coverage or the signal over noise ratio. The price to pay is an accurate time and frequency synchronization all over the coverage area. 9.2.2.2 SFN operation The same content is received from distribution network and transmitted at the same time by all the transmitters of the SFN, with the same waveform strictly and exactly. Due to the nature of the signals, SFN operation have some constraints recalled below: • frequency synchronization: if fk denotes the ideal RF position of the kth carrier, then each transmitter should broadcast this kth carrier at fk ± (Δf / 1 000); • time synchronization: the Guard Time has been introduced to cope with the different echoes in terrestrial channels, and also for SFN purpose. Thus, to avoid Guard Time wasting, a strong synchronization constraints must be put on the transmitters : a few µs difference should be the maximum (less than 10 % of the Guard Time). ETSI 229 ETSI TS 102 584 V1.1.2 (2010-XY) The different constraints impose an absolute time reference like the GPS for all transmitters, and the transmission of start signal ("top") to allow all transmission at the same time. This is performed through the Mega Frame in DVB-T/H and through the SH Frame in DVB-SH. 9.2.2.3 SHIP solution The general architecture of the SFN synchronization scheme is provided here in figure 9.1. Transit delay < 1 second DVB-SH SFN Transmitters adapters Service IP SHIP platform encapsulator inserter Contribution Distribution Network Network Figure 9.1: SFN synchronization architecture Distribution network characteristics: • heterogeneous network: satellite, terrestrial; • differential transit delay performance from one path to another. Synchronization principles: • SH Frame Information Packet inserter: insertion of a GPS based timestamp (±0,1 µs accuracy) in the SH-FRAME indicating the transmission time of the beginning of the next SH Frame; • SFN adapters in the transmitters (repeaters): buffering of incoming MPEG-TS packets and transmission of SH Frame aligned with GPS relative time stamp. The definition and specification of the SHIP is provided in EN 302 583 [1]. The complete mechanism is described in clause 7.6.1.2. 9.2.3 Synchronization of Satellite and Terrestrial repeaters for the Common Content 9.2.3.1 SH-B and SH-A/MFN cases In this configuration, content is broadcast on two separate carriers, each having dedicated modulation schemes. Two demodulators are required in the receiver; at this stage, no specific time or frequency synchronization is required. Each demodulator outputs a time de-interleaved SH Frame, one for the satellite path and one for the terrestrial path. Typical combination scheme consists in combining content prior to FEC decoding (see clause 7.3.2.3). It benefits from the FEC code combining performance and provides straightforward seamless reception. This requires time synchronization of both SH Frames, and controlled difference between symbol rates. DVB-SH framing provides TDM and OFDM SH Frames of identical period; this eases the synchronization process in the receiver, as it provides simple repetitive and predictable processing. At network level, only relative delays have to be handled. Typically, the satellite component has the longest delay: it is the combination of the transmission path (about 250 ms), plus the length of the Physical Layer interleaver (potentially up to several seconds). Thus, at terrestrial repeater level, these satellite delays must be compensated to minimize the level of buffering memory with the receiver. Residual variation delay is less than a few milliseconds, which much less that the size of the memory required for physical layer time de-interleavers. Synchronization between transmit signals (satellite and terrestrial repeaters) is performed using the SHIP information, in relation with configurable geographical-dependant information in each repeater site. In the receiver, time synchronization between both components is ensured by the SH-Frame (Signalling Field for the TDM component, and OFDM frame for the OFDM component). ETSI 230 ETSI TS 102 584 V1.1.2 (2010-XY) 9.2.3.2 SH-A case 9.2.3.2.1 Principle The global satellite coverage and the geographically distributed overlap of hybrid satellite/terrestrial zones require specific dispositions to ensure deriving maximum benefit of SFN capabilities to mobile broadcast services networks; this involves time phase and frequency shift control over the several possible paths used to transport broadcast signals to the User Equipment. e) satellite-originating broadcast signals are everywhere pregnant and hence constitute a de facto reference to which the terrestrial segment must be adjusted to ensure the signals transmitted over the different paths arriving at the receiver should be coherent in time, and frequency; f) DVB-SH is the reference ETSI standard applicable to hybrid satellite/terrestrial coverage architectures; its waveform definition [1] provides through the SH frame a "container" structure for SFN synchronization data (SHIP), adapted from the DVBT/H standards; g) Each hybrid satellite/terrestrial network requires the elaboration of customized SFN synchronization data, that are dependent on the actual specifics of the satellite orbital behaviour, which must distinguish the following cases: - geostationary orbit (GEO); - near-Geostationary orbit; h) Relative Time phase and frequency shift variations between the satellite direct path and the indirect repeated last mile path affect the architecture performances and engineering optimization (cell size, etc.) of the ground network, and must be tuned carefully. Hybrid SFN engineering considerations lead to ensure that: • time phase performances synchronization between direct and indirect path of some fraction (typically a few percents) of the Guard Interval duration • frequency synchronization performances between direct and indirect paths of some fraction (typically a few thousandths) of subcarrier spacing (the which e.g. is 1 709 Hz for a 7 MHz band and a 4 K FFT size) This requires that all equipments be synchronized onto a unique and stable clock reference and/or a unique and stable frequency reference. This is ensured by the synchronization of all terrestrial equipment (IPE/ Satellite gateway modulation stages/ Repeater including frequency Up and down converters ) onto the GPS 1 pps clock and/or 10 MHz frequency references. However, the limiting effect of the Hybrid SFN technique is the so-called self-interference of the hybrid network. If the relative delay between the signal arriving from the satellite and the terrestrial repeater are delayed more than the amount allowed by the guard interval, SFN self interference occurs. The strength of such signals depends on the propagation conditions, which will vary with time. The self-interference of a Hybrid SFN for a given transmitter is reduced by selecting a large enough guard interval, which would reduce the useful multiplex througput. It should be noted that the performance impact of signals characterised by a differential delay exceeding the guard interval may depend on receiver design and on relative signals powers. In a hybrid network, with a GEO or near GEO satellite, the hybrid SFN zone around a given transmitter is an ellipse, of which one of the focuses is located at the transmitter; This ellipse offers a limited Hybrid SFN constructive zone in the south direction (in the north hemisphere) towards satellite position. When necessary, the size of the ellipse can be increased by advancing transmitter time of emission of certain amount of time. For convenience, this amount of time can be an integer number (N) of Guard Intervals. Typically one Guard Interval advance is selected. In that case, the first bullet point of previous paragraph must be reformulated as: a. time phase performances synchronization between direct and indirect path of some fraction (typically a few percents) of the Guard Interval duration (Modulo N GI). ETSI 231 ETSI TS 102 584 V1.1.2 (2010-XY) 9.2.3.2.2 Application with near GEO With active house keeping, GEO satellite orbit should be kept within less than ±0,1 ° in both directions. The near GEO satellites may be inclined by as much as some few percents (typically up to (±) 5/6°. Though, correction procedures can be simplified for GEO satellite, we propose the analysis for the worst case, that is to say near GEO orbits. 9.2.3.2.2.1 Time delay The direct path through this satellite experiences changes in the absolute time delay involving four location parameters: i) Satellite gateway location j) Cell location in the hybrid satellite/terrestrial coverage zone k) Ground repeater location within the terrestrial cell l) UE location in the cell Maximum cell range resulting from CGC engineering results from ground network planning architecture and dimensioning choices. The following schematic displays the several path components contributing to relative delays between the direct satellite broadcast path and the indirect repeated last mile path. The distribution network for repeater content can be of different nature as for DVB-T/H networks. In the shown case, distribution network is using satellite S2.                 Figure 9.2: Overall system ETSI 232 ETSI TS 102 584 V1.1.2 (2010-XY) The satellite apparent motion cause propagation time variation on the gateway-satellite path and on the satellite-repeater path (and on satellite to any location in the cell path). Depending on the satellite orbital characteristics and coverage considered in the hypothesis, the time delay variations may range to about ±4 ms in the whole coverage. A pre-compensation of the time delay variation must be done at the Gateway location. It is ensured by comparing the locally received signal with a local reference delayed by twice the gateway to satellite reference position path duration. This pre-compensation is generally sufficient for GEO satellites. For near GEO satellites, this pre-compensation reduces significantly the time delay variations on the coverage, however a differential delay of up to about ±400 µs remains between the direct and indirect paths. These differential delay variations must be compensated for at the level of each repeater using satellite ephemeris position to reach 10 % of the guard time (around 11µs in QPSK 5 MHz 2k GI=1/4). The technique used to convey this information is out of the scope of the current standard, however data could be sent by in-band techniques like SHIP private section. In a near GEO case with 6 ° inclination, the following absolute delay variations can be observed, as shown in following figure, during 24 hours. Two repeaters at different locations are used. • DT1 : delay variation : satellite to repeater1 minus nominal value; • DT2 : delay variation : satellite to repeater 2 minus nominal value; • DTG : delay variation : satellite to Gateway minus nominal value. DT1 2500 DT2 2000 DTG DT1-2 1500 DT1-G 1000 DT2-G Delta T (µs) 500 0 -500 -1000 -1500 -2000 0 3 6 9 12 15 18 21 24 Time (h) Figure 9.3: Examples of absolute delay variations (one way). The differential following delay variations are shown in next figure. • DT1-G = DT1-DTG; • DT2-G = DT2-DTG; • DT1-2 = DT1-DT2. ETSI 233 ETSI TS 102 584 V1.1.2 (2010-XY) 800 DT1-2 600 DT1-G DT2-G 400 200 Delta T (µs) 0 -200 -400 -600 -800 -1000 0 3 6 9 12 15 18 21 24 Tim e (h) Figure 9.4: Examples of differential delay variations 9.2.3.2.2.2 Frequency shift In parallel to the time variations, the satellite apparent motion results in Doppler shifts of the frequency by about a few hundred Hz (typical value circa ±300 Hz) which also require compensation. The absolute Doppler variations are shown here below. • DF1 : Delta F satellite/repeater 1; • DF2 : Delta F satellite/repeater 2; • DFG : Delta F satellite/Gateway. ETSI 234 ETSI TS 102 584 V1.1.2 (2010-XY) 400 300 200 100 Delta F (Hz) 0 DF1 -100 DF2 DFG -200 DF1-2 DF1-G -300 DF2-G -400 0 3 6 9 12 15 18 21 24 Time (h) Figure 9.5: Examples of absolute Doppler variations (2,2 GHz on downlink) A pre-compensation of the Doppler shifts must be done at the Gateway location. It is ensured by comparing the locally received signal with a local reference. This pre-compensation is generally sufficient for GEO satellites. For near GEO satellites, this pre-compensation reduces significantly the Doppler shifts on the whole coverage, however it remains a differential delay up to about ±100 Hz. These differential delay variations must be compensated for at the level of each repeater using satellite ephemeris velocity data the same way the time compensation is achieved. 9.2.3.2.3 Architecture The satellite gateway provides signal processing for the satellite direct and indirect paths, and elaborates all SFN-related data required by the CGC elements (repeaters). The main services offered by the gateway are: • transmission of data received from the IP Encapsulator towards the User Equipment via the satellite (S-Band), and towards the terrestrial repeaters through a commercial distribution satellite (Ku/Ka-Band); • global SFN control and Time and frequency shifts corrections computation in a synchronized way between the satellite path and the repeater path. The same MPEG TS are broadcast by S-band satellite (S1) and by a commercial satellite (S2). The MPEG-TS transmitted by the gateway to S1 will be DVB-SH modulated. This DVB-SH modulated signal, after transposition by S1 to a S-band carrier frequency fo, is directly received by the User Terminals (direct path) under S1 coverage. The same MPEG-TS transmitted by the gateway to S2 will be DVB-S2 modulated in the Ku/Ka-band. S2 retransmits this signal in the Ku/Ka-band. The repeaters under the S2 coverage then demodulate the received DVB-S2 signal, and retransmit the extracted MPEG-TS using the same DVB-SH modulation scheme at the same previous S-band fo frequency: the terrestrial DVB-SH modulated signal are then be also received by the User Terminal. The SFN time and frequency synchronization processing in the gateway as illustrated by figure 9.6 consists in: • comparing actual received S band signal with an internal reference; • compensating time and frequency in the DVB-SH modulator sending to S1 ensuring perfect time and frequency compensation for the gateway location; and • delivering to the repeaters through adequate means the relevant ephemeris information (one means could be to introduce such ephemeris into SHIP private functions that is sent over the global MPEG2 TS) for correcting remaining differential time variation and frequency shift, if required, at the repeater level. ETSI 235 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 9.6: Hybrid SFN satellite gateway modulator 9.3 Signalling As presented in clause 4, different signalling information are used in a DVB-SH system: • physical layer: SHIP (all modulations), TPS (OFDM) and Signalling Field (TDM) are used; • section: PSI signalling; • link: MPE and MPE-IFEC; • higher: ESG, files, etc. 9.3.1 SHIP, TPS and SF Relevance of SHIP, TPS and SF signalling are detailed in clause 7. Procedures for signalling SHIP, in particular in complex situations, is detailed in [21]. 9.3.2 MPE and MPE-IFEC Relevance of MPE and MPE-IFEC signalling are detailed in clause 6. 9.3.3 PSI/SI case Precise specification of the PSI/SI and SHIP can be found in [21]. Precise specification of the ESG can be found in [37]. The Service Information (SI), is not protected by the MPE-FEC so SI reception quality will heavily depend on burst and error rates witnessed before MPE-IFEC decoding. The main PSI/SI needed by an IPDC/DVB-H terminal are: • NIT Network Information Table; • INT IP/MAC Notification Table; • PAT Program Association Table; • PMT Program Map Table; and • SDT Service Description Table. ETSI 236 ETSI TS 102 584 V1.1.2 (2010-XY) Table 9.1: PSI/SI tables reception performance Table Average size Size Repetition Reason for acceptable Remarks (bytes) (TP) period (ms) performance quality NIT 3 000 to 6 000 20 to 37 10 000 Static over the network Size depends on number of TS and cells INT 500 to 13 000 4 To 72 30 000 Static over TS Size depends on neighbour cell (HF and NHF) signalling policy and scenario PAT 24 to 64 1 100 Static over the TS Size depends on scenario Repetition period (SH-A/B) PMT 140 to 1 400 1 to 8 100 May be variable depending Size depends on scenario on the location (SH-A/B) and local content policy (HF and NHF) Repetition period SDT 50 to 450 3 2 000 Static over ST Size depends on scenario (HF and NHF) (SH-A/B) and local content policy (service_availability) Concerning PAT/PMT, the size of the respective tables and the repetition rate (10 times per second) are such that robustness is not a problem, even in ITS channels where TS PER of 10 % are typical. The PAT is not variable but the PMT may vary depending on the location (this is a DVB-SH particularity). In order for the terminal to capture local variations, especially when local content is inserted, a specific procedure is described within the [21]: the terminal has to buffer for SH-frame durations before deciding on the actual value of the PSI with specific rules for erasing the incorrect values. Case of partial losses that can induce incorrect conclusions are managed by integrating over several SH-frames. SDT, NIT and INT tables are made quasi-static and constant over the TS so the receiver does not have to actually receive them each time it is switched on or each time it performs handover, provided the tables are stored in the receiver. So SI tables do not depend on geographical location within a country: - This case is naturally applicable for the hybrid frequency and is similar to the DVB-H case. In this case, NIT, INT and SDT are defined to cover a full spot beam so they could be stored in the receiver and updated only when they have changed. Any change of content would then be signalled in the PMT so a receiver would immediately be aware of this and could start downloading the updated tables. Robustness will not be a problem in this case. In this case, a receiver in SI acquisition mode would likely be tune on during the acquisition time in order to optimize the reception probability. In typical ITS environment, the average TP error rate is around 10 % to 15 % but these errors are distributed over 20 % of bursts (or 80 % of burst are error free). By listening continuously to burst, the receiver will quickly acquire the table since the repetition interval is in the order of a few seconds. - This case is also applicable for non hybrid frequencies, a situation described by [21] as “content regionalization”. If the SI were variables, it would have been necessary to actually receive these tables more frequently. However, due to the design choices made in [Erreur ! Source du renvoi introuvable.], unique SI are maintained throughout the TS, wherever the terminal location. This will ease a lot the reception quality of INT, NIT and SDT. In conclusion, INT and SDT are unique within the TS whereas the NIT is unique throughout the network and will not change (except when roaming between network, case not described here). If one can assume that the receiver accesses the SI each time it is switched on and in connection with each handover, then this SI has to be receivable in a robust way without too much delay (preferably before the next burst). Due to the fact that the SI is tailor-made for the specific area, the size of the SI tables can be made highly limited in size. It is then possible to repeat the tables much more frequently than the required minimum (every 2 s form the SDT, every 10 s for NIT, every 30 s for INT). If the table sizes are small, they could be repeated, e.g. every second, and the probability of correct reception would increase dramatically thanks to the redundancy provided by the repetitions. This optimization procedure is also detailed in EN 302 304 [3], clause 9.3 and PSI/SI tailoring are detailed in the [Erreur ! Source du renvoi introuvable.]. ETSI 237 ETSI TS 102 584 V1.1.2 (2010-XY) The PSI/SI to be used for DVB-SH will most probably be quasi-static over the hybrid frequency, since all content related information is sent over IP. This information can, therefore, in principle, be stored in the receiver, which will make access much less time critical. By the redundancy of the repetition of the PS/SI tables, correct reception is guaranteed, sooner or later in all reception conditions. In cases where fast access of SI, not previously stored, is required, this can be accomplished over non-hybrid frequency by increasing the repetition rate of the NIT, INT and SDT. With a repetition rate of, for example, one second, correct reception of SI tables can be obtained within a few seconds, also in very bad channel conditions. 9.3.4 ESG case ESG are protected by MPE-IFEC, files are protected by IPDC content delivery mechanism. Specific mechanisms described in [ESGoverSHIG] manage regionalizaton of ESG and their protection by means of repetition (carrousel). 9.4 Considerations on the use of repeaters and their feeder links This clause contains guidelines and recommendations for the use of repeaters (Complementary Ground Components) in DVB-SH networks. 9.4.1 TR(a) On-channel regenerative repeaters. The TR(a) CGC on-channel regenerative repeaters are broadcast infrastructure transmitters which complement reception in areas where satellite reception is difficult especially in urban areas. They are basically designed as any broadcast transmitters. Local content insertion is possible. GPS Feeder synchronization Antenna link system Network Data Modulator and Power Termination processing RF signal Amplifier Output filter generation Local Control Unit insertion To and from Network operation Center Figure 9.7: TR(a) functional block diagram The TR(a) repeaters are composed of the main following functional blocks: • A network termination function that interfaces with at least one feeder link. This feeder link ensures the backhauling function and brings the MPEG TS flows to the repeater in such a way that the SFN synchronization remains possible. The backhauling can be of any appropriate kind providing that the Quality of Service is compatible with the overall target (typically ten times better than the repeater to subscriber QoS). This aspect of feeder link QoS is not part of these guidelines. One of the most appropriate backhauling means is via a Ku or Ka-band geo-stationary satellite transponder (DVB-S or DVB-S2) thus offering the advantage of ensuring a wide area coverage of any number of repeaters. Other backhauling means are: Optical fibre ETSI 238 ETSI TS 102 584 V1.1.2 (2010-XY) network, microwave link, high speed DSL, etc. A mix of any compatible means is possible provided that the overall latency remains compatible with the SFN alignment. The network termination function ensures the interfacing with the backhauling network at the PHY and MAC levels. For adequate network management the network termination also ensures the monitoring of the quality of the incoming signal and provides the information to the control unit. A repeater can include several types of interfaces for either ensuring redundancy between feeder links or for the insertion of local contents or both. • A data processing function that ensures the extraction of the selected bouquet(s) of MPEG TS from the backhauling flow(s). This data processing function might also be able to re-multiplex the MPEG TS in order to optimize the use of the feeder link. In particular local content can be extracted from the feeder link to generate the local multiplex. In that condition, only the envelope of all local contents is sent rather than the sum of all local multiplexes. This data processing function also achieves the SFN synchronization after extraction of the mandatory parameters of the SHIP packets. The time reference for the SFN synchronization is typically a 1 pps GPS pulse (or Galileo in the future). In case of loss of the time reference signal the TR(a) should stop transmitting (no radiated signal) or hold the transmission as long as the time shift can be guaranteed to be lower than 25 % of the guard interval and then stop transmitting. The extraction of the optional SHIP parameters is advisable. Note that the SFN synchronization as derived from the DVB-T/H assumes that the S band satellite is geo-stationary and that its residual relative movement can be compensated by the broadcast head end. • A control unit function or Operation and Maintenance function that monitors and controls the repeater and is able to be linked to a centralized network management. If not linked to such a network operation centre the control function should be able to record the key parameters in such a way that the history can be later recovered for analysis. If no link to the network operation centre is possible then the use of optional SHIP parameters could be envisaged. • A modulator function that generates the COFDM waveform followed by an up-converter or directly at RF. The RF frequency must also be synchronized by means of a GPS (Galileo) signal to ensure the correct SFN function. • A power amplifier or set of power amplifiers to amplify the carrier signal up to the desired level. The output power limit is determined by the applicable safety regulation on one hand and to the radio planning target on the other hand. • An optional output filter in order to comply with the output spectrum limits NOTE 1: For example, ETSI EN 302 574-1 in the case of the S-UMTS band. • The antenna system that radiates the RF signal. Typical TR(a) parameters: • Feeder interfaces: - L-band interface from Ku or Ka-band LNB (950 MHz to 2 150 MHz) - ASI - Ethernet (10/100 baseT or Gigabit Ethernet, optical or electrical) • Synchronization: - GPS receiver • Output power (S-UMTS case): - More than 45 dBm per carrier per cell site • Other characteristics: - MER : > 23 dB ETSI 239 ETSI TS 102 584 V1.1.2 (2010-XY) NOTE 2: The MER (Modulation Error Ratio) value includes all the signal impairments as measured at the antenna level. The indicated value assumes that there is no non-regenerative means between the repeater and the subscriber terminal or that a non-regenerative devices inserted has no significant effect on the MER value seen by the terminal. Primary supply (48 VDC/110-230 VAC) with-w/o back up Indoor cabinet or shelf or outdoor cabinet (with or GPS L band signal w/o air conditioning) Data processing unit Antenna sub-system Power amplification DVB-S/S2 Network termination unit • Reception of TS RF up conversion • Mapping of TS • SFN synchronization • COFDM modulation GigEthern • Optional DVB-H, -SH processing Ku-Ka receiving • part (dish + LNB) ASI Control unit: O&M functions (local, remote) Terrestrial network (e.g. fiber optics, microwave link , ADSL…) Local operation:  Installation, commissioning  Configuration  Maintenance and repair Operation and Maintenance Centre – Network Manager  Monitoring  Reconfiguration  Performance monitoring Head end and or contribution Figure 9.8: Example of implementation diagram of TR(a) 9.4.2 TR(b) Non-regenerative gap-fillers 9.4.2.1 Generalities The non-regenerative gap-fillers (or simply gap-fillers) are personal or public devices of limited coverage providing local on-frequency re-transmission and/or frequency conversion; typical application is indoor enhancement under satellite coverage; no local content insertion is possible. The main benefits of the gap-fillers, when compared to regenerative repeaters, are easier deployment and lower cost. The delay induced by the whole process of reception, amplification and transmission must be substantially shorter than the guard interval of the used DVB-SH mode (a typical delay is 5 µs), so that a receiver receiving both signal from a gap filler and signal from a satellite or a regenerative repeater does not have to deal with interference but with a constructive addition of signals. Two difficulties arise when using a non-regenerative gap filer is the filtering and the difficulty to ensure the remote management of the gap-fillers when the deployment and operation have to be kept simple and economical. ETSI 240 ETSI TS 102 584 V1.1.2 (2010-XY) 9.4.2.2 Filtering issues in non-regenerative gap-fillers The main obstacle in the deployment of on-channel gap-fillers is a problem inherent to its logic. The transmitted signal may be fed back to the input of the gap-fillers, thus creating a feedback-loop which generates two kinds of problems: ripple in the transfer function of the device, and, at worst, instability of the device. Decoupling:  with delay Gain : G  Figure 9.9: Illustration of an on-channel gap-filler coupling effect 20 18 Peak to peak ripple (dB) 16 14 12 10 8 6 4 2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Gain margin (dB) Figure 9.10: Peak to peak ripple as a function of gain margin The amplitude ripple value in dB is given by:  − M   1 + 10 20  R = 20 log M  , M: margin in dB = -20 logG (with G < 1)  1 − 10 − 20    There is also a group delay distortion and ripple but of negligible impact in OFDM: if the relative delay between input signal and re-injected loop-back signal is  then the time delay variation in function of the frequency f is given by: dϕ Gβ − cos 2πfτ θ =− = τ .Gβ . dω 1 + G ² β ² − 2Gβ cos 2πfτ Common values are 20logG around 70 dB and -20log greater than 80 dB. Some on-channel gap-fillers include internal echo-cancellers in their devices. This element adds an internal decoupling to the external decoupling, thus allowing for a higher total effective decoupling. Improvements of more than 15 dB have been reported. The use of echo-cancellers in on-channel non-regenerative repeaters thus brings two benefits: • the deployment of repeaters in sites where it would otherwise reveal itself unfeasible; and • a reduction of in-band distortion thus improving quality of the received signal in the area covered by the repeater. ETSI 241 ETSI TS 102 584 V1.1.2 (2010-XY) If no filtering is included in the gap-filler transfer function then other channels and also close or adjacent channels belonging to other services will be re-transmitted thus impacting the coverage of these other services. Proper site spectrum survey is thus necessary. In certain cases however this on the contrary can be advantageously used in the sharing of gap filling infrastructures between different services or applications. Typical examples are subways, tunnels and large indoor public areas (commercial malls, airports). Regarding the gap-filler itself it is also important to note that protection filtering might be required in order to protect it against strong nearby sources that could drive it into non-linear behaviour and thus non-linear distortion and inter- modulation generation. Signal from DVB-SH transmitter Strong interfering signal from other service Figure 9.11: Filtering issues in non-regenerative gap-fillers Recent gap-filler designs include efficient filtering in order to properly select the channel(s) to be retransmitted but this does not always overcome the issue of input overloading by strong out of band signal. 9.4.2.3 Frequency synchronized transposing non-regenerative gap-fillers Such gap-fillers receive a signal on frequency F1 and re-transmit it on frequency F2 without any SFN resynchronization apart from the F2 carrier frequency. Demodulation at F1 then modulation at F2 of the received signal is not recommended so as to maintain the through time latency as low as possible. The carrier frequency F2 has to be synchronized to ensure the SFN between all the transposing gap-fillers this can be efficiently done by using a GPS locked frequency reference. 9.4.3 TR(c) Mobile transmitters These transmitters are used to build "moving complementary infrastructures" on-board public transportation such as trains. Depending on waveform configuration and radio frequency planning, local insertion may be possible. Depending on the application both types of repeaters/gap-fillers can be employed for mobile coverage: regenerative and non-regenerative transmitters. In all cases it must be kept in mind that temporary interference with fixed repeaters might occur. 9.4.3.1 Regenerative TR (c) Obviously if local insertion is required then regenerative TR(a) type is needed. A typical service that could be offered e.g. on-board trains could be pre-recorded programs locally inserted. That kind of service would be similar to what is offered now on-board planes. The backhauling of the main content is the most critical issue since it requires a relatively high data rate link between the vehicle and the fixed network. However depending on the desired capacity this will be easily achievable and actually achieved for other data services (WiMAX). Because there is no need for SFN with fixed sources it is also possible in that particular application and in the case of a hybrid network to recover the signal from the S-band satellite preferably by means of a relatively high gain antenna (6 dBi to 8 dBi) on the roof of the mobile and to ETSI 242 ETSI TS 102 584 V1.1.2 (2010-XY) demodulate this signal before re-modulation on the same frequency. The re-generated signal would then be redistributed through a radiating distributed local installation at a much higher level compared to the residual signal coming direct from the satellite. S-band signal from satellite Low profile 6 – 8 dBi receiving antenna, regeneration and amplification On-board radiating network Figure 9.12: Example of possible TR(c) implementation on-board train 9.4.3.2 Non-regenerative TR(c) If no local insertion is required then a non-regenerative solution could be envisaged. However attention has to be paid in the link budget and final C/N value depending on the main receiving antenna G/T. All the other drawbacks of non regenerative on-channel solutions (see TR(b)) apply but indeed in that case a shared installation with 3G networks could be contemplated. A transposing non-regenerating solution would help in solving the isolation constraint of the on-channel gap-fillers but the frequency availability or local usage might not be possible. 10 Reference Terminals 10.1 Top-level design considerations 10.1.1 Terminal categories To address a wide range of market sectors, DVB-SH allows a large freedom in terminal implementations. Three main categories can be identified and are considered hereafter : • category 1: car-mounted terminals (also called "vehicular"); • category 2: portable TV devices with 2 sub-categories: - 2a: large screen ( 10'') portable devices, battery or mains powered; - 2b: pocketable (handheld) TV devices, mainly battery powered; • category 3: handheld terminals with an embedded cellular telecom modem (or "convergence" terminal). Car-mounted terminals can especially benefit from the nation-wide coverage and allow designers to include many of the advanced features of DVB-SH (Seamless complementary satellite/CGC coverage, SH-B configuration, high-order modulations, long time interleaver, etc.). Portable TV devices with large screen are mainly stationary during reception. They could have an attached antenna, but also detachable antenna accessories. This latter case allows for high reception performance thanks to optimization of the antenna position by the user (i.e. find a LOS reception from satellite or optimized position for good reception from CGC). ETSI 243 ETSI TS 102 584 V1.1.2 (2010-XY) Handheld terminals can especially benefit from the outdoor and indoor coverage in built up areas (similar to 3G coverage), but are more challenging due to their small form factor, the large number of functions to be integrated, limited battery power and coexistence with other active radio functions. Category 2b has common characteristics with handset. Most often, pocketable terminals embed a large number of multimedia features without necessitating coexistence with radio modems. Other features which are at the discretion of the manufacturers/markets include: • antenna dedicated and/or optimized to satellite link; • number of antennas and branches for diversity gain; • power of processor embedded in terminal; • embedded memory for physical layer processing; • host memory for PVR functions. 10.1.2 Mobility aspects 10.1.2.1 Mobile channels DVB-SH receiver mobility environments are defined in clause 11. 10.1.2.2 Antenna and diversity considerations Due to the relative narrow bandwidth of the frequency bands considered (see clause 4.1.4.1 on frequency allocations), antenna performances could be better than that of their wide-band UHF counter part. Of course, this statement is not valid in the case of multi-band antenna (antenna shared between different missions on different bands). For handheld with integrated linearly polarized antenna, -3 dBi gain can be achieved with omni-directional pattern (for example at 2,2 GHz/30 MHz bandwidth). If the antenna is external, antenna gain can be as high as 0 dBi for linear (0 dBic for circular antennas and omni-directional for both). The polarization choice depends on targeted terminal (optimization for satellite link or not). For vehicular or external dedicated antenna, 4 dBic gain can be reached at elevation optimized to target satellite orbit and position (for example, antenna for SDARS system). Further details are given in clause 10.4.5. Antenna diversity reduces the effect of the fast fading. As shown in figure 10.110.1 in a receiver exploiting antenna diversity, output signals obtained from several antennas are linearly combined using adjustable complex-value weighting factors before being decoded. Implementations may differ: • by the antenna system's characteristics: number of antennas, relative positions, orientation and characteristics of each antenna (polarization, radiation pattern, etc.); • by the algorithm used to compute and eventually iteratively adapt the weighting factors. Antennas x1 x2 x3 xN  w1*  w2*  w3*  wN* y= wi *x i DVB-SH i Decoder Figure 10.1: Antenna diversity receiver In mobile reception conditions (especially for Terminal Category 1), antenna diversity gain is expected to permit a reduction of terrestrial repeaters transmit power by up to 6 dB for the same coverage. It should also allow increasing the ETSI 244 ETSI TS 102 584 V1.1.2 (2010-XY) mobile's maximum speed for correct reception. This result is given considering a multipath Rayleigh channel and no long time interleaver benefit. Tests results do not evidence contradictory effect between diversity and interleaving gains. For satellite reception, at least 3 dB diversity gain can be achieved in LMS channels. The antenna diversity gain will increase with the mobile speed reduction. Of course, diversity with higher order (3 or 4) could be envisaged on Terminal Category 1. In indoor reception (especially for Terminals Categories 2a, 2b and 3) the mobile channel dynamic is typically slower than for outdoor mobile reception. However, considering the fading Rayleigh environment, diversity gain could be as high as 6 dB. Experimental results confirm this level of performance advantages (see Clause A.13).It is to be remarked that the measured diversity gain may be impaired by issues like gain unbalance between antennas, antenna fading correlation dependency on terminal antennas respective positions and considered environment. 10.1.3 Service aspects As for DVB-H, a trade-off has to be made between burst length, time-slicing off time period, power saving effectiveness and access time. Receivers should minimize "wake up" time (time between RF switch-on and start of demodulation) by implementing various techniques such as keeping and using formerly acquired demodulation parameters, using fine evaluation of the off time period, etc. As for DVB-H, time slicing allows seamless handover by scanning adjacent cells signals during idle time periods. Another aspect in DVB-SH is the receiver Class (memory, see also clause 10.2) and its associated time interleaver depth capability. Issues to be considered are: • backward compatibility of receiver Class 1 with long interleaver used by the transmitter; • time slicing burst length / off time period; • access (or zapping) time. Coexistence (especially for Category 3 receivers) For handheld receivers with an embedded cellular telecom modem, DVB-SH reception should not prevent or hinder cellular operations (initial synchronization, roaming, handover, communication). Reciprocally, reception of DVB-SH service should be possible in parallel with cellular operations. However, DVB-SH quality may be degraded by cellular uplink signal (see Clause B). Two receiver states may be considered: • cellular function in standby, i.e. intermittent uplink transmissions (cell search, handover in mobile conditions): DVB-SH reception if active should provide nominal picture quality; and • cellular function in active communication, i.e. uplink traffic transmission: DVB-SH reception if active may or may not be displayed, depending on user selection. When DVB-SH service display is switched off during call, demodulation should continue in the background and display should be automatically resumed at the end of the cellular call. Coexistence conditions described in Clause B consider a worst case: • DVB-SH full service while cellular handheld operation: standby and communication; • low level DVB-SH downlink signal; • maximum power cellular uplink; • no DVB-SH time interleaver protection. 10.2 Memory requirements for DVB-SH processing As shown in figure 10.2 the DVB-SH receiver baseband processing shows a similar high-level architecture as that of the DVB-H counterpart, namely analog-to-digital converter, OFDM demodulator, de-interleaver, Forward Error Correction (FEC), demultiplexing, multi protocol de-capsulation, MPE-FEC, IP filter and terminal host interfacing. New functions are mainly the Turbo decoder replacing the convolutional one of DVB-H and the physical layer time de-interleaver. ETSI 245 ETSI TS 102 584 V1.1.2 (2010-XY) Analog RF Turbo decoder A/D OFDM de-interleaver 4 Mbits Memory for processing DVB-SH receiver 4 Mbits Memory for de-interleaver MPEG-TS IP filter & host interface MPE-FEC MPE de-multiplexing Figure 10.2 The Turbo decoder needs extra hardware which is available as an off-the-shelf building block from 3G technology development. The main challenge for implementing the Physical Layer Interleaver is its memory requirement which is dependent to the DVB-SH Receiver Class: Class 1 Receivers Class 1 are required to handle an interleaver profile comprising one full SH-frame with 816 CU. Using the convolutional interleaver approach which typically halves the amount of memory needed, this transforms to: • interleaver lengths of up to 240 ms (QPSK, uniform) and 120 ms (16QAM, uniform); • capability to store and process up to 408 CU or 6 528 IU (can be handled with approx. 4 Mbits of memory). Since support of DVB-H is likely for DVB-SH receivers, the memory dedicated to the DVB-H MPE-FEC could be allocated to the de-interleaver. In this case, the DVB-H MPE-FEC, if ever transmitted, is not processed. The choice between MPE-FEC processing or interleaving processing should be managed at system level. Therefore, DVB-SH Class 1 receivers could be realized with the same amount of memory than previous DVB-H receivers. Class 2 Receivers Class 2 are required to handle an interleaver profile comprising 64 full SH-frames with 52 224 CU. Using the convolutional interleaver approach which typically halves the amount of memory needed, this transforms to: • interleaver lengths of up to 30 s (QPSK, non-uniform) and 15 s (16QAM, non-uniform); • capability to store and process up to 26 112 CU or 417 792 IU (can be handled with 256 Mbits memory). 10.3 DVB-SH reference receiver model 10.3.1 Reference model The receiver performance is defined according to the reference model shown in figure 10.3. ETSI 246 ETSI TS 102 584 V1.1.2 (2010-XY) Field Strength Antenna E Gain Noise DVB-SH IP Decapsulator IP-Out Figure F Extended Front end & DVB-SH Time MPE- IP-De- MPE- Input tuner Demodulator Slicing FEC encapsulation FEC Power P TS- RF-Reference IP-Reference point Reference point point FER- MFER- Reference Reference point Point Figure 10.3: Reference model Reference points are defined for: • RF; • transport stream; • frame errors before MPE-FEC; • frame errors after MPE-FEC and extended MPE-FEC; • IP-stream. All the RF receiver performance figures are specified at the RF-reference point, which is the input of the receiver. 10.3.2 Receiver for Vehicular terminals Depending on terminal categories, various antenna and front-end solutions will be embedded in receiver. For terminal category 1 described in figure 10.4, according to vehicular reception constraints (clause 4.2.6.1), antenna with optimized diagram pattern and polarization will be used. Additional Low Noise Amplifier will be connected directly to antenna to optimize noise figure and sensitivity (see details in clause 10.4). Field Strength Antenna E Gain Ga DVB-SH tuner Demodulator Pin LNA RF NF Filter RF-Reference point Figure 10.4: Radio receiver architecture for Terminal category 1 10.3.3 Receiver for Terminals with telecom modem For category 3 terminals shown in figure 10.5, if the terminal includes telecom RF modems, a filter is mandatory between the antenna and the (optional) LNA and/or the tuner to protect the latter elements from telecom modem signal interference. However, this filter in front of the LNA degrades the DVB-SH receiver sensitivity. ETSI 247 ETSI TS 102 584 V1.1.2 (2010-XY) Field Strength Antenna E Gain Ga DVB-SH tuner Demodulator Pin RF LNA NF Filter optional RF-Reference point Figure 10.5: Radio receiver architecture for Terminal category 3 Basic receiver architectures for all terminal categories will be derived from both architectures presented above. For SH-A system architecture, these synopsis will be used without any variants. For SH-B system architecture, the tuner will have two outputs each of them will correspond to TDM and OFDM signals respectively. These outputs are generated after down conversion into the tuner from two different RF frequencies, one is provided by satellite(s), the other by CGC segment. Demodulator for SH-B will have two inputs in accordance to figure 4.7 of clause 4.2.6. For antenna diversity implementation, several receiver branches will be added to the single one. After diversity combining, transport stream will be provided to an unique input of DVB-SH IP decapsulator. 10.4 Minimum signal input levels for planning 10.4.1 Noise floor for vehicular receiver The receiver should have a noise figure better than 2 dB at the reference point, at sensitivity level of each DVB-SH mode. This corresponds to the following noise floor power levels: • Pn = -103,2 dBm [for 8 MHz OFDM channels, BW = 7,61 MHz]; • Pn = -103,7 dBm [for 7 MHz OFDM channels, BW = 6,66 MHz]; • Pn = -104,4 dBm [for 6 MHz OFDM channels, BW = 5,71 MHz]; • Pn = -105,2 dBm [for 5 MHz OFDM channels, BW = 4,76 MHz]; • Pn = -110,2 dBm [for 1,7 MHz OFDM channels, BW = 1,52 MHz]. 10.4.2 Noise floor for handheld receiver The receiver should have a noise figure better than 4,5 dB at the reference point, at sensitivity level of each DVB-SH mode This corresponds to the following noise floor power levels: • Pn = -100,7 dBm [for 8 MHz OFDM channels, BW = 7,61 MHz]; • Pn = -101,2 dBm [for 7 MHz OFDM channels, BW = 6,66 MHz]; • Pn = -101,9 dBm [for 6 MHz OFDM channels, BW = 5,71 MHz]; • Pn = -102,7 dBm [for 5 MHz OFDM channels, BW = 4,76 MHz]; • Pn = -107,7 dBm [for 1,7 MHz channels, BW = 1,52 MHz]. 10.4.3 Minimum C/N requirements This clause provides in table 10.1 typical and possible DVB-SH receiver implementation losses. For AWGN minimum C/N requirement determination, it is recommended to add the implementation losses reported in table 10.1 to the C/N values given in clause 7. ETSI 248 ETSI TS 102 584 V1.1.2 (2010-XY) Table 10.1: AWGN Implementation Losses Modulation Implementation losses (dB) Implementation losses (dB) Typical Receiver Possible Receiver TDM- QPSK 0,5 0,5 TDM- 8PSK 1,0 0,75 TDM- 16APSK 1,5 1,0 OFDM- QPSK 1,1 0,5 OFDM- 16QAM 1,5 1,0 The implementation losses are defined with respect to an ideal receiver with perfect channel estimator. The implementation losses for typical receivers constitute an upper bound based on early receivers. The possible receiver implements more sophisticated channel estimation algorithms, reducing implementation losses. • Under TU6 channel conditions, the implementation losses for typical receivers are 1,6 dB for QPSK and 2 dB for 16-QAM. • Under TU6 channel conditions, the implementation losses for possible receivers are 1 dB for QPSK and 1 dB for 16-QAM. • Under Rice and Rayleigh terrestrial channel conditions, a provision of 2 dB for implementation loss is recommended. Nevertheless, in all TU 6 computations of Chapter 11, measured C/N values are used. 10.4.4 Minimum input levels 10.4.4.1 Sensitivity for vehicular receiver At RF-reference point, the receiver should observe reference criterion for a wanted signal level greater than Pn. • Pn = -103,2 dBm + C/N [for OFDM 8 MHz]; • Pn = -103,7 dBm + C/N [for OFDM 7 MHz]; • Pn = -104,4 dBm + C/N [for OFDM 6 MHz]; • Pn = -105,2 dBm + C/N [for OFDM 5 MHz]; • Pn = -110,2 dBm + C/N [for OFDM 1,7 MHz]. where C/N is specified in clause 7.2.2.6 and is depending on the channel conditions and DVB-SH modes. 10.4.4.2 Sensitivity for handheld receiver At RF-reference point, the receiver should observe reference criterion for a wanted signal level greater than Pn. • Pn = -100,7 dBm + C/N [for OFDM 8 MHz]; • Pn = -101,2 dBm + C/N [for OFDM 7 MHz]; • Pn = -101,9 dBm + C/N [for OFDM 6 MHz]; • Pn = -102,7 dBm + C/N [for OFDM 5 MHz]; • Pn = -107,7 dBm + C/N [for OFDM 1,7 MHz]. where C/N is specified in clause 7.2.2.6 and is depending on the channel conditions and DVB-SH modes. 10.4.5 Antenna considerations Receiver antennas must have good performance in various environments, receiving signal from satellite or/and CGCs. In this clause, intrinsic and integration characteristics are highlighted depending on the targeted terminal. ETSI 249 ETSI TS 102 584 V1.1.2 (2010-XY) 10.4.5.1 Antenna for terminal category 1 Antenna for vehicular terminals can be listed in two categories: • antenna with circular polarization optimized for satellite reception; and • antenna with linear polarization optimized for terrestrial reception. Antennas with circular polarization for vehicular terminals have been developed for several systems like Satellite Digital Audio Radio System ( SDARS) in US (XM Radio, Sirius). These antennas have relatively good circularly polarized gain at high elevation angle and acceptable linearly polarized gain at the horizon. They efficiently receive both satellite and terrestrial signals. Antenna gain in Left or Right Hand Circular Polarization (LHCP/RHCP) can be +2 up to +8 dBic, depending on angle aperture and directivity required. Antenna gain in linear vertical polarization is less than -1 dBi at horizon. These antennas are designed in various technologies: small ground plane dependent patch etched on ceramics antennas, ground independent mast antennas such as quadrifilars, turnstile consisting in two 90° crossed dipoles able to generate circular polarization. Voltage Standing Wave Ratio (VSWR) can be less than 1,5 on 50 Ω load for narrow band operation, this performance maximize antenna efficiency. A Low Noise Amplifier (LNA) with a noise figure less than 1 dB can be connected directly at the antenna output to improve sensitivity. G/T of more than -21 dB/°K can be achieved. For antennas with linear polarization, the practical standard for car reception is λ/4 monopole which use the metallic roof as the ground plane. Planar structures may be also used. Due to the use of the metallic roof, antenna pattern (and consequently the gain) is very dependent of the position of the antenna on the vehicle. Considering the 2 GHz MSS band, gain on the whole demi-sphere in linear vertical polarization is more than 0 dBi. Due to depolarization losses, the gain in RHCP is less than -3 dBic. Other constraints have to be considered: Number of radio systems integrated in vehicle (Telecom modem car kit, GPS..), larger antenna structures can be aesthetically displeasing when mounted on roof or glasses… Diversity features that increase the antenna number can lead to use a mix of technologies described above. 10.4.5.2 Antenna for terminal category 2a This clause deals with antennas for Portable digital TV sets. These terminals are defined to be equipped with screen size larger than 25 cm (more than 10" diagonal), stationary during reception and antenna attached to the receiver. Integration of high performance antennas with circular polarization and high gain is now challenging due to the flat design of the terminal. In this case the manufacturers have to make compromise between design and performances. The performances of this terminal could be between the one of category 1 and category 2b, 3. If performances are favoured, by using for example a detachable antenna accessory, the terminal G/T could be as high as to -25 dB/°K (position adjusted by the user in line of sight of the source). Diversity will be implemented advantageously with regard to mechanical size and potential distance between antennas. 10.4.5.3 Antenna for terminal category 2b and 3 This clause deals with antennas for pocketable digital TV receivers and handhelds with embedded Telecom modem(s), both being battery operated. Two solutions are possible, one with an external antenna, the other with internal antenna. The majority of handset makers design terminals with internal antennas. A first consideration concerning antenna design is volume availability. Also, coexistence with a lot of components decreases expected gain due to dielectric and magnetic losses. This is why, low bulk antennas have less efficiency and more susceptibility to detuning. It is not conceivable to use multi band antenna that provide poor performances and coupling. Narrow band antenna will be considered as the basic antenna reference. For handheld 2b and 3 categories, better reception sensitivity will be reached with external antenna. Antennas for handhelds can also be ordered in two categories: • antennas with circular polarization optimized for satellite reception; and • antennas with linear polarization optimized for terrestrial reception. ETSI 250 ETSI TS 102 584 V1.1.2 (2010-XY) Circular polarization: the antenna pattern could be optimized for some pre-defined elevations, the main drawback is the position given by the user to the terminal. This leads to favour omni-directional pattern design rather than directional ones (at least in a hemisphere). The same assumptions have to be done with antenna with linear polarization. Typical figures are given below for 2 GHz MSS band and in LOS measurement conditions: • external RHCP antenna: gain 0 dBic (circular), gain in linear –3 dBi; • internal RHCP antenna: gain -3 dBic (circular), gain in linear –6 dBi; • external linear vertical polarized antenna: gain in linear 0 dBi, gain in circular –3 dBic; • internal linear vertical polarized antenna: gain in linear -3 dBi, gain in circular –6 dBic; • external RHCP antenna removable and adjustable by the user: gain +3 dBic, gain linear 0 dBi (antenna with directivity). Antenna behaviour in multipath environment is more complex due to depolarization effect and integration of power in a sphere. 10.4.5.4 Antenna diversity considerations Coming back on antenna diversity, different measurements on antenna diversity gain have been performed over the past years in various channels conditions and in S band ( 2170-2200 MHz) ([i.33] and [i.40]). These references have been reporting the gain obtained using dual antenna . It is thus recommended to apply the following values • 3,5 dB in reception conditions A, C and D • 5,5 dB in reception conditions B1 and B2.(I 33) • 3 dB under satellite reception conditions., and more generally 10 ⋅ log10 ( N ) , where N is the number of antennas. 10.4.6 Maximum Input Power for Wanted and Unwanted Signals The maximum allowed input level on the RF-reference point will depend on the antenna characteristics and RF front end architecture chosen (see clauses 10.3.2 and 10.3.3). Therefore, maximum value will differ with terminal categories. For terminal category 3: The maximum total average power from the wanted and unwanted signals should be less than +15 dBm. This value corresponds to a decoupling between DVB-SH antenna and modem antenna of more than 15 dB to 18 dB depending of the Telecom frequency band considered. For terminal category 1 and 2a,2b: The maximum total average power from the wanted and unwanted signals should be less than –25 dBm. This value corresponds to mobile phone up link emissions close to the DVB-SH terminal at more than 1 meter from the source. Down link signal providing by base stations emissions will be below this value. The assumptions considering interoperability with other radio systems, are given in Clause B. In particular, real interferer environments of DVB-SH receiver will be highlighted including interfering signal patterns (with number of carriers, level and frequency band). In this Clause, recommendations will be done for receiver linearity and selectivity. 10.4.7 G/T considerations Table 10.2 gives typical figures of basic receivers (without diversity consideration) related to the terminal categories. G/T calculations are based on various assumptions and the boundaries between classes of terminals are not so clear-cut. Active antenna corresponds to an antenna with an attached LNA. Polarization losses are defined in the respective satellite and terrestrial budget links. Assumptions: • vehicular receiver (category 1): NF=2 dB (LNA 0,8 dB, filter 2,5 dB, tuner 3 dB) • handheld receiver (category 2b and 3): NF=4,5 dB (filter 1,5 dB, tuner 3 dB). ETSI 251 ETSI TS 102 584 V1.1.2 (2010-XY) Table 10.2: Receiver typical G/T versus terminal category Usage Handheld Handheld Portable Vehicular Classification category 3 category 2b category 2a category 1 Antenna polarization L or C L or C L or C C Tuner DVB-SH DVB-SH DVB-SH DVB-SH Antenna gain (dBi) or (dBic) -3 0 2 4 Total Gain (dB) -3 0 2 4 Antenna Temperature (K) 290 290 200 150 Active antenna No No Yes Yes Noise figure (dB) max 4,5 4,5 3 2 Total noise temperature (° K) 817 817 489 320 G/T (dB/K) -32,1 -29,1 -24,9 -21,0 11 Network planning DVB-SH system is a hybrid system with a satellite and a terrestrial components. Terrestrial component network planning technique have been developed over the past 50 years for two distinct services • Classical TV broadcast in UHF, VHF, from for high towers, using high power transmitter. • Cellular, mobile system from GSM/DCS to UMTS and more recently from WiMAX technology. Although based on the same physics, practice differ in some extent to the very different markets they address: • High power for the first one, low/medium power for the second. • Roof top reception for the first one, indoor handset service for the second one. It is not the objective of this clause to propose an integrated framework, but to expose the main rules of each world for the network planning exercise. On the other hand the satellite world, also existing for 50 years, has obviously specificities: line of sight, shadowing, high path loss, and uses Land Mobile Satellite propagation models, for availability computation. The present document is presenting an overview of different methods of radio network planning and using different dimensioning parameters which values are compiled from different sources and experiences. The proposed values can not be seen as absolute references, especially on the cellular network approach, as the used parameters values can vary from a source to another, from a country to another. That is why a range of typical parameters values is proposed, and some examples provided for network planning 11.1 Introduction to DVB-SH network planning As DVB-SH system is a hybrid system with a satellite and a terrestrial component, the coverage considered in the DVB-SH network planning can be split in three areas: • Terrestrial only coverage: defined as the area served only by the complementary ground component. • Satellite only coverage: defined as the area served only by the satellite component. • Hybrid coverage: defined as the area where both the satellite and the terrestrial signals can be received simultaneously and potentially combined. Concerning terrestrial coverage, there are typically two possible methods to perform radio network planning: • The first one is a typical broadcast approach based on the computation of the minimum field strength requirements, for different classes of terminals and different propagation channels. The broadcaster method has been developed for the use of high transmitting towers with very high power, and aiming at roof top coverage rather than street level receivers. ETSI 252 ETSI TS 102 584 V1.1.2 (2010-XY) • The second one is based on cellular network radio planning tools, and is using the minimum received power level as planning criterion. Most of these tools uses only this method, and allow network planners to determine (given a certain transmitting power per site) the number of sites necessary to cover a city with a given quality of service. The cellular network method has been developed for lower tower and also lowers to medium power to cover dense cities at street level receivers, and targeting indoor coverage. This clause will give also in some examples the link budgets with the required transmitter power in some typical examples. The link between the two methods is insured through the formula below with E representing the field strength and Pimin the minimum received power level, and Ga the antenna gain in dBi. E [ dBV/m ] = Ps min [ dB(mW) ] − Ga [ dB] + 77, 2 + 20 log10 f [ MHz ] One of the objectives of this clause is to compile long time practices in network planning domain applicable to DVB-SH network planning, that is to say review parameters and their possible value ranges enabling both type of network planning. These approaches can also be extended for hybrid coverage considering the satellite as an additional repeater that allows a) reduce the number or the power of terrestrial sites or increase the quality of the service in the hybrid coverage area. For satellite coverage, the number of transmitters is dependent on the number of satellites serving the satellite coverage area. While in the terrestrial case we have a cluster of cells serving a geographically limited area, for the satellite the coverage can be a continent-size area served with several linguistic or regional spots. The objective of the satellite network planning is to assess the required satellite power given a certain coverage area and a certain quality of service for a given data rate. This exercise is described in the clause "Satellite Link Budgets", but it is done in a different way as in terrestrial network planning. • in terrestrial links, given the coverage probability in some conditions, the required shadowing margins are computed and allow to dimension the transmitter; • in satellite links, given the satellite EIRP in some conditions, the obtained link margins are used in simulations to compute the availability in the given conditions. Thus, by reverse engineering, given a an availability, required satellite EIRP can be estimated. Therefore, after this introduction, the clause comprises the following clauses: • definition of reception condition; • coverage definition, following a parallel track for the three approaches; • network planning factors. Then, following also a parallel track: • network planning methods and examples based on minimum filed strength; • network planning methods and examples based on minimum received signal level; • satellite network planning; and • for next release: hybrid network planning. 11.2 DVB-SH reception conditions 11.2.1 Introduction Reception conditions depend on the environment, the mobility conditions and the kind of terminal. The environments relevant for DVB-SH system are defined in clauses 4.2.1 to 4.2.3 and Clause A.7 by propagation channel models. DVB-SH terminal categories are defined in clause 10. Furthermore, DVB-SH introduces two classes of receivers with different capability of processing time diversity elements in the transmitted waveform (see clauses 6.5, 7.3.3, and 10.2). Concerning the mobility conditions, two main kinds of reception can be defined: ETSI 253 ETSI TS 102 584 V1.1.2 (2010-XY) • Pedestrian speed reception where the channel conditions exhibit relatively low levels of fast fading (from movement of nearby objects such as trucks) and slow deep fades due to slow movements of the antenna and nearby blocking objects (≤ 3 kmph). The portable terminal with external or integrated antenna is used indoors or outdoors at a mean height of 1,5 m above ground. • Mobile reception applies to the use of Car-mounted Terminals (Terminal Category 1) with speeds higher than 3 kmph. It is assumed that the receiving antenna is external and at a minimum height above ground level of 1,5 m. Terminal Category 2b and 3 used for direct reception within a car or train or any kind of vehicle could also be considered as a case of mobile reception. It can be called in vehicle reception. 11.2.2 DVB SH reception conditions The five cases of reception conditions defined in DVB-H, called Class A, B1, B2, C and D (see EN 302 304 [3] and [19]), fit also the DVB SH terrestrial variety of contexts. An additional dimension can be introduced if we consider also the DVB-SH environments defined in clause 4: rural, urban, and suburban. These classifications can also be used for satellite or hybrid coverage, although not all reception cases are applicable in all environments due to the intrinsic power limitation of the satellite link. Conversely, some reception conditions in certain environments are only applicable to satellite coverage, due to the geographical limitation of the terrestrial coverage. An additional case can also be defined in DVB-SH; this case refers to the satellite indoor coverage achieved due to a domestic gap filler (defined as TR (b) in the system specifications) that receives and amplifies the satellite signal. Note that this partition can apply equally to SH-A and SH-B system architectures. In all scenarios, and in the rest of the document, receiver is placed at 1,5 m above ground. ETSI 254 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.1 Reception Situation Characteristic Environment Coverage Channel Typical channel parameter Condition s Characteristics Relevant DVB-SH parameters Reception Outdoor Up to 3 kmph Rural Satellite Stationary : Low LOS :AWGN/Rice K>10 : condition A pedestrian delay/low spread Additional margin to cope with fading For shadowed, Rice below 7 dB Time interleaving to mitigate effects Low speed: large LMS channel model at low signal variation speed. Time interleaving Urban Terrestrial Stationary : TU6 channel? /low code rate Rayleigh/Very low improves. Antenna diversity Doppler also improves Low speed / Rayleigh Higher margins to cope with /low Doppler slow fading effects Suburban Terrestrial, For terrestrial same as For terrestrial same as Hybrid above above No hybrid channel No hybrid channel model model available available Reception Light-indoor Up to 3 kmph, Rural See condition lightly shielded note (1) B1 building Urban Terrestrial Channel is the same TU6 channel? /low code rate as Reception A with improves. Antenna diversity high penetration also greatly improves margins Suburban Terrestrial, Same as above for Hybrid terrestrial No hybrid channel model available Reception Deep- Up to 3 kmph, Rural See condition indoor highly shielded note (1) B2 building Urban Terrestrial Channel is the same TU6 channel? /low code rate as Reception A with improves. Antenna diversity higher penetration also greatly improves margins as in B1 Suburban Terrestrial Same as above (lower Same as above margins) Reception Mobile Up to Rural satellite Large signal strength LMS channel model at condition (vehicle) 200 kmph variation depending medium/high speeds for C with roof- on environment different environments top antenna Urban Terrestrial  Channel models like TU6  cover this scenario at least Delay spread for low or medium power depends mainly on repeaters. network Critical SFN scenarios characteristics require channel models with higher delay spread. Suburban Terrestrial, For terrestrial same as For terrestrial same as Hybrid above above No hybrid channel No hybrid channel model model available available Reception Mobile Up to Rural See condition (portable) 130 kmph note (2) D in-car Urban Terrestrial Multiple Rayleigh Channel models like TU6 fading paths cover this scenario at least Delay spread for low or medium power depends mainly on repeaters. network Critical SFN scenarios characteristics require channel models with higher delay spread. Suburban Terrestrial Same as above Same as above ETSI 255 ETSI TS 102 584 V1.1.2 (2010-XY) NOTE: According to clause 10 classification, terminals used for the different reception conditions are the following: reception conditions A, B1, B2 and D are associated to terminal categories 2a, 2b and 3 (handset, handheld and portable): (1) for Reception conditions B1 and B2 in the rural environment, under satellite coverage, it is assumed that the satellite signal is assisted by terrestrial repeater (TR (b)). The link budget applies to these TR(b), not to the end- user terminal; reception condition C is associated to terminal category 1 (vehicular): (2) for Reception condition D in the rural environment, under satellite coverage, it is assumed that the satellite signal is assisted by terrestrial repeater (TR (c)). The link budget applies to these TR(c), not to the end-user terminal. 11.3 Coverage definition In general, coverage is defined by a reference area and a percentage of that area where the signal is received with a given required quality of service a given percentage of time. Consideration is also given to the edge of the signal area (whether satellite beam, or terrestrial transmit signal). The edge is taken as the contour beyond which the signal degrades (either from leaving the satellite beam footprint, or moving too far from the terrestrial transmitter) below the point required to deliver the given QoS level. DVB-SH coverage is made up of three different components: satellite only coverage, terrestrial only coverage, and hybrid coverage. The coverage area definition is differing, first between satellite and terrestrial coverage, and second between the broadcaster and cellular network approaches. 11.3.1 Broadcaster approach In defining the coverage area for each reception condition, a three level approach is taken (see EN 302 304 [3]). Receiving location coverage: the smallest unit is a receiving location with dimensions of about 0,5 m. In the case of portable antenna reception, it is assumed that optimal receiving conditions will be found by moving the antenna or moving the handheld terminal within 0,5 m in any direction. Such a location is regarded as covered if the required carrier-to-noise and carrier-to-interference values are achieved for 99 % of the time. Small area coverage: the second level is a "small area" (typically 100 m × 100 m). Coverage area: the third level is the coverage area. The coverage area of a transmitter, or a group of transmitters, is made up of the sum of the individual small areas in which a given class of coverage is achieved. 11.3.2 Cellular approach Cell coverage: rather than using the three level approach coverage area is defined as the aggregation of several circular (approximately) cells defined by their radius. The transmitter is in the centre of each cell and can use omnidirectional antenna or trisector or any other type of antennas. Cell radius: when using cellular network methods, coverage are is defined as a cell of radius R, defined, as the extreme points where the link budgets is satisfied to meet the required level of signal. Cell edge: is defined at distance R of the transmitter. 11.3.3 Satellite coverage For satellite-only coverage: the basic covered area is the satellite antenna footprint, typically tailored to a satellite spot beam. However, not all points in the satellite coverage area are reached with the same quality. Depending on the service definition, service coverage will have to take into account environments such as shadowing, which can be caused by trees, tall buildings, or even low buildings in the case of low satellite elevation angle. 11.3.4 Quality of coverage The coverage is characterized by a certain quality of service characterized by a quality criterion. For each propagation channel (Gaussian, Rice, TU6, etc.) and each modulation and coding state, this criterion is satisfied by a minimum C/N value, which is used in network planning.(called network planning value). ETSI 256 ETSI TS 102 584 V1.1.2 (2010-XY) The cell coverage probability is defined as the percentage of locations inside the cell where the criterion is satisfied or where C/N is above the minimum corresponding value. Under terrestrial coverage (TU6), Clause A.12 utilizes two quality criteria: • FER 5 %, equivalent to MFER 5 %, which is used in DVB-H: 5 % of frames in error; and • ESR5; defined in Clause A.8: 1 s maximum in error every 20 s, which is more demanding and a fulfilment criterion defined by ESR5(20). As shown in Clause A.12, less than 1 dB in C/N separates the two criteria. In the computations associated to TU6, an ESR5(20) fulfilment criterion is chosen. For satellite channel, ESR5(20) fulfilment criterion is chosen. BMCO [19] defines the different qualities of service corresponding to different coverage probabilities, and depending on the different reception conditions. They are defined as "good" and "acceptable": • the terrestrial coverage area is declared "good": - at least 99 % of receiving locations within the area are covered for reception conditions C and D; - at least 95 % of receiving areas at the edge of the area are covered for reception conditions A and B; • the terrestrial coverage area is declared "acceptable": - at least 90 % of receiving locations within the area are covered for reception condition C and D; - at least 70 % of the receiving locations at the edge of the area are covered for reception conditions A and B. Table 11.2: Coverage probability for terrestrial coverage (broadcast method) "Good" "Acceptable" Reception Condition A 95 % edge 70 % edge Reception Condition B 95 % edge 70 % edge Reception Condition C 99 % overall 90 % overall Reception Condition D 99 % overall 90 % overall In the cellular coverage, there is no specific qualification of coverage quality or required percentage of coverage (cell or edge of cell)With given percentage of edge or overall coverage. There are some coverage obligation under the regulation process. Usually, coverage probability can be defined on the whole cell, or at the edge of cell coverage. There is no specific rule. The most usual qualities of coverage are 90 % and 95 % overall coverage for all reception conditions. Hence, other coverage probability can be used upon request of operator. Under satellite coverage: under satellite coverage, there are no typical values of coverage probability. In the satellite clause, the obtained coverage probability are provided in the different examples. 11.4 Network planning factors 11.4.1 Introduction The network planning factors constitute the ensemble of parameters that are needed in the different network planning computation and tools; As they are common to the different methods, they are described in this clause. One can distinguish three categories of criteria: • channel dependant factor: relative to the shadowing and in building penetration margins for instance; • technology dependant factor : bandwidth, noise factor, antenna gain, etc.; and ETSI 257 ETSI TS 102 584 V1.1.2 (2010-XY) • network architecture factor or implementation optional parameter: SFN topology gain, antenna diversity gain. 11.4.2 Channel dependant factor The first factor is the in building or in vehicle entry or penetration losses, applicable to reception conditions B and D. Depending on the different terrestrial planning methods, the range of values is somehow different, and so the taken values are give in each corresponding clause. The other factor is called location correction factor, or shadowing margins or fading margins in the satellite links. Correction location factor is defined by the product: Cl = µ*σ in EN 302 304 [3] and [19] where: µ is called the distribution factor, and σ is the standard deviation of the distribution. of the signal variation. The possible values are provided in clause 11.5.3.1. The term shadowing margins is used currently in the cellular network approach. The most commonly used propagation model is Log normal attenuation with the standard deviation σ. Considering the distance to transmitter d and R the cell radius, probability that signal exceeds a defined threshold at distance d in an area dA (elementary area) is given by:    d    1  −Cl + 10n log  R    Pe = 1 − erf      2  2σ        where: Cl is the shadowing margin associated with the coverage probability and σ is the shadowing standard deviation. So, when being at edge of cell, d = R and the equation becomes: 1 Pe = [1 − erf (a)] 2 with a = − Cl σ 2 It must be noticed that the ratio Cl / σ is the location correction factor µ. The relationship between overall coverage probability and shadowing margins is given by the Jakes formula, obtained by integrating the previous equation all over the cell. 1  1 − 2ab    1 − ab    Pc = 1 − erf (a ) + exp   1 − erf    2  b²    b   where: Pc is overall cell coverage probability, a the same variable as above and 10n log(e) b= σ 2 n is the path loss exponent (usually 3,5 in cellular networks); e the exponential constant. This probability is calculated at each location with the appropriate values, taking into account standard deviation. There are many ways to use these formulas, whether a cell edge coverage or a whole cell coverage is required. As the requirements are given in terms of whole coverage, the margins will be computed accordingly, and the cell edge coverage indicated. ETSI 258 ETSI TS 102 584 V1.1.2 (2010-XY) In the satellite link budgets, there are different ways to consider the link margins: • for stationary reception, a Rice channel is assumed, leading to similar computation as above, providing availability versus link margins. Then link budgets give the possible link margins, and the corresponding availability; • for mobile reception (above 3 kmph), the link margins are given by the link budgets, and simulations provide the availability depending on conditions. 11.4.3 Technology dependant factors By this we mean the antenna gain, frequency band and the receiver noise factor. The different possible frequency bands have been explored in clause 4. In this context, the computations are done in S band (2 170 MHz to 2 200 MHz) all over the document, but L band is also a candidate band, and the results can be easily adapted. The noise bandwidth can have all the values described in the DVB-SH standard. For the computed examples provided in this clause, a noise bandwidth of 4,76 MHz is assumed, corresponding to the OFDM 5 MHz case. The terminal related factors are the Noise Factor, the antenna gain,, the antenna polarization, and the antenna temperature, resulting in the factor of merit G/T, used in the satellite link budgets. The different terminal categories and their related characteristics are described in.clause 10. 11.4.4 Network architecture or implementation optional factors This clause concerns some factors depending on some specific system architecture, mainly SFN gain and on implementation antenna diversity gain, that may additionally be taken in account by the Network planning tool. m) The SFN configuration The SFN topology can be considered in two cases: • terrestrial only coverage; • hybrid satellite/terrestrial coverage. In the terrestrial only coverage, it is an important feature of the DVB-SH. The SFN gain will depend on the network topology. When overlap is made possible by construction and the cell transmitters are properly synchronized, there is SFN gain with respect to a single-cell case. Similarly, the UMTS networks are also built as Single Frequency Network, probably more known under the name of networks with frequency reuse factor "1". In UMTS networks, the handover procedure occurs when a mobile is at the edge of two or three cells, where signal received from different nodes B are of similar strength; Thanks to the Rake receiver, terminal can acquire and combine the different signals coming from the Nodes B, increasing signal strength by the amount of the so called Soft Handover Gain (SHG). This point is studied in the literature. Different references provide analytical and experimental studies of Handover gain, leading to 3 dB to 4,6 dB Soft Handover gain. In DVB-SH/H SFN networks at the cell boundaries, the signals can be added coherently, thanks to the Guard Interval, leading to SFN gain. This gain can be translated into the following advantages: • increase of the coverage probability when dealing with equal transmit power as for a single cell; • decrease of Transmit power while keeping the same coverage and cell radius (equivalent to shadowing margins reduction); • increase of cell coverage. The resulting so called SFN gains are in a range between 3 dB and 5 dB. ETSI 259 ETSI TS 102 584 V1.1.2 (2010-XY) The network planner will take into consideration the requirements related to the SFN configuration. Namely, the deployment of the transmitters must be such that the signal coming from different stations can be added coherently (i.e. relative delay falls within the guard interval). When this is not respected the network will endure an additional level of self interference that may degrade the overall performance. The hybrid SFN topology and possible associated gain is not studied in this release of the Implementation Guidelines. n) Receiver antenna diversity Antenna diversity has been studied and implemented for a long time in the telecommunication industry, and also by the car equipment manufacturers, as it can provide improvement of reception quality. The antenna diversity is not a standard feature. Implementation depends on wavelength and terminal size. Typically, the use of S band allows implementing dual antenna diversity on handsets, while it will be more adapted for PDA in L band for instance. As reported, in different trial results, dual diversity gains using Maximum Ratio Combining are the following • 3,5 dB in conditions A, C and D • 5,5 dB in conditions B1 and B2 • 3 dB in Satellite Reception conditions It is not a standardized feature, though mentioned in clause 10, but as its efficiency has been on field proven, it will be used as an option in the different computations. 11.5 Terrestrial Network Planning based on minimum equivalent field strength (broadcast approach) 11.5.1 Introduction Traditional TV broadcast is based on roof service characteristics, typically 10 m height, from high towers, high powered sites, to high gain Yagi type receiving home antennas. Quality of service provided is measured through the computation of the minimum median equivalent field strength. Broadcast Network planning tools are usually proprietary to each broadcaster, that allow them to determine the transmitters optimal topology, and accordingly defined delivered quality of coverage. For DVB-H, these methods have been extensively discussed at the BMCO forum and have been enhanced to take into account the specifics of indoor service planning to individual terminals, such as in building penetration losses, shadowing margins (hereunder defined as location correction factor), thus introducing near ground coverage (1,5 m) in urban and rural areas, and also technology terms such as emission antenna gain, and reception antenna diversity gain. Only few DVB-H networks are actually deployed on large scale, thus providing little experience feedback; likewise actual cellular topologies and network planning based on these tools have limited background for broadcast services, and mostly rely on experiences from mobile radio services planning (detailed in the cellular approach clause). The following document gives an illustration of applicable methods and typical parameterizations as recommended by the BMCO forum. Mobile services, derived methodologies are presented in the cellular approach In the following, different calculations are made for the different reception conditions identified earlier and are given for S band only, but they can be extended easily to L band or to any other usable frequency band. 11.5.2 Quality of coverage The quality of coverage is provided in clause 11.2.4. 11.5.3 Network planning based on field strength computation methodology To compute the median filed strength, the following reception cases are considered for terrestrial reception: • reception condition A (outdoor portable); ETSI 260 ETSI TS 102 584 V1.1.2 (2010-XY) • reception condition B1 and B2 (indoor portable); • reception condition C Mobile (vehicle with roof top antenna); • reception condition D (portable, in vehicle with no roof top antenna). The C/N values will be the same as used in the previous clause on minimum signal level. To calculate the minimum median power flux density or equivalent field strength needed to ensure that the minimum values of signal level can be achieved at the required percentage of locations, the following formulas are used: Φmin = Ps min – Aa Aa = G + 10 log10 (²/4) Ps min f E = 4πη ⋅ Ga c or, in dB: E [ dBV/m ] = Ps min [ dB(mW) ] − Ga [ dB] + 77, 2 + 20 log10 f [ MHz ] This equation gives the link between the two possible methods, where: η = 120π Ω; Emin = Φmin + 120 + 10 log (120π) = Φmin + 145,8; Φmed = Φmin + Pmmn + Cl + Lo (Reception condition A or C); Φmed = Φmin + Pmmn + Cl + Lb + Lo (Reception condition B); Φmed = Φmin + Pmmn + Cl + Lv + Lo (Reception condition D). In case of antenna diversity use, all value of minimum power flux density must be written differently: Φmed/div = Φmed - Gdiv Emed = Φmed + 120 + 10 log (120π) = Φmed + 145,8 An equivalent formula is given here below: Emed = NF + 10*Log(kBT B) + C/N – G + 107,2 + 20*Log(f) + Lb (or Lv) + µ*σ + Lo (other losses) where: C/N : RF signal to noise ratio required by the system (dB) for the required performance and modulation/coding scheme; Φmin : minimum power flux density at receiving location (dBW/m2); Emin : equivalent minimum field strength at receiving location (dBV/m); Aa : effective Antenna Aperture; G : antenna gain (dBi); Lb : building penetration loss (dB); Lv : vehicle entry loss (dB); Lo : all other losses (polarization mismatch,.etc.); Psmin : minimum receiver signal input power (dBW); ETSI 261 ETSI TS 102 584 V1.1.2 (2010-XY) Pmmn : allowance for man made noise (dB); Cl : location correction factor (dB); Φmed : minimum median power flux density, planning value (dBW/m2); Emed : minimum median equivalent field strength, planning value (dBV/m). For calculating the location correction factor Cl a log-normal distribution of the received signal is assumed. The location correction factor can be calculated by the formula: Cl = µ*σ where: µ is the distribution factor, and σ is the standard deviation of the distribution. It must be noticed that this value of µ depends on the model assumed for the distribution (normally lognormal is considered for shadowing), as well as the standard deviation. The product is in fact the taken shadowing margins. Possible values are given in further clauses. 11.5.4 Network planning factors 11.5.4.1 Location Correction 11.5.4.1.1 Location correction for reception condition A In reception condition A (portable outdoor), the location variation factor is given by: Cl = µ*σ The common used value for σ is 5,5 dB (see ITU-Recommendation M.1225 [11]). This gives the table for the different qualities of coverage. Table 11.3: Location correction factor for reception condition A Edge of area coverage (%) Distribution factor (µ) Location correction factor (dB) 70 0,52 3 95 1,64 9 11.5.4.1.2 Building penetration losses and location correction for reception conditions B1 and B2 In reception conditions B, signal are subject to in building penetration losses. The location correction factor at indoor locations is the combined result of the outdoor variation and the variation factor due to building attenuation. These distributions are expected to be uncorrelated. The standard deviation of the indoor field strength distribution can therefore be calculated by taking the root of the sum of the squares of the individual standard deviations. As a consequence, the location variation of the field strength is increased for indoor reception. σ = (σ²o + σ²p)1/2 Table 11.4 gives the building penetration losses values proposed. Table 11.4: Building penetration losses Band S band Penetration loss Loss Condition B1 14 Condition B2 18 ETSI 262 ETSI TS 102 584 V1.1.2 (2010-XY) As an example, table 11.5 gives the location correction factor in deep indoor.(B2 reception) with overall: σ = 10 dB Table 11.5: Location correction factor for reception condition B2 in S band Edge of area coverage (%) Distribution factor (µ) Location correction factor (dB) 70 0,52 5,2 95 1,64 16,4 11.5.4.1.3 Location correction for reception conditions C and D Mobile reception is defined as a reception with moving receiver or at location where large objects moving around the receiver. Two cases are possible: • mobile reception with handset terminal inside a vehicle, car or inside any moving object (Reception condition D); • mobile reception with a vehicular terminal (Reception condition C). Clause 7 includes the values that should be used for planning in mobile reception. It should be taken into account the great influence of Turbo code FEC on the C/N and maximum Doppler shift (and therefore maximum speed in a particular channel).Any way, link budgets are performed with speed below the limit. (a) Location percentage requirements for mobile reception (conditions C and D) For reception categories C and D, some additional margins are added to cope with mobile environments (see EN 302 304 [3]). Table 11.6: Location correction factor for reception conditions C and D Overall area coverage (%) Distribution factor (µ) Location correction factor (dB) 90 1,28 7 99 2,33 12,8 (b) Vehicle entry loss For mobile reception inside cars or any other vehicle entry loss must be taken into account. For instance, an entry loss is 7 dB is taken for in car penetration losses. The location correction factors will be the same for in vehicle reception. 11.5.4.2 Network architecture and implementation optional factors (a) The SFN possibility The SFN gains appear at the cell overlap points. The field strength computation is made all over the area and not at specific points. So SFN gain will not be considered in this clause. (b) Receiver antenna diversity As recalled in section 11.4.4 b) thre will be two cases: single antenna reception and dual antenna reception The proposed gains are the following • 3,5 dB in conditions A, C and D • 5,5 dB in conditions B1 and B2. The filed strength reductions are equal to the gain provided above. ETSI 263 ETSI TS 102 584 V1.1.2 (2010-XY) 11.5.5 Field strength computation examples The objective of the link budget is to provide the minimum field strength values at 1,5 m height. Normally, the exercise can be made for all possible values of C/N. We limit the examples to a few values corresponding to some situations in the Clause A.12. Extrapolation from the proposed values is straightforward. An example of each reception condition is provided in S band, only to limit the number of tables. Antenna diversity is used for reception conditions A, B and D, though it can also be implemented for vehicle roof top reception. Two modulation and coding schemes are studied: • QPSK 1/3; and • 16QAM 1/3. The required C/N values are extracted from the clause 12.. In the following clause, we give an example of table, and a synthesis for different examples. The C/N values are coming from laboratory measurements and thus include the implementation losses, and are given for FER 5 %. It must be noticed that maximum speed in the tables (conditions C and D) is 50 kmph, below the CN + 3 dB Doppler limit. 11.5.5.1 Field strength computation for reception condition B2 The studied case is with Handheld Category 3 in reception condition B2(3 kmph) and the link budgets are computed for 99 % of ESR5(20) fulfilment at different coverage qualities as indicated in the table 11.7.(Tableau à reclculer avec les valeurs dE C§N) Table 11.7: Median field strength in reception condition B2 with QPSK 1/3 and 16 QAM 1/3 single antenna Modulation and coding QPSK 1/3 16QAM 1/3 Frequency band (MHz) 2 182,5 Equivalent Noise Bandwidth B (MHz) 4,76 Receiver Noise Figure NF (dB) 4,5 Minimum C/N required by system (dB) 2,8 9,1 Min receiver signal input power Psmin (dBW) -129,90 -123,60 Min. equivalent receiver input voltage, 50  Us min (dBµV) 7,1 13,4 Antenna gain relative in dBi G (dBi) -3 Effective Antenna aperture Aa (dBm²) -31,23 Min. power flux density at receiving location  min (dBW/m²) -99 -92 Min. equivalent field strength at receiving location Emin (dBµV/m) 47 53 In building or in car penetration losses Lb (dB) 18 Location correction factor for 70 % edge of Cl (dB) 5,2 coverage Minimum median power flux density at 1,5 m med (dBW/m²) -75 -69 Minimum median equivalent field strength at Emed (dBµV/m) 1,50 m 70 77 Location correction factor for 95 % edge of Cl (dB) 16,4 coverage Minimum median power flux density at 1,5 m med (dBW/m²) -64 -58 Minimum median equivalent field strength at 1,5 m Emed (dBµV/m) 82 88 Table 11.8: Median field strength in reception condition B2 with QPSK 1/3 and 16 QAM 1/3 dual antenna ETSI 264 ETSI TS 102 584 V1.1.2 (2010-XY) Modulation and coding QPSK 1/3 16QAM 1/3 Frequency band (MHz) 2 182,5 Equivalent Noise Bandwidth B (MHz) 4,76 Receiver Noise Figure NF (dB) 4,5 Minimum C/N required by system (dB) 2,8 9,1 Min receiver signal input power Psmin (dBW) -135,40 -129,10 Min. equivalent receiver input voltage, 50  Us min (dBµV) 1,6 7,9 Antenna gain relative in dBi G (dBi) -3 Effective Antenna aperture Aa (dBm²) -31,23 Min. power flux density at receiving location  min (dBW/m²) -104 -98 Min. equivalent field strength at receiving location Emin (dBµV/m) 42 48 In building or in car penetration losses Lb (dB) 18 Location correction factor for 70 % edge of Cl (dB) 5,2 coverage Minimum median power flux density at 1,5 m med (dBW/m²) -81 -75 Minimum median equivalent field strength at Emed (dBµV/m) 1,50 m 64,83 71,13 Location correction factor for 95 % edge of Cl (dB) 16,4 coverage Minimum median power flux density at 1,5 m med (dBW/m²) -70 -63 Minimum median equivalent field strength at 1,5 m Emed (dBµV/m) 76 82 11.5.5.2 Synthesis Table 11.9 gives a synthesis of the previous link budgets based on minimum filed strength. Table 11.9: Overall synthesis (Field Strength in dBµV/m) single antenna reception Reception Quality of Service cases QPSK 1/3 16 QAM 1/3 conditions/terminal Condition A/ Handheld Cat 3 70 % edge 50 57 95 % edge 56 63 Condition B2/ Handheld Cat 3 70 % edge 70 77 95 % edge 82 88 Condition C/Vehicle Cat 1 90 % overall 47 52 99 % overall 53 58 Condition D/ Handheld Cat 3 90 % overall 60 66 99 % overall 66 72 Table 11.10: Overall synthesis (Field Strength in dBµV/m) dual antenna reception Reception Quality of Service cases QPSK 1/3 16 QAM 1/3 conditions/terminal Condition A/ Handheld Cat 3 70 % edge 47 54 95 % edge 53 60 Condition B2/ Handheld Cat 3 70 % edge 65 72 95 % edge 77 83 Condition C/Vehicle Cat 1 90 % overall 44 49 99 % overall 50 55 Condition D/ Handheld Cat 3 90 % overall 57 63 99 % overall 63 69 11.5.6 Use of field strength based Radio Network Planning Tool example for DVB-SH Network Design For further study. ETSI 265 ETSI TS 102 584 V1.1.2 (2010-XY) 11.6 Terrestrial Network Planning based on minimum received signal level (cellular approach) Network Planning tools allow to compute precise site distribution and cell ranges. They use propagation models to predict with high accuracy path loss, and different cell ranges. These models integrate field based calibration. Some Practical Network Planning tools, include a propagation model called Standard Propagation Model, based on Cost Hata model with different correction terms allowing range of application down to 200 m from the BTS. Others suppliers provide tools based on ITU-R Recommendation P.526 [27] model. The complete description, parameterization of the tool, and procedures of use is not in the range of the present document. A network planning tool will compute the coverage using an intermediate parameter called "the minimum signal level for coverage". This parameter corresponds to the minimum signal power level that will ensure service reception. It takes into account propagation characteristics associated to the considered type of scattering environment. The network planning tool will plot coverage maps. This clause gives an overview of the common practices of the cellular industry (cellular operators), and provides a first level approximation of DVB-SH networking, including the computation of cell range, given the transmitter EIRP. After some considerations on network topology, we provide the basic equations used in the network planning tools. 11.6.1 Cellular network topology Without going into complex topological details, a cellular network is composed of an aggregation of cells that can be considered in first approximation as circle of radius R.. This radius defines the edge of cell and is considered in the link budget computation.. Figure 11.1 shows a regular cellular network with trisectorized cells and in SFN possible topology. R Figure 11.1: Trisectorized cell topology For the trisectorized antenna on the left of the figure, the coverage area of the three sectors is approximated by a regular hexagon. This approximation makes it easy to show the overall coverage area in the figure on the right, as the hexagons completely cover the plane area without overlap. The surface of a cell is the following S= 1,95 R². Given the intersite distance D, relationship between D and R is the following: D = 3R/2. This formula can be used to provide estimates of the different of required sites to cover an area. ETSI 266 ETSI TS 102 584 V1.1.2 (2010-XY) 11.6.2 Quality of coverage Given a certain quality criterion, different whole cell coverage probability. can be considered. The 90%coverage probability means that 90 % (for instance) of the cell can receive at least the minimum signal level required for insuring the target C/N value. Usually 90 % and 95 % overall cell coverage are considered. This corresponds respectively to 77 % and 88 % edge of cell coverage. But other values can be as well required. Usually, network dimensioning is made for indoor service with handheld at 90 % or 95 % cell coverage., Such indoor coverage probabilities imply de facto in very high outdoor coverage probability, above 99 %. 11.6.3 Network planning tool A typical numerical network planning tool uses the plotting of the Minimum Received Signal Level. Starting by the following equation: Pn = NF + 10* log (k ∗ T0 ∗ B); Ps min = Pn + C/N. where: NF is the Noise Factor of the receiver; K: Boltzmann constant; T0: receiver temperature; B: noise bandwidth of the signal; C/N is the required C/N for a given quality of service, depending on the modulation, coding and propagation channel. So the minimum received signal input power at antenna input can be written: Pamin = Ps min – G + Lp = C/N – G + Lp + NF + 10* log (k ∗ T0 ∗ B). It must be noticed that power unit is dBm, Where: G is the antenna gain in dBi and Lp is possible polarization mismatch losses. On the other hand the basic link budget equation can be written as follows, considering the received C/N: C/N = EIRPTx – PL – Ms – Lb – Lv+ G – NF – 10* log (k ∗ T0 ∗ B) + GSFN (equation 1). where: EIRPTx: transmitter EIRP; PL : path loss; Ms : shadowing margins, following a Log Normal law of standard deviation ; Lb : building penetration loss (dB) (if any); Lv : vehicle entry loss (dB) (if any); Lp : polarization losses (if any). EIRPTx = Txpower – (losses) + Tx (antenna gain) where: transmitted power is in dBm, losses in dB, and antenna gain in dBi. Path loss is computed using different path loss models that are included in the network planning tool. There is a variety of PL models (Cost 231 Hata, Welfish Ikegami, Xia-Bertoni) that are usually on field calibrated. For the purpose of ETSI 267 ETSI TS 102 584 V1.1.2 (2010-XY) global link budgets for rough order of magnitude estimates, the Cost Hata model is used in the link budgets with some calibrated factors. After some elementary algebra, equation (1) can be written: C/N – G + Lp + NF + 10* log (k ∗ T0 ∗ B) + Ms + Lb + Lv – ( GSFN) = EIRPTx –PL (equation 2). The left part of the equation is the Minimum Received (Rx) Signal level, independent of transmitted power. Other definitions can be introduced: • System Gain (SG): SG = EIRPTx – Pamin; • MAPL (Maximum Allowable Path Loss ) = SG – Ms - Lb – Lv + GSFN. Leading to equivalent definition of the Minimum Received (Rx) Signal level. 11.6.4 Network planning factors 11.6.4.1 Shadowing margins In cellular network radio planning it is usual to distinguish between different environments, while this classification does not exist in EN 302 304 [3]. • Dense Urban (DU): dense cities like Paris; • Urban (U): cities like immediate Paris suburbs called also "Dense Individual, Mean collective"; • SubUrban (SU): residential areas, called also "Mean individual, mean collective" including as well industrial zones; • Rural (RU): open areas : small towns and villages, highways, fields. The last case is not applicable for DVB-SH, as rural areas will be under satellite coverage. Shadowing margins follow a Lognormal law of standard deviation usually in the range of 6 dB to 10 dB, or even more for some authors [i.11], [i.10]. For network planning exercise, a common value of 8 dB is assumed for urban and suburban areas in [i.11], [i.12], [i.13], [i.14], and [i.15]. Concerning in building penetration losses, there is also a wide range of values. For instance, deep indoor penetration losses are in the range of 18 dB to 21 dB, and sometimes more. In the following clauses, we give the possible ranges of values, and give an example for network planning purpose. 11.6.4.1.1 Shadowing margins for reception condition A As proposed value for standard deviation is 8 dB, table 11.11 gives the shadowing margins for various conditions and different cell coverage. Usual coverage probability values whole cell are 90 % and 95 %. Table 11.11: Shadowing margins for reception condition A DU/U/SU Cell Coverage (%) Shadowing margins (dB) 90 5,5 95 8,7 11.6.4.1.2 In building penetration losses and shadowing margins for reception condition B In building penetration margins values depend on the different materials and the environments. Table 11.12 presents the possible range of values, and the sued values in the network planning. But the taken values can vary from an operator to another. ETSI 268 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.12: Indoor penetration margins in S band for reception condition B Environment/B1 or B2 DU/B2 U/B2 SU/B1 Indoor penetration margins range (dB) 18 to 21 15 to 18 13 to 15 proposed value for network planning 18 15 13 NOTE: For the shadowing margins, the same formula is applied when dealing with the combination of indoor and outdoor signal variations. The following standard deviation values are taken: • outdoor standard deviation: 8 dB for DU, U and SU; • indoor standard deviation: 6 dB for all cases. The shadowing margins are provided in table 11.13. Table 11.13: Shadowing margins in reception conditions B1 and B2 Environment DU U SU Standard deviation (dB) 10 10 10 Shadowing margins @ 90 % cell coverage 7,6 7,6 7,6 Shadowing margins @ 95 % cell coverage 11,6 11,6 11,6 11.6.4.1.3 Shadowing margins for reception conditions C and D In reception condition C, the same shadowing margins will be used as for reception condition A, as we are in outdoor reception. In reception conditions D, an in vehicle penetration loss of 7 dB is required, and the shadowing margins are the same as for condition C. 11.6.4.2 Network architecture optional parameters This clause concerns some criteria depending on some specific system architecture, mainly SFN gain and antenna diversity gain. (a) The SFN configuration The used topology is similar to the UMTS one. As the link budgets are computed at cell edge, SFN gain can be taken into account and 4,7 dB gain is used in the link budgets. (b) Receiver antenna diversity As recalled in section 11.4.4 b) there will be two cases: single antenna reception and dual antenna reception The proposed gains are the following • 3,5 dB in conditions A, C and D • 5,5 dB in conditions B1 and B2. 11.6.5 Link budgets examples for DVB-SH network planning based on minimum received signal level For cellular network, global estimation uses link budgets for budgetary purposes. As it gives only a rough order of magnitude of the coverage, Cost Hata model is also commonly used below 1 km. In the link budgets, the power of each transmitter is provided, with different losses (cable losses, diplexer losses if any: between 3 dB and 4 dB overall), the antenna gain (18 dBi per sector in the following examples), resulting in the transmitted EIRP. Diplexer losses are considered when the DVB-SH transmitter is sharing equipments with other BTS (3G for instance. When stand alone repeater, non diplexer loss is considered. ETSI 269 ETSI TS 102 584 V1.1.2 (2010-XY) The results of the link budgets are the Minimum Signal Level for Network Planning (green line), and also the cell radius, and cell surface, given the reception condition, the terminal properties, and the required C/N. Two modulation and coding schemes are studied: • QPSK 1/3; and • 16QAM 1/3. The required C/N values are extracted from the clause 12 (Table A .7?). In the following clause, we give an example of table, and a synthesis for different examples. The C/N values are coming from laboratory measurements and thus include the implemnation losses, and are given for FER 5 %. It must be noticed that maximum speed in the tables (conditions C and D) is 50 kmph, below the CN + 3 dB Doppler limit. The following cases are provided: • reception condition A: Outdoor pedestrian in dense urban area; • reception condition B (B1 and B2): indoor pedestrian in dense urban area; • reception condition C: vehicle at 50 kmph in suburban area; • reception condition D: in car user at 50 kmph in suburban area. As in the clause 11.5, we give an examples for reception condition B, and the synthesis for the other cases. 11.6.5.1 Link budgets for Reception condition B2 The studied case is with Handheld Category 3 and the link budgets are computed for 99 % of ESR5(20) fulfilment over 95 % overall coverage in dense urban. ETSI 270 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.14: Link budget for reception condition B2 @ 95 % with Tx power of 12 W and QPSK 1/3 SHA Terrestrial Link Budgets Dense Urban Dense Urban value/ Radio interface parameters Unit value/No Diversity Diversity Channel bandwidth MHz 5,00 5,00 Frequency MHz 2 182,50 2182,50 Mode 2 048,00 2048,00 Radio interface mod code QPSK1/3 QPSK1/3 Antenna type TRI TRI 1/Guard Interval 8 8 Guard interval 0,125 0,125 Total number of subcarriers 1 705,00 1705,00 Number of data subcarriers 1 512,00 1512,00 Tu duration µs 358,40 0,18 GI duration µs 44,80 358,40 Ts duration µs 403,20 44,80 Subcarrier spacing kHz 2,79 403,20 Useful bandwidth occupancy MHz 4,76 2,79 Useful data rate at MPEG2-TS interface Mbit/s 2,490 2,490 Transmitting end Tx Tx Power amplifier per carrier and 12,0 sector W 12,0 Tx Power at antenna input dBm 40,8 40,8 Cable loss dB 3,0 3,0 Diplexer loss dB 0,7 0,7 Tx antenna gain dBi 18,0 18,0 EIRP dBm 55,1 55,1 ERP W 323,0 323,0 Receiving end Rx Rx Rx antenna gain dBi -3,0 -3,0 Polarization mismatch dB 0,0 0,0 Noise figure dB 4,5 4,5 Antenna temperature K 290,0 290,0 Ambient temperature K 290,0 290,0 kT dBm/Hz -174,0 -174,0 Equivalent Rx band dBm/Hz 66,8 66,8 Rx noise floor dBm -102,7 -102,7 Required C/N dB 2,8 2,8 Rx sensitivity dBm -99,9 -99,9 Minimum Rx level at antenna dBm -96,9 -96,9 Minimum Signal Level for Network Planning dBm -72,01 -77,51 System Gain dB 152,0 152,0 Margins Average building penetration loss dB 18,0 18,0 Target level of signal penetration B2 B2 Std shadowing outdoor dB 8,00 8,00 Std shadowing indoor dB 6,00 6,00 Std dev of Fading Margin dB 10,00 10,00 Propagation constant 3,5 3,5 Shadow fading Margin - whole cell dB 11,6 11,6 Coverage Probability - whole Cell 95,0 % 95,0 % SFN network gain dB 4,7 4,7 Rx gain (antenna diversity) dB 0,0 5,5 ETSI 271 ETSI TS 102 584 V1.1.2 (2010-XY) MAPL dB 127,1 132,6 Cell range computation Cost 231-HATA model Cell range km 0,593 0,850 Surface km2 0,69 1,41 H-BTS m 30,0 30,0 H-MS m 1,5 1,5 K1 dB 138,08 138,08 K2 dB 35,22 35,22 Kc dB -3,0 -3,0 a(CPE) dB 0,00 0,00 Note 1 : Cell radius is 593 m in dense urban environment with single antenna reception Note 2 : Cell radius is 850 m in dense urban environment with dual antenna reception 11.6.5.2 Overall synthesis As in the minimum field strength approach, the tables 11.15 and 11.15provides a synthesis concerning different reception conditions. Table 11.15: Overall synthesis @ 95 % coverage with single antenna reception Reception Condition A in dense urban B2 in dense urban C in suburban D in suburban Modulation /coding QPSK 1/3 16 QAM QPSK 1/3 16 QAM QPSK 1/3 16 QAM QPSK 1/3 16 QAM 1/3 1/3 1/3 1/3 Power per sector (W) 7 27 12 32 7 25 7 25 Required C/N (dB) 2,8 9,1 2,8 9,1 2 7,2 2 7,2 Minimum Signal Level for -93 -87 -72, -66 -100 -95 -87 -81 Network planning (dBm) Cell range (km) 1,997 1,940 0,593 0,519 5,102 5,202 2,111 2,157 Table 11.16: Overall synthesis @ 95 % coverage with dual antenna reception Reception Condition A in dense urban B2 in dense urban C in suburban D in suburban Modulation /coding QPSK 1/3 16 QAM QPSK 1/3 16 QAM QPSK 1/3 16 QAM QPSK 1/3 16 QAM 1/3 1/3 1/3 1/3 Power per sector (W) 7 27 12 32 7 25 7 25 Required C/N (dB) 2,8 9,1 2,8 9,1 2 7,2 2 7,2 Minimum Signal Level for -98 -90 -77 671 -104 -98 -90 -85 Network planning (dBm) Cell range (km) 2,510 2,439 0,850 0,744 6,413 6,552 2,653 2,711 It must be noticed, that its is easy to extrapolate the Minimum Signal Level to other values of C/N. 11.6.6 Use of radio network planning tool example for DVB-SH network design The Radio Network Planning Tool allows to compute and show coverage maps based on the minimum received power level required to ensure service reception (depends on scattering environments). The design is based on the use of a Radio Network Planning Tool (RNPT) and a geographical database. The inputs of the network design are: • the DVB-SH link budget for a precise configuration of the RNPT; • the clutter and terrain model databases; • the minimum signal levels for coverage (design level); and ETSI 272 ETSI TS 102 584 V1.1.2 (2010-XY) • eventually the characteristics of existing 2G/3G sites if available for an eventual reuse of the antennas and feeders. (This is not a requirement: it can exist stand alone repeaters). The outputs of the network design are: • coverage maps with received power level or C/N+I level; • number of sites; • coverage statistics. The goal of the network design process with the RNPT is to reach the minimum received power level defined in the link budget (i.e. coverage) for all scattering environments over the area considered in the design). A RNPT allows the designer performing the coverage of one area by positioning transmitting sites within the database. The Minimum Rx signal level to be considered in the RNPT can be computed as follows: Minimum Rx signal level = Tx EIRP – Maximum Allowable Path Loss (MAPL) Using the Link budget formula provided above, it can be further computed from as follows: Minimum Rx signal level = Min Rx input power + (Shadowing Margin + Building Penetration loss [see note 1]) - (SFN gain [see note 2] + Rx diversity gain) NOTE 1: If Indoor coverage is targeted. Depending on the situation it can be also in vehicle penetration losses. NOTE 2: SFN impact can also supported by the tool. For a highly accurate network design especially in dense urbanized areas, ray-tracing simulations can be considered for a better reliability of the coverage of the network. For these simulations a deterministic propagation model (ray optical propagation model) is used, requiring a 3D building high-resolution database. This combination of ray-tracing and 3D vector database allows a higher accuracy and reliability of the coverage maps. In some tools, SFN gain is added "manually", other advanced tools integrates already the SFN gain. 11.7 Satellite Network Planning For the satellite network planning, the following reception categories are considered: • Reception condition A (outdoor portable) for rural environment; • Reception condition B1 with domestic gap filler; • Reception condition C Mobile (roof top antenna). For satellite network planning purposes, the following satellite to terminal propagation environments will be considered: 6) suburban or rural areas (villages); 7) open area or also rural (open field); 8) rural intermediate tree shadowing (road crossing a forest). We have excluded from this analysis the urban and rural heavy tree shadowing environments as the satellite is not expected to deliver the DVB-SH service in these kind of environments. 11.7.1 Methodology for Satellite Coverage Calculation The DVB-SH satellite coverage is a composite of the three environments mentioned earlier: suburban, open area and intermediate tree shadowing. Each environment has distinctive propagation characteristics and therefore DVB-SH physical and link layer performances are expected to be different. Moreover the required quality criteria in terms of ESR5(20) is also dependent on the propagation conditions. For terrestrial channels that can be characterized as Rayleigh channels the requirement is 99 % of ESR5(20) fulfillment; however, the satellite channel presents very different ETSI 273 ETSI TS 102 584 V1.1.2 (2010-XY) dynamics and therefore this requirement cannot be directly translated. At the present moment, we lack sufficient information to conclude on the required ESR5(20) fulfillment necessary to guarantee a good video quality via satellite. The calculations presented in this clause are based on the assumption that a 90 % ESR5(20) fulfillment is required, however the methodology is considered of general validity. To get satellite coverage figure over a wide area it is necessary to: • classify the satellite coverage region according to three environment categories listed above and associate a probability of occurrence pE (i ) ≤ 1, i = 1, 2,3 for each of them over the satellite coverage; • select typical mobile terminal speed for each environment; • select an ESR5(20) target fulfillment rate per environment FESR 5 (i ) [ ]min , i = 1, 2,3 . As explained earlier, this target fulfillment rate depends on the environment considered. For the calculations on the present document we will consider that for the satellite environments 90 % of ESR5(20) fulfillment is required: • define the system parameters required to compute the link budgets providing LOS C/N and C/I values as described in the following clause; • define the DVB-SH waveform parameters required to perform the simulations as described in Clause A.3; • analyze the ESR5(20) fulfillment rate FESR 5 (i ), i = 1, 2,3 achieved in each of the environments given the selected system and waveform parameters (see Clause A.12 for typical results); • compute the satisfaction index χ (i ), i = 1, 2,3 for each satellite coverage environment defined as: χ (i) = { 1 if FESR 5 ( i ) ≥[ FESR 5 ( i )]min 0 if FESR 5 ( i ) <[ FESR 5 ( i )]min i = 1, 2,3 • compute the overall satellite coverage CSAT (in % as): 3 CSAT (%) = 100 pE (i ) χ (i ) ; i =1 where pE(i) is the probability of occurrence of each environment. Note that the above described methodology assumes that the satellite elevation angle is not changing in an appreciable way over the coverage region. In case of large satellite beams, as it is the case of a global beam covering a large part of the continent, the above described procedure must be extended by splitting the satellite coverage over regions of similar elevation angle. 11.7.2 Basic formulas for Satellite link budgets calculation As was done for the terrestrial case, the basic formulas for the calculation of the Signal to Noise ratio C in the N satellite link are recalled hereafter. Satellite link budget is made of two components: uplink (from Gateway to Satellite) and a downlink (from Satellite to User Terminal). The formula for the uplink C calculation can be expressed as N follows: C  1 1 G   N  = kB EIRPGW L  T   U U   SAT where: k: Boltzmann's constant; B: Noise Bandwidth; EIRPGW : Gateway Equivalent Isotropic Radiated Power; ETSI 274 ETSI TS 102 584 V1.1.2 (2010-XY) LU: Uplink losses; G  T  : Satellite Antenna Gain over the Satellite Noise Temperature;   SAT PGWTX GGW EIRPGW = LT L FTX where: PGWTX is the transmitted power of the Gateway; GGW is the gateway antenna gain; LT is the de-pointing losses; and LFTX is the feeder loss between the transmitter and the antenna; LU = LFS L A where LFS is the free space losses (see formula below) and LA is the atmospheric attenuation losses; 2  4πR  L FS =   with R defined as the distance between the gateway and the satellite.  λ  In a similar manner, for the downlink the following formula applies:  C  1 1 G    = EIRPSAT N   D kB  L D  T  ES   where EIRPSAT: satellite Equivalent Isotropic Radiated Power; LD: Downlink losses; G   T  : Terminal Antenna Gain over the Terminal Noise Temperature;   ES P G EIRPSAT = SAT SAT ; LT L FTX where PSAT is the transmitted power of the Satellite; GSAT is the satellite antenna gain; LT is the de-pointing losses; and LFTX is the feeder loss between the transmitter and the antenna. LD = LFS L A ; where LFS is the free space losses (same formula as above considering R as the distance between the satellite and the terminal); and LA is the atmospheric attenuation losses in the downlink. ETSI 275 ETSI TS 102 584 V1.1.2 (2010-XY) Additional to the uplink and downlink C components, the link budget needs to take into account the interference N contribution. In general, interference is modelled as an additional source of noise: −1 −1 −1  C  C  C  N + I  =  +    N  I  where I is the interference component caused by multiple sources of interference: intermodulation (IM), co-channel interference (CC), and adjacent channel interference (ACI): I = I IM + I CC + I ACI . In the following we will assume a transparent GEO satellite system (no modulation or coding on-board) which is limited by the downlink component (from satellite to user mobile terminal). This last assumption is consistent with the DVB-SH scenario where small hand-held or mobile terminals are targeted. The basic link budget for transparent satellite systems is computed as follows: −1 −1 −1  C   C   C  N +I  =  +  .    N + I  uplink  N + I  downlink When the downlink part is driving the link budget, the overall link budget can be approximated by the downlink component:  C   C  N + I  ≈ N + I  .     downlink 11.7.2.1 Satellite link margin In order to close the link, the available Signal to Noise plus Interference ratio needs to be greater than the required Signal to Noise ratio: for given waveform configuration (MODCOD):  C  C  N + I  ≥ N  .     REQUIRED MODCOD This required C/N guarantees the target BER and it depends on the Modulation and Coding and on the channel (in general considered as being AWGN). Link margin is generally defined as the difference between the available Signal to Noise plus Interference ratio and the required signal to noise ratio:  C  C  M =  (dB ) −  N  (dB ) . N + I    REQUIRED MODCOD 11.7.2.2 Satellite fade margin When there is a fade in the downlink (cause by shadowing, partially blockage of the signal, etc.) we can expect the nominal signal power to be degraded by an attenuation α. C fade = C.α When the main source of interference comes from the intermodulation and from the co-frequency beams the fade affecting the carrier also affects the interference. Therefore the ratio C/I remains unchanged.  C.α   C   I .α  =  I  if I = I IM + I CC .     ETSI 276 ETSI TS 102 584 V1.1.2 (2010-XY) And therefore the overall C/(N+I) in the presence of fade affecting both the useful signal and the main sources of interference is as follows: −1 −1 −1  C   Cα  C  N +I  =  +  .    N  I  Fadedownlink Now, by fade margin we refer to the value of α so that the link budget can be closed (i.e. is equal or greater than the required): −1   Cα  −1  C  −1  C    +I   =  .  N        N  REQUIRED MODCOD And therefore, the fade margin (in dB) can be calculated by:  C −1 −1  C    C  M fade = −α (dB ) =   (dB ) + 10 log    −  . N    N  REQUIRED MODCOD  I     When the link budget is not limited by the interference, i.e. when C/I>> required C/(N+I) and C/(N+I)≈C/N; or when the interference is dominated by the components not affected by the same fade as the signal (i.e. adjacent channel interference or co channel interference from other systems), then the fade margin is the same as the link margin: C   C  M fade ≈   (dB ) −   ≈M . N   N + I  REQUIREDMODCOC 11.7.3 Required margins computation The above stated formulas can be used to calculate the Signal to Noise ratio received by the terminal in a AWGN channel, but in order to close the link in the propagation channels applicable to DVB-SH, it is required to assess the additional fade endured by the satellite signal due to the mobile propagation channel. Contrary to the terrestrial case, the satellite propagation channel is considered non-frequency selective. Therefore, the fade associated with the satellite to mobile propagation channel will homogeneously affect the satellite signal along all its bandwidth. This clause analyses the required fade margins for the satellite link. As stated before, these margins depend on the reception conditions and on the environment considered. For reception condition A, the methodology proposed here considers that the terminal is in direct LOS and in a static situation (neither the terminal nor the surrounding geometry changes or changes are very slow compared to physical and upper layer countermeasures available in DVB-SH). In this static condition, time diversity mechanisms such as the physical layer interleaver or the link layer FEC do not provide any advantage. Fade due to multipath needs to be counteracted via a modulation and coding that allows for a sufficient link margin (i.e. Mfade>Fade) or via spatial diversity mechanisms such as antenna diversity (the latter not analyzed in the following). Concerning reception conditions C and D, the mechanisms introduced in DVB-SH to counteract the signal fluctuations in the satellite to mobile channel impact the link margin required to guarantee the service quality (ESR5(20)). If in general the margin is calculated so that the link budget is closed under fade conditions for a certain modulation and coding (and a given BER), in the presence of physical layer FEC interleaver and/or link layer FEC, the quality criteria (ESR5(20)) can be met even if the fade temporarily exceeds the available margin. In fact, within the physical layer or link layer protection time, the fade can exceed the available margin for a period of time as long as this period does not go beyond the correction capabilities of the interleaver or link layer. The correction capabilities depend on the specific implementation of the physical or link layer protection. For any given link margin, the overall signal quality performance depends on this correction capability and on the dynamics of the channel (i.e. kind of environment and terminal speed). Therefore, it is not possible to derive a required fade or link margin value of general validity. The margins proposed in the clause dedicated to mobile reception are based on the available physical layer simulation results for different environments. A methodology will be also proposed on how this results can be used to calculate the availability over the coverage area. ETSI 277 ETSI TS 102 584 V1.1.2 (2010-XY) 11.7.3.1 Margins required reception condition A In portable reception conditions addressed in this clause we consider that the terminal is in a static situation where the fade due to multipath can be counteracted only with sufficient link margin. In the case of reception condition A (i.e. outdoor pedestrian), satellite reception is associated to a rural environment. Direct reception of satellite signal to handheld terminals (terminal categories 2a, 2b) is achieved only when the terminal is in direct Line-of-Sight. In these conditions, the satellite-to-mobile terminal channel can be considered as a Ricean channel with a Carrier to Multipath Ratio that depends on the geometry of the environment. In order to calculate the required margin associated to the Ricean channel, the following calculations apply: Let us consider a signal s with a Signal to Noise Ratio in AWGN channel hereafter called [C/N]AWGN (calculated according to the formulae given in the previous clause). This signal experiences a fade that instantaneously degrades the received C/N. This instantaneous C/N, hereafter called γ, can be expressed as follows: C  γ =  r2 where r is the fading process amplitude.  N  AWGN In the case of a Ricean channel the distribution of the fading amplitude r is given by the Rice distribution (see also Clause A.7): r ( ) − r 2 + s 2 / 2σ 2 pR ( r ) = σ2 e I0 ( ), rs σ2 where σ2 is the variance of either the real or the imaginary components of the multipath; and s is the amplitude of the LOS signal component. To find the distribution of the instantaneous Signal to Noise ratio γ from the Rice distribution, we use the identity: dr pΓ ( γ ) = pR ( r ) . dγ We also define Rice factor K as in Clause A.7 and we normalize to unity the LOS signal power: s2 K= ; s2 = 1. 2 2σ Therefore, we obtain the following expression for the Probability Density Function (PDF) of the instantaneous Signal to Noise ratio γ:       −K C  γ +1   K  N   γ pΓ ( γ ) = e    AWGN  I0  2K . C   C   N    AWGN  N    AWGN    And the CDF of γ can be expressed by:     γ CDFΓ ( γ ) =  px ( x ) dx = 1 − Q1  2 K , 2 K . γ −∞  C    N    AWGN    ETSI 278 ETSI TS 102 584 V1.1.2 (2010-XY) where Q1 is the Marcum's Q function: ∞ Q1 (a, b ) = e −(a )  ( ) I (ab) a t 2 +b 2 / 2 b t b>a>0 t =o According to the above expressions, the following figure represents the cdf of the degradation of instantaneous Signal to Noise Ratio with respect to the Signal to Noise Ratio in AWGN for different values of Carrier to Multipath ratio K. Figure 11.2: Ricean fade power loss CDF for different Rice factors K (in dB) In a mobile channel, this instantaneous Signal to Noise ratio γ, is a time variant variable. When the terminal is in a static situation, as it is the case of category A reception, the Signal to Noise ratio becomes a position dependent variable. The time probability can be mapped into coverage probability and therefore the margins required to reach a certain percentage of the coverage can be calculated using the CDF above. This approach is homologous to the location correction factor used in the terrestrial coverage calculations seen earlier.   C     C   C  C    Pout = Pr γ <    = Pr   −γ >   −      N  REQ      N  AWGN  N  AWGN  N  REQ   C  C  Where the quantity   − represents the Fade Margin as defined earlier (in the case of no  N  AWGN  N  REQ   interference). ETSI 279 ETSI TS 102 584 V1.1.2 (2010-XY) According to the previous calculations, the following figure summarizes the fade margin required for different values of Rice factor K and different percentages of covered area: Figure 11.3: Fade margin required for different carrier to multipath ratios and coverage probability Typical values for the Rice factor K range between 5 dB and 10 dB for hand-held terminal. Values can be higher in open areas. It has to be noted that in the case of static reception conditions, the presence of a long physical layer interleaver or link layer FEC does not provide a real advantage in terms of additional coverage or availability since these mechanisms are effective to counteract the fluctuations of fade that occur only when the terminal is on the move. Instead the presence of mobile terminal antenna diversity may greatly help to reduce the required link margin for this case. 11.7.3.2 Margins required in reception conditions C Mobile reception conditions for satellite refer to reception condition C for the rural environment. Mobile speeds analyzed in this clause cover also low speeds (3 kmph). The Empirical Roadside Model briefly described in clause 4 allows determining the shadowing endured by the satellite link for a certain coverage probability in an equivalent way as the location correction factor does for the terrestrial network planning. However, as discussed previously, DVB-SH implements a long physical layer interleaver or/and a Link Layer FEC that alters the amount of shadowing that can be endured by the satellite link (without causing outage) when the terminal is moving with a certain speed. Therefore, this model cannot be directly translated into a required shadowing margin for our network planning calculations. ETSI 280 ETSI TS 102 584 V1.1.2 (2010-XY) In order to take into account the time diversity introduced by the DVB-SH physical and link layer, the network planning needs to rely on the simulations of the physical and link layer over a channel model with representative fade/interfade durations. In the case of DVB-SH, many simulations have been performed using the time series generated with the well-known Perez-Fontan LMS model [16]. Clause A.12 presents the results obtained for different environments, speeds and fade margins. These results give an indication of the achievable ESR5(20) criteria in different environments for a given LOS C/N. The required margin for satellite reception is heavily dependent on the satellite elevation angle, the location, the type of environment, the mobile terminal speed and on the receiver physical and link layer configuration (overall redundancy, length and configuration of the physical layer interleaver and/or link layer FEC, etc.). Therefore, the tables in Clause A.12 present a set of representative DVB-SH cases but cannot be considered to have general or universal applicability. Clause A.12 simulations are organized by environments that are considered representative of DVB-SH usage (Intermediate Tree Shadowing, Suburban, etc.). Network planners need to take into consideration the percentage of time where a DVB-SH user is under each of the analyzed environments in order to derive the final availability. 11.7.4 Antenna diversity Gain The use of antenna diversity in the receiver can improve the link budget and allow to counteract multipath fading. In combination with Maximum Ratio Combining, the C/N gain of the antenna diversity with 2 antennas has been estimated between 3 to 5 dB both in laboratory tests and in on field trials. 11.7.5 Link budget examples This clause presents several examples of satellite link budgets calculated following the formulae given above. It also presents the available fade margin for different physical layer configurations. The selected link budget parameters are chosen to be representative of a DVB-SH GEO satellite system, although the link budget parameters (EIRP and interference contribution, etc.) always depend on the specific system configuration. For these link budget calculations two satellite EIRP values have been selected: 63 dBW and 68 dBW which are representative of low and medium power satellites. It has also been considered an overall downlink C/I component of 14 dB, where the interference contribution comes from the inter-modulation noise and from the co-channel intra system interference. Both interference components are affected by fade in the same amount as the signal and therefore the distinction between fade and link margin defined in clause 11.7.2.2 Satellite fade margin applies. In the following link budgets, there are four columns representative of the four different categories of terminals defined in clause 10. Main terminal characteristics (polarization, antenna gain, noise figure) are extracted from clause 10. 11.7.5.1 Link budget for SH-A Table 11.17 represents the link budget for SH-A configuration in the case of a satellite with an effective EIRP towards the analyzed beam of 63 dBW and a 5 MHz channel. Table 11.18 shows the same link budget but considering antenna diversity of order 2 in the receiver. ETSI 281 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.17: SH-A link budget for 5 MHz channel and 63 dBW EIRP SH-A, 5 MHZ channel Portable Handheld category Portable Vehicular Unit category 2b category category 3 (see note) 2a 1 OFDM noise bandwidth MHz 4,76 4,76 4,76 4,76 Up-link C/(N+I) dB 19,5 19,5 19,5 19,5 Satellite transmission Tx frequency GHz 2,2 2,2 2,2 2,2 EIRP effective/beam dBW 63,0 63,0 63,0 63,0 Sat to RX Terminal propagation Free Space Loss dB 190,7 190,7 190,7 190,7 Atmospheric attenuation dB 0,5 0,5 0,5 0,5 Total attenuation dB 191,2 191,2 191,2 191,2 Terminal Rx reception Terminal G/T dB/K -32,1 -29,1 -24,9 -21,0 Polarization losses dB 3,0 3,0 3,0 0,0 Downlink Results C/N down-link dB -1,8 1,2 5,5 12,3 C/I down-link total dB 14,0 14,0 14,0 14,0 Down-link C/(N+I) dB -1,9 1,0 4,9 10,1 Total C/(N+I) dB -1,9 0,9 4,7 9,6 NOTE: The results of this link budget are also representative of a handheld terminals with circular polarization. Table 11.18: SH-A link budget for 5 MHz channel and 63 dBW EIRP and antenna diversity DVB-SH-A, 5 MHZ channel Handheld Handheld Portable Vehicular Unit Category 3 category 2b category 2a category 1 OFDM noise bandwidth MHz 4.76 4.76 4.76 4.76 Up-link C/(N+I) dB 19.5 19.5 19.5 19.5 Satellite transmission Tx frequency GHz 2.2 2.2 2.2 2.2 OBO dB 2.0 2.0 2.0 2.0 EIRP effective/beam dBW 63.0 63.0 63.0 63.0 Sat to RX Terminal propagation Free Space Loss dB 190.7 190.7 190.7 190.7 Atmospheric attenuation dB 0.5 0.5 0.5 0.5 Total attenuation dB 191.2 191.2 191.2 191.2 Terminal Rx reception Terminal G/T dB/K -32.1 -29.1 -24.9 -21.0 Polarisation losses dB 3 3 3 0 Antenna Diversity Gain dB 3 3 3 3 Downlink Results C/N down-link dB 1.2 4.2 8.4 15.3 C/I down-link total dB 14.0 14.0 14.0 14.0 Down-link C/(N+I) dB 1.0 3.8 7.4 11.6 Total C/(N+I) dB 0.9 3.6 7.1 10.9 This link budget can be closed for different physical layer configurations, given different fade margins. In order to close the link, the required C/N and implementation margin values shown in clauses 7.2.2.6 and 10.4.3 respectively are chosen. Tables 11.19 to 11.21 show the link budget closure for three different physical layer configuration (Code rate 1/2, 1/3 and 1/5) for the single antenna case. ETSI 282 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.19: Link budget closure for code rate 1/2 at physical layer and 63 dBW EIRP Handheld Portable Portable Vehicular category category category category Physical Layer Unit 3 2b 2a 1 CODE 1/2 at PL DVB-SH mode 2K 2K 2K 2K Number of carriers 1 705 1 705 1 705 1 705 Number of useful carriers 1 512 1 512 1 512 1 512 Useful symbol duration msec 358 358 358 358 Guard Time fraction ¼ ¼ ¼ ¼ Modulation Order QPSK QPSK QPSK QPSK Physical Layer Coding rate ½ ½ ½ ½ Useful bit rate at physical layer Mb/s 3,38 3,38 3,38 3,38 OFDM noise bandwidth MHz 4,76 4,76 4,76 4,76 Required C/N at physical layer at BER 10^-5 dB 1,4 1,4 1,4 1,4 Implementation Loss in AWGN dB 1,1 1,1 1,1 1,1 Received C/(N+I) dB -1,9 0,9 4,7 9,6 LOS Margin at Physical Layer w.r.t. AWGN dB -4,4 -1,6 2,2 7,1 Available Fade Margin at Physical Layer dB -4,7 -1,7 2,5 9,4 Table 11.20: Link budget closure for code rate 1/3 at physical layer Handheld Portable Portable Vehicular category category category category Physical Layer Unit 3 2b 2a 1 CODE 1/3 at PL DVB-SH mode 2K 2K 2K 2K Number of carriers 1 705 1 705 1 705 1 705 Number of useful carriers 1 512 1 512 1 512 1 512 Useful symbol duration msec 358 358 358 358 Guard Time fraction ¼ ¼ ¼ ¼ Modulation Order QPSK QPSK QPSK QPSK Physical Layer Coding rate 1/3 1/3 1/3 1/3 Useful bit rate at physical layer Mb/s 2,25 2,25 2,25 2,25 OFDM noise bandwidth MHz 4,76 4,76 4,76 4,76 Required C/N at physical layer at BER 10^-5 dB -0,9 -0,9 -0,9 -0,9 Implementation Loss in AWGN dB 1,1 1,1 1,1 1,1 Received C/(N+I) dB -1,9 0,9 4,7 9,6 LOS Margin at Physical Layer w.r.t. AWGN dB -2,12 0,73 4,54 9,40 Available Fade Margin at Physical Layer dB -2,22 0,78 5,02 11,87 ETSI 283 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.21: Link budget closure for code rate 1/5 at physical layer Handheld Portable Portable Vehicular category category category category Physical Layer Unit 3 2b 2a 1 CODE 1/5 at PL DVB-SH mode 2K 2K 2K 2K Number of carriers 1 705 1 705 1 705 1 705 Number of useful carriers 1 512 1 512 1 512 1 512 Useful symbol duration msec 358 358 358 358 Guard Time fraction ¼ ¼ ¼ ¼ Modulation Order QPSK QPSK QPSK QPSK Physical Layer Coding rate 1/5 1/5 1/5 1/5 Useful bit rate at physical layer Mb/s 1,35 1,35 1,35 1,35 OFDM noise bandwidth MHz 4,76 4,76 4,76 4,76 Required C/N at physical layer at BER 10^-5 dB -3,6 -3,6 -3,6 -3,6 Implementation Loss in AWGN dB 1,1 1,1 1,1 1,1 Received C/(N+I) dB -1,9 0,9 4,7 9,6 LOS Margin at Physical Layer w.r.t. AWGN dB 0,58 3,43 7,24 12,10 Available Fade Margin at Physical Layer dB 0,60 3,59 7,83 14,68 According to previous link budgets, only code rate 1/5 allows closing the link for the handheld terminals. With antenna diversity, the link budget can be closed for all terminals with code rate 1/3 as seen in Table 11.22. In the most performing terminals, the antenna diversity provides also a clear advantage in terms of fading margin. It is worth to note that antenna also counteracts the multipath fade (if antennas are sufficiently separated) therefore the gain assumed in this link budgets is clearly conservative. Table 11.22: Link budget closure for code rate 1/3 at physical layer Handheld Handheld Portable Vehicular Physical Layer Category 3 category 2b category 2a category 1 CODE 1/3 at PL DVB-SH mode 2K 2K 2K 2K Number of carriers 1705 1705 1705 1705 Number of useful carriers 1512 1512 1512 1512 Useful symbol duration msec 358 358 358 358 Guard Time fraction 1/4 1/4 1/4 1/4 Modulation Order QPSK QPSK QPSK QPSK Physical Layer Coding rate 1/3 1/3 1/3 1/3 Useful bit rate at physical layer Mb/s 2.25 2.25 2.25 2.25 OFDM noise bandwidth MHz 4.76 4.76 4.76 4.76 Required C/N at physical layer at BER 10^-5 dB -0.9 -0.9 -0.9 -0.9 Implementation Loss in AWGN dB 1.1 1.1 1.1 1.1 Received C/N dB 0.9 3.6 7.1 10.9 LOS Margin at Physical Layer wrt AWGN dB 0.7 3.4 6.9 10.7 Available Fade Margin at Physical Layer dB 0.8 3.8 8.0 14.8 In the same way, the tables 11.23 to 11.26 apply for the a medium power satellite radiating 68 dBW towards the beam. ETSI 284 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.23: SH-A link budget for 5 MHz channel and 68 dBW EIRP DVB-SH-A, 5 MHZ channel Handheld Portable Portable Vehicular Physical Layer Unit category category category category 3 2b 2a 1 FDM noise bandwidth MHz 4,76 4,76 4,76 4,76 Up-link C/(N+I) dB 19,5 19,5 19,5 19,5 Satellite transmission Tx frequency GHz 2,2 2,2 2,2 2,2 EIRP effective/beam dBW 68,0 68,0 68,0 68,0 Sat to RX Terminal propagation Free Space Loss dB 190,7 190,7 190,7 190,7 Atmospheric attenuation dB 0,5 0,5 0,5 0,5 Total attenuation dB 191,2 191,2 191,2 191,2 Terminal Rx reception Terminal G/T dB/K -32,1 -29,1 -24,9 -21,0 Polarization losses dB 3,0 3,0 3,0 0,0 Downlink Results C/N down-link dB 3,2 6,2 10,4 17,3 C/I down-link total dB 14,0 14,0 14,0 14,0 Down-lin k C/(N+I) dB 2,8 5,5 8,8 12,3 Total C/(N+I) dB 2,8 5,4 8,5 11,6 Table 11.24: Link budget closure for code rate 1/2 at physical layer Handheld Portable Portable Vehicular category category category category Physical Layer Unit 3 2b 2a 1 CODE 1/2 at PL DVB-SH mode 2K 2K 2K 2K Number of carriers 1 705 1 705 1 705 1 705 Number of useful carriers 1 512 1 512 1 512 1 512 Useful symbol duration msec 358 358 358 358 Guard Time fraction ¼ ¼ ¼ ¼ Modulation Order QPSK QPSK QPSK QPSK Physical Layer Coding rate ½ ½ ½ ½ Useful bit rate at physical layer Mb/s 3,38 3,38 3,38 3,38 OFDM noise bandwidth MHz 4,76 4,76 4,76 4,76 Required C/N at physical layer at BER 10^- 5 dB 1,4 1,4 1,4 1,4 Implementation Loss in AWGN dB 1,1 1,1 1,1 1,1 Received C/N dB 2,76 5,35 8,49 11,57 LOS Margin at Physical Layer w.r.t. AWGN dB 0,3 2,9 6,0 9,1 Available Fade Margin at Physical Layer dB 0,3 3,3 7,5 14,4 ETSI 285 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.25: Link budget closure for code rate 1/3 at physical layer Handheld Portable Portable Vehicular category category category category Physical Layer Unit 3 2b 2a 1 CODE 1/3 at PL DVB-SH mode 2K 2K 2K 2K Number of carriers 1 705 1 705 1 705 1 705 Number of useful carriers 1 512 1 512 1 512 1 512 Useful symbol duration msec 358 358 358 358 Guard Time fraction ¼ ¼ ¼ ¼ Modulation Order QPSK QPSK QPSK QPSK Physical Layer Coding rate 1/3 1/3 1/3 1/3 Useful bit rate at physical layer Mb/s 2,25 2,25 2,25 2,25 OFDM noise bandwidth MHz 4,76 4,76 4,76 4,76 Required C/N at physical layer at BER 10^- 5 dB -0,9 -0,9 -0,9 -0,9 Implementation Loss in AWGN dB 1,1 1,1 1,1 1,1 Received C/N dB 2,76 5,35 8,49 11,57 LOS Margin at Physical Layer w.r.t. AWGN dB 2,56 5,15 8,29 11,37 Available Fade Margin at Physical Layer dB 2,76 5,75 9,99 16,84 Table 11.26: Link budget closure for code rate 1/5 at physical layer Handheld Portable Portable Vehicular category category category category Physical Layer Unit 3 2b 2a 1 CODE 1/5 at PL DVB-SH mode 2K 2K 2K 2K Number of carriers 1 705 1 705 1 705 1 705 Number of useful carriers 1 512 1 512 1 512 1 512 Useful symbol duration msec 358 358 358 358 Guard Time fraction ¼ ¼ ¼ ¼ Modulation Order QPSK QPSK QPSK QPSK Physical Layer Coding rate 1/5 1/5 1/5 1/5 Useful bit rate at physical layer Mb/s 1,35 1,35 1,35 1,35 OFDM noise bandwidth MHz 4,76 4,76 4,76 4,76 Required C/N at physical layer at BER 10^- 5 dB -3,6 -3,6 -3,6 -3,6 Implementation Loss in AWGN dB 1,1 1,1 1,1 1,1 Received C/N dB 2,76 5,35 8,49 11,57 LOS Margin at Physical Layer w.r.t. AWGN dB 5,26 7,85 10,99 14,07 Available Fade Margin at Physical Layer dB 5,57 8,56 12,80 19,65 ETSI 286 ETSI TS 102 584 V1.1.2 (2010-XY) 11.7.5.2 Link budget for SH-B Table 11.27 represents the link budget for SH-B configuration in the case of a satellite with an effective EIRP towards the analyzed beam of 63 dBW and a 5 MHz channel. Table 11.27: SH-B link budget for 5 MHz channel and 63 dBW EIRP DVB-SH-B, 5 MHZ channel Handheld Portable Portable Vehicular Unit category category category category 3 2b 2a 1 TDM occupied bandwidth MHz 4,89 4,89 4,89 4,89 Up-link C/(N+I) dB 20,0 20,0 20,0 20,0 Satellite transmission Tx frequency GHz 2,2 2,2 2,2 2,2 EIRP effective/beam dBW 63,0 63,0 63,0 63,0 Sat to RX Terminal propagation Free Space Loss dB 190,7 190,7 190,7 190,7 Atmospheric attenuation dB 0,5 0,5 0,5 0,5 Total attenuation dB 191,2 191,2 191,2 191,2 Terminal Rx reception Terminal G/T dB/K -32,1 -29,1 -24,9 -21,0 Polarization losses dB 3,0 3,0 3,0 0,0 Downlink Results C/N down-link dB -1,3 1,7 5,9 12,8 C/I down-link total (adjacent, co-channel, NPR, etc.) dB 14,0 14,0 14,0 14,0 Down-link C/(N+I) dB -1,4 1,5 5,3 10,3 Total C/(N+I) dB -1,4 1,4 5,2 9,9 As before, this link budget can be closed for different physical layer configurations. Tables 11.28 to 11.30 show the link budget closure for three different physical layer configuration (Code rate 1/2, 1/3 and 1/5) according to the thresholds and implementation margins shown in clause 7 for the TDM case. Table 11.28: SH-B link budget closure for code rate 1/2 at physical layer and 63 dBW EIRP Physical Layer Handheld Portable Portable Vehicular category category category category 1 3 2b 2a CODE 1/2 at PL TDM Roll-off - 0,15 0,15 0,15 0,15 TDM Symbol rate - 4,25 4,25 4,25 4,25 Modulation - QPSK QPSK QPSK QPSK Physical Layer Coding rate - 1/2 1/2 1/2 1/2 Useful bit rate at physical layer (minus 8 % of pilots) Mb/s 3,91 3,91 3,91 3,91 TDM noise bandwidth MHz 4,25 4,25 4,25 4,25 Required C/N at physical layer at BER 10^-5 in dB 1,1 1,1 1,1 1,1 AWGN Implementation Loss in AWGN channel dB 0,5 0,5 0,5 0,5 Received C/N dB -1,4 1,4 5,2 9,9 Link budget results LOS Margin at Physical Layer w.r.t. AWGN dB -3,0 -0,2 3,6 8,3 Available Fade Margin at Physical Layer dB -3,2 -0,2 4,0 10,9 ETSI 287 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.29: SH-B link budget closure for code rate 1/3 at physical layer and 63 dBW EIRP Physical Layer Handheld Portable Portable Vehicular category category category category 3 2b 2a 1 CODE 1/3 at PL TDM Roll-off - 0,15 0,15 0,15 0,15 TDM Symbol rate - 4,25 4,25 4,25 4,25 Modulation - QPSK QPSK QPSK QPSK Physical Layer Coding rate - 1/3 1/3 1/3 1/3 Useful bit rate at physical layer (minus 8 % of Mb/s 2,61 2,61 2,61 2,61 pilots) TDM noise bandwidth MHz 4,25 4,25 4,25 4,25 Required C/N at physical layer at BER 10^-5 in dB -1,2 -1,2 -1,2 -1,2 AWGN Implementation Loss in AWGN channel dB 0,5 0,5 0,5 0,5 Received C/N dB -1,4 1,4 5,2 9,9 Link budget results LOS Margin at Physical Layer w.r.t. AWGN dB -0,7 2,1 5,9 10,6 Available Fade Margin at Physical Layer dB -0,8 2,2 6,5 13,3 Table 11.30: SH-B link budget closure for code rate 1/5 at physical layer and 63 dBW EIRP Physical Layer Handheld Portable Portable Vehicular category category category category 3 2b 2a 1 CODE 1/5 at PL TDM Roll-off - 0,15 0,15 0,15 0,15 TDM Symbol rate - 4,25 4,25 4,25 4,25 Modulation - QPSK QPSK QPSK QPSK Physical Layer Coding rate - 1/5 1/5 1/5 1/5 Useful bit rate at physical layer (minus 8 % of Mb/s 1,56 1,56 1,56 1,56 pilots) TDM noise bandwidth MHz 4,25 4,25 4,25 4,25 Required C/N at physical layer at BER 10^-5 in dB -3,9 -3,9 -3,9 -3,9 AWGN Implementation Loss in AWGN channel dB 0,5 0,5 0,5 0,5 Received C/N dB -1,4 1,4 5,2 9,9 Link budget results LOS Margin at Physical Layer w.r.t. AWGN dB 2,0 4,8 8,6 13,3 Available Fade Margin at Physical Layer dB 2,0 5,0 9,2 16,1 The same approach can also be repeated for a satellite radiating 68 dBW per beam, obtaining 5 dB more margin for the same physical layer considered previously. As in SHA link budget example, same calculation regarding the antenna diversity gain can also be applied in this case. 11.7.5.3 Implementation losses and link budgets In all previous link budgets tables, the used implementation losses are the typical values presented in Table 10.1 of Chapter 10. The same table presents “Possible Receiver Implementation Losses” that are 0,25 to 0,6 dB below the typical values. For instance, in OFDM QPSK, implementation losses are 0,5 dB instead of 1,1 dB for typical receivers. The use of these implementation losses would increase consequently the obtained fade margins. 11.7.5.4 Example of availability calculation for reception condition A The fade margin derived in the above link budget can be used to calculate the availability obtained for portable category of reception in a given area conditioned to the multipath factor over that area. Tables 11.31 and 11.32 show the availability obtained for the different physical layer configurations and for the different terminal types. The availability results are derived by matching the fade margin obtained in each of the link budgets with the fade margin required for a given multipath factor shown in figure 11.2. ETSI 288 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.31: Availability for reception condition A and 63 dBW of EIRP EIRP=63 dBW QPSK 1/2 QPSK 1/3 QPSK 1/5 SH-A Fade margin Availability Fade margin Availability Fade margin Availability available available available -4,7 dB Out of -2,2 dB Out of 0,6 dB 70 % (if Coverage Coverage K>15 dB) Terminal Category 2b -1,7 dB Out of 0,8 dB 70 % 3,6 dB 80 % (see note) Coverage (K>12 dB) 90 %(if K>10 dB) Terminal Category 2a 2,5 dB 80 % 5,0 dB 90 % 7,8 dB 95 % 90 % (if 95 % (if 99 % (if K>10 dB) K>5 dB) K>6 dB) NOTE: Also representative of terminal category 3 (handheld) with circular polarized antenna. Table 11.32: Availability for reception condition A and 68 dBW of EIRP EIRP=68 dBW QPSK 1/2 QPSK 1/3 QPSK 1/5 SH-A Fade margin Availability Fade margin Availability Fade margin Availability available available available Terminal Category 3 0,3 dB Coverage 2,8 dB 80 % 5,6 dB 90 % only in AWGN 90 % (if 95 % (if K>10 dB) K>6 dB) Terminal Category 2b 3,3 dB 80 % 5,75 dB 90 % 8,6 dB 95 % 90 % 95 %(K>5 dB) 99 %(if (K>6 dB) K>5 dB) Terminal Category 2a 7,5 dB 95 % 10 dB 95 % 12,8 dB 95 % 99 % (if 99 % (if 99 % (if K>7 dB) K>4 dB) K>3 dB) 11.7.5.5 Example of availability calculation for reception condition C The link budgets shown above have been used in the physical layer simulations presented in Clause A. These simulations give quality in terms of ESR5(20) obtained in a set of scenarios representative of DVB-SH satellite to mobile channel. Tables 11.33 to 11.36 summarizes the percentage of ESR5(20) fulfilment for the Intermediate Tree Shadowing (ITS) and SubUrban (SU) environment. The terminal considered is a vehicular terminal (category 1) with a long physical layer interleaver. Table 11.33: ESR5(20) fulfilment for SH-A for the category 1 (vehicular) terminal in reception condition C SH-A Physical Layer Long physical layer interleaver QPSK 1/3 16QAM 1/5 63 dBW EIRP Available Fade Margin = 11,9 dB Available Fade Margin = 9,3 dB Terminal category 1 (Vehicular) (After implementation losses) (After implementation losses) Uniform Uniform-Late Uniform Uniform-Late Mobile ITS 50 kmph 100 % 99 % 95,5 % 86 % Channel SU 50 kmph 100 % 100 % 100 % 99 % ETSI 289 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.34: ESR5(20) fulfilment for SH-A in reception condition A with terminal category 2b with circular polarized antenna SH-A Physical Layer Long physical layer interleaver QPSK 1/3 16QAM 1/5 68 dBW EIRP Available Fade Margin = 5,0 dB Available Fade Margin = 2,2 dB Terminal category 2b with linear (After implementation losses) (After implementation losses) polarized antenna and 3 with Uniform Uniform-Late Uniform Uniform-Late circular polarized antenna Mobile SU 3 kmph 84 % 75 % 51 % 42 % Channel Table 11.35: ESR5(20) fulfilment for SH-B for Category 1 (vehicular) terminal in reception condition C SH-B Physical Layer Long physical layer interleaver QPSK 1/3 8PSK 2/9 63 dBW EIRP Available Fade Margin = 13,3 dB Available Fade Margin = 11,2 dB Terminal category 1 (Vehicular) (After implementation losses) (After implementation losses) Uniform Uniform-Late Uniform Uniform-Late Mobile ITS 50 kmph 97 % 96 % 95,3 % 93,7 % Channel SU 50 kmph 98,2 % 97,3 % 99 % 96,8 % Table 11.36: ESR5(20) fulfilment for SH-B in reception condition A with terminal category 2b with circular polarized antenna SH-A Physical Layer Long physical layer interleaver QPSK 1/3 8PSK 2/9 68 dBW EIRP Available Fade Margin = 6,5 dB Available Fade Margin = 4,4 dB Terminal category 2b with linear (After implementation losses) (After implementation losses) polarized antenna and 3 with Uniform Uniform-Late Uniform Uniform-Late circular polarized antenna Mobile SU 3 kmph 77,9 % 72,6 % 75,2 % 73,5 % Channel 11.7.6 Methodology for Satellite Coverage Calculation for reception condition C The DVB-SH satellite coverage is a composite of several environments: urban, suburban, open area, heavy tress shadowing and intermediate tree shadowing. Each environment has distinctive propagation characteristics and therefore DVB-SH physical and link layer performances are expected to be different. Moreover the required quality in terms of ESR5(20) is also dependent on the propagation conditions. For terrestrial channels that can be characterized as Rayleigh channels the requirement is 99 % of ESR5(20) fulfilment; however, the satellite channel presents very different dynamics and therefore this requirement cannot be directly applied. At the present moment, we lack sufficient information to conclude on the required ESR5(20) fulfilment necessary to guarantee a good video quality via satellite. The calculations presented in this clause are based on the assumption that a 90 % ESR5(20) fulfilment is required, however the methodology is considered of general validity. ETSI 290 ETSI TS 102 584 V1.1.2 (2010-XY) In the following, a methodology to calculate the satellite coverage area served with a given quality is presented. This methodology has been developed and used in the framework of the ESA contract “role of satellite in 4G mobile networks” [Erreur ! Source du renvoi introuvable.]. To get satellite coverage figure over a wide area it is necessary to: • Classify the satellite coverage region according to the environment categories listed above. In order to do so, the topographic data provided by the Global Land Cover Facility (GLCF, see the reference [Erreur ! Source du renvoi introuvable.]) gives the type of land cover of each location under the coverage. It enables to attribute a specific user environment to every location . • Using the topographical data, associate a probability of occurrence pE (i ) ≤ 1, i = 1, 2,3 for each of the environments over the satellite coverage; • select typical mobile terminal speed for each environment; • select an ESR5(20) target fulfilment rate per environment FESR 5 (i ) [ ]min , i = 1, 2,3 . As explained earlier, this target fulfilment rate depends on the environment considered. For the calculations on the present document we will consider that for the satellite environments 90 % of ESR5(20) fulfilment is required, • define the system parameters required to compute the link budgets providing LOS C/N and C/I values as described in the following clause; • define the DVB-SH waveform parameters required to perform the simulations as described in Appendix A.3; • analyze the ESR5(20) fulfillment rate FESR 5 (i ), i = 1, 2,3 achieved in each of the environments given the selected system and waveform parameters (see Clause A.12 for typical results); • compute the satisfaction index χ (i ), i = 1, 2,3 for each satellite coverage environment defined as: χ (i) = { 1 if FESR 5 ( i ) ≥[ FESR 5 ( i )]min 0 if FESR 5 ( i ) <[ FESR 5 ( i )]min i = 1, 2,3 • compute the overall satellite coverage CSAT (in % as): 3 CSAT (%) = 100 pE (i ) χ (i ) ; i =1 where pE(i) is the probability of occurrence of each environment. Note that the above described methodology assumes that the satellite elevation angle is not changing in an appreciable way over the coverage region. In case of large satellite beams, as it is the case of a global beam covering a large part of the continent, the above described procedure must be extended by splitting the satellite coverage over regions of similar elevation angle. 11.7.7 Example of satellite coverage calculation Previous clauses have explained the link budget calculations and the margins required for different reception conditions and different terminals. It has to be noted that physical layer simulations refer to one specific environment (Intermediate tree shadowing or suburban) and characterized the quality of DVB-SH physical layer (with a specific code rate and interleaver length) under that environment. In this context, the environments are those defined by the LMS model used for the physical layer simulations [17][16]. In this example, we consider a satellite coverage area composed of 6 linguistic beams covering 8 service zones: Belgium, France, Germany, Italy, Poland, Portugal, Spain and UK. ETSI 291 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 11.4: Example of Satellite Coverage area . A world land cover database based on the measurements of the Advanced Very High Resolution Radiometer (AVHRR) from the NOAA polar-orbiting satellite is used to determined the distribution of the different environments under the satellite coverage. Here, the smallest resolution is used: ( 0.5’ x 0.5’), (around 1 km2 at NADIR). A 0.5’ x 0.5’ cell defines one location and is characterized by a homogeneous land cover. These database provides 14 classes of land environments. Table 11.37 provides a conservative correspondence matrix between the land cover classes of the AVHRR database and the LMS propagation environments used for the physical layer assessment. Table 11.37: land cover classes and LMS Perez-Fontan Classes. AVHRR land cover classes Pérez-Fontan classes Heavy Tree Evergreen needle leaf forest Shadowing (HTS) Evergreen broadleaf forest HTS Deciduous needle leaf forest HTS Deciduous broadleaf forest HTS Mixed forest HTS Intermediate Tree Woodland Shadowing (ITS) Wooded grassland ITS Closed shrub land Open Open shrub land Open Grassland Open Cropland Open Bare ground Open Urban and built-up Urban/suburban The following figure shows the percentage of coverage per environment considering the data from AVHRR and the classification done in Table 11.37 in each of the 8 service zones: ETSI 292 ETSI TS 102 584 V1.1.2 (2010-XY) Belgium France Germany Italy Poland Portugal Spain UK 100 90 80 Repartition of the environments [%] 70 60 50 40 30 HTS 20 ITS 10 OPEN SUB 0 Figure 11.5: Distribution of environments in the satellite coverage area. According to Figure 11.5, most of the coverage area can be characterised as open or Intermediate tree shadowing environment. Suburban and urban environments represent less than 4% of the area. However, the concentration of population in those areas make the analysis of the suburban environment relevant for the satellite. Contrary to that, the Heavy Tree Shadowing environment can represent up to 15% of the coverage in certain service zones, however, the population is likely to be rather scarce in those areas. The satellite antenna diagram in this example with a resolution of 0.2° x 0.2°) has been used to compute both the antenna gain and the carrier-to-interference ratio (C/I) due to the frequency re-use (see Figure 11.6 and Figure 11.7). Figure 11.6: Example of Satellite Antenna Gain over the coverage area ETSI 293 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 11.7: Example of Satellite Antenna C/I for a frequency reuse 1:2 Link budgets have been computed using the same resolution as the antenna diagram. The C/(N+I) values over the coverage obtained from the link budgets can be translated into service availability (ESR5) using the environmental data described earlier. The results of the link budgets are interpolated onto a given service zone with a 0.5’ x 0.5’ resolution and associated with the type of propagation environment. To each environment, type of modulation, interleaver length and code rate, there is a graph that links the C/N with the ESR5(20) quality (see Clause A.12). This process is depicted in Figure 11.8. 100 E SR5(20) fulfillment (%) 95 90 85 80 Satellite Coverage 75 6 7 8 9 10 11 12 13 14 C/N (dB) Urban SH-B (TDM) , Class 2 with Uniform Long Sub-Urban Interleaver Profile – QPSK1/3 ITS environment - 50 kmph HTS 100 ITS E SR5(20) fulfillment (%) 95 90 Open 85 80 75 0 2 4 6 8 10 12 14 C/N (dB) SH-B (TDM), Class 2: with Uniform Long Interleaver Profile – QPSK 1/3 Sub Urban environment - 50 kmph Figure 11.8: ESR5 versus C/N performance association for different environments over the coverage area ETSI 294 ETSI TS 102 584 V1.1.2 (2010-XY) The result of this process is a map that provides the Service availability over the satellite coverage for a given interleaver profile, code rate, modulation type and mobility situation (see Figure 11.9) Figure 11.9: Example of service availability over France The last stage requires the setting of a targeted service availability. In this example, the target service availability is given in terms of ESR5 and in this document is set to be 90% for all satellite environments. 11.8 Hybrid network planning 11.8.1 Introduction While classical Fixed TV broadcast either satellite-based or terrestrial-based addresses roof top service to directional antennas, Mobile TV broadcast addresses terminals with very reduced RF performances with regard to roof top antennas, located at street level (for outdoor services) or at various levels of floors with insuperable penetration losses for a satellite signal. Hence Global Mobile Broadcast solutions that rely on satellite umbrellas also require Complementary Ground Components to distribute the service to users in the urban/dense urban environments. Architectures will rely: • Either on the SH-B version of the DVB-SH standard, where satellite and terrestrial components use different modulations (TDM versus OFDM) and different frequency channels or even bands; satellite/terrestrial overlap zone is just another case of service management across the spectrum • Either on the SH-A MFN version of the DVB-SH standard, where satellite and terrestrial components use the same modulation (OFDM) but different frequency channels or even bands; satellite/terrestrial overlap zone is likewise just another case of service management across the spectrum • Either on the SH-A SFN version of the DVB-SH standard, where satellite and terrestrial components use the same modulation (OFDM) and the same frequency channel (and thus band); satellite/terrestrial overlap zones require co-engineering on both segments in order to maintain SFN on the whole hybrid coverage and avoid destructive interferences; this entails constraints on parameters such as Guard Interval for the CGC. ETSI 295 ETSI TS 102 584 V1.1.2 (2010-XY) It is not easy to derive the hybrid SFN gain as the satellite to terrestrial repeater power ration will rapidly evolve while the mobile user is moving from some rural or sparse suburban context at the theoretical terrestrial border to standard suburban or even some level of urban environment in more central zones. In fact, not only the SFN gain can be observed, which is beneficial for the QoS, but also some destructive interference can appear, due to the fact that the terrestrial receiver coverage outside the terrestrial SFN area may create interference with the satellite signal. This is because the terrestrial gap fillers signals may be seen in LOS propagation for oriented towards open fields, especially in the direction of the satellite (south) and the differential delay of terrestrial signal versus the satellite one may exceed the OFDM guard interval An other consequence of the hybrid architecture and frequency reuse in the case of multi beam satellite is the possible existence of interference on terrestrial coverage of a given country coming from adjacent country beam at the same frequency, due to beam secondary lobes. So the following subjects are covered in the different subsections. • Hybrid network system model • Impact on coverage : improvements and interferences • Interference limited zones in multibeam hybrid networks 11.8.2 Hybrid network system model 11.8.2.1 Overview The following picture gives an overview of the system architecture that must be considered in the following sections. Figure 11.10 : Overall System In the most general situation, the system is composed of the following elements under the DVB-SH satellite coverage. • One or more repeaters • A DVB-SH receiver, mainly vehicle mounted receiver ETSI 296 ETSI TS 102 584 V1.1.2 (2010-XY) Though some hybrid situation can occur in urban areas, the ideal locations for hybrid situation are suburban areas, where satellite signal can be received with a reasonable quality of service and which can be under one or many terrestrial repeaters. Of particular interest is the case where a repeater is covering the edge of the area, at the limit of rural area, open or ITS. For the terrestrial channel, we consider TU6 channel at 50 km/h (3 km/h?) in suburban environment, using Cost Hata model. For the satellite channel, three possible cases, based on Perez-Fontan LMS channel, can be used • Sub urban channel • Open channel: obviously when repeater is at the edge of an open area • Intermediate Tree Shadowing There are many differences when considering SFN (SHA) and MFN cases. In MFN case, the two signals are received in different frequency blocks (inside the same frequency band or not), and simultaneous reception (though ad hoc terminals are not yet existing) requires two receiving chains (two antennas, two tuners two demodulator but probably a single decoder). The two signals (OFDM/OFDM or TDM/OFDM) can be received without any stringent SFN conditions, and even outside the Hybrid SFN geometric zone, the signal from the repeater can be added to the satellite signal, provided it is received with a sufficient level to be detected and combined. In the case of SFN (hybrid SHA), the two signals are received through the same chain (two chains, but with the same content in case of receiving diversity), they are added intrinsically as in terrestrial SFN and decoded. 11.8.2.2 Hybrid channel model and Geometrical Aspects The hybrid channel can be novelised as the addition of the terrestrial signal following a typically TU6 channel model, and the satellite channel, following a LMS channel model The following picture, that represents the emulator block diagram, gives a schematic view of hybrid channel model. The terminal will receive the signal from the two sources: • The terrestrial repeater(s) following a TU6 propagation channel, or any repeaters • The satellite following a LMS propagation channel The signals are impaired differently, suffering from multipath, shadowing and blocking effects on the different path. Doppler effects will also be different. Receiver will perform an estimation of the right window to remove the CP, and then will use a channel estimator in order to demodulate. ETSI 297 ETSI TS 102 584 V1.1.2 (2010-XY) Satellite CGC Component AC/CC/UMTS ACI/CCI interference Application Satellite Satellite Mobile + + Modulator Channel Encapsulator S imulator Application CGC SFN Modulator Channel Encapsulator Figure 11.11: Possible lab configuration The geometrical Analysis intends to determine in the case of a single repeater the Hybrid SFN zone using a very simplified model. The schematics of the geometry is represented here below; Figure 11.12: Schematics of the geometrical analysis The repeater is called R1 and is considered as the time reference. The direction N/S is indicative, in fact, the considered direction is approximately SE/NW, if we are located around Las Vegas (115 °W), and satellite ICO G1 is at 91° W, and also around Paris with W2A satellite located at 10 ° E. The positive axis is towards north direction. And we make a rough computation along this axis. Satellite is seen under an elevation angle α. So if we consider a point M at distance x from the repeater, the satellite signal will arrive at the following relative time to the repeater: x cos(α)/c . If x is positive (north of the repeater), satellite signal will arrive later that at the repeater. In case M is south of the repeater, satellite signal will arrive in advance compared to the repeater. ETSI 298 ETSI TS 102 584 V1.1.2 (2010-XY) The origin of time is defined by the time of arrival of the satellite signal at the repeater, which is also the time of transmission of the repeater signal. This synchronisation can be realised thanks to the SHIP in the SH frame. Now, at point M, with these conventions x • The signal from the repeater arrives at time at point M c x cos(α ) • The signal from the satellite arrives at time c So the general hybrid SFN condition can be written as: x x cos(α ) . − ≤ GT c c Where GT is the guard time. Introducing do =c*GT So the Hybrid SFN condition can be written as x − x cos(α ) ≤ d o Two possible cases are therefore foreseen: x >0 and x<0, and the Hybrid SFN conditions ca be written do • x≤ when M is located north of the repeater 1 − cos(α ) do • x≤ when M is located south of the repeater 1 + cos(α ) Some examples are necessary to illustrate the previous equations. In all following cases, GT = 89.6 µs for GI =1/4 /2k/5 MHz Let us first consider the case of point located North of the repeater • α=33°: x max = 166 km (around Paris) • α =45°: x max = 91 km • α =41 °: x max = 109.59 km When M is located South, the SFN condition can be written as • α=33 ° : x max = 14.6 km • α =45 ° : x max = 15.75 km • α =41 ° : x max = 15.3 km ETSI 299 ETSI TS 102 584 V1.1.2 (2010-XY) These figures suggest that the hybrid SFN zone is an ellipse north oriented, as roughly described in next figure.    Figure 11.13 : One repeater plus satellite SFN coverage @ 41 ° elevation With GI 1/8 dimensions are roughly divided by two. The hybrid SFN zone si quite narrow in the southern part; 11.8.3 Impact on coverage 11.8.3.1 The Possible zones in the case of SHA Hybrid SFN In the case of SHA Hybrid SFN, and when considering a hybrid coverage zone there are three main areas with special scenarios. Nevertheless while transiting from one area to another the transition is smooth: • When approaching the transmitter, the terrestrial signal is much more powerful than the satellite one, and there is no impact on the reception quality • When satellite and terrestrial signals are of comparable strength, they can add until the mobile is out of the SFN zone defined by the Guard Interval. Leaving the overlap zone the terrestrial component fades out smooth. This situation is especially important when the system is working in DVB-SH-A-SFN, because the combining gain is maximum here. • When the mobile is out of the Hybrid SFN zone, the terrestrial signal can act as an interferer whose strength depends on terrestrial repeater distance, EIRP, antenna diagram, propagation channel. Again this case mainly concerns the pure SH-A-SFN case, as no (in-band) interference can occur in the MFN case • When terrestrial signal C/N vanishes ( below -10 dB ), there is no more interference, and we are in a satellite only coverage zone. It is in principle necessary to distinguish the SH-A-SFN case from the MFN case (SH-A or SH-B, which is quite the same). The difference between MFN and SFN is the type of interference, for SFN the interference is in-band (e.g. if the guard interval is violated), for MFN the interference is in another frequency band and can be mitigated by receiver filtering. Following picture shows a qualitative representation of the different zones. ETSI 300 ETSI TS 102 584 V1.1.2 (2010-XY)         Figure 11.14: Hybrid System Overview The different computations have shown that the dimensions of the ellipse are quite high in the north edge around 100 km with a repeater located in Paris, and a satellite at 10 ° East,and GI of 1/8. On the southern direction, distance is around 8 km; So except the southern direction, the geometrical SFN zone is quite huge, and the risk and level of interference are very low. This is why the different simulations and measurements have been concentrated on the critical cases. The different results are presented in Chapter 7 and Clause A.13 The main parameters to consider are the terrestrial received C/N and the time difference between the satellite signal time of arrival and the repeater signal time of arrival at the receiver. The time reference is defined as the time of arrival of the satellite signal at the repeater. This time difference is called  T. Summarizing some of different laboratory measurements, the following graph, illustrates the different zones ETSI 301 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 11.15: Illustration of the different zones in rural areas with 16 QAM 1/3 It is possible to distinguish the different zones • When received terrestrial signal C/N is above required threshold, we can consider that we are under terrestrial coverage, and the satellite signals provides some improvement, but is not essential. • The second zone (starting from right in the diagram) corresponds to a zone where satellite signal is improved by the terrestrial one. Even if the satellite signal is quite low (for instance if received by a handset, the combination of the two signals result in a good quality signal • The third zone corresponds to an improvement of the satellite signal with a “residual” terrestrial signal. • The fourth zone corresponds to possible interference zone between the satellite and terrestrial signals. The border between the two previous zones corresponds to a differential delay of around 160 % of the GI. • The last zone corresponds to satellite only signal as terrestrial signal is falling below -10 dB, but it can have some interference impact. In the case of suburban area, the interference zone is reduced as shown in the following figure. ETSI 302 ETSI TS 102 584 V1.1.2 (2010-XY) 16QAM : Hybrid Channel Suburbain 50 km/h Performances Overview 12 dB   10 dB        C/N for 90% availability  8 dB 6 dB 30m high tower 45m high tower 4 dB 60m high tower Satellite Only 2 dB 0 dB -10 dB -5 dB 0 dB 5 dB 10 dB C/N (Ter)    Figure 11.16: Illustration of the different zones in Sub Urban areas with 16 QAM 1/3 11.8.3.2 Hybrid coverage improvement The previous sections have shown that the interference can occur in the southern direction . The way to mitigate the interference is to increase the SFN zone : of course the first way is to use higher Guard Time, but this will not solve all the situations; An important feature that can be implemented is the timing advance • Usually, the time of transmission of a repeater is aligned with the time of arrival of the satellite signal at the repeater • In the timing advance mechanism, time od transmission is advance by a integer multiple of the use Guard Time • By doing so, the dimensions of the ellipses are multiplied by the same integer number The following pictures provide plots of hybrid SFN zones using W2A satellite at 10 °E,and one repeater in two different locations: Paris and Frascati.,(ESRIN) 5 MHz and GI 1/8 (44,6 µs) are used. The figures plots the hybrid SFN contours with no time advance, one GI advance and 8 GI advance. .So the contours represent maximum time difference of one GI, two and nine GI. ETSI 303 ETSI TS 102 584 V1.1.2 (2010-XY)      Figure 11.17: Hybrid SFN contour lines in Paris And at ESRIN ETSI 304 ETSI TS 102 584 V1.1.2 (2010-XY)     Figure 11.18: Hybrid SFN contour lines in Frascati The different plots show clearly that the DFN zone is improved by using timing advance ; 11.8.3.3 Combining gain Some conclusions and rules can be derived from the previous graphs; • The hybrid channel architecture gives an evident gain of quality of services, as it increases in most of the cases the ESR (5) ratio. • In the northern direction, the differential delay is below 100 % of the GI on along distance, even for GI 1/8, and there is no risk of interference, as the terrestrial signal becomes very low at the edges. • On the southern direction (in sight of the satellite), the gain is noticeable, even at more than 100 % of the GI. • When the delay is more than 150 % of the GI approximately , there can be some interference at te edges of the contour, but they can be mitigated by different means: using a large GI when possible, implement timing advance at repeaters level, or reducing radiated power towards southern direction; The repeater could use two sector antenna at the edge of the zone and oriented towards northern directions; ETSI 305 ETSI TS 102 584 V1.1.2 (2010-XY) Hence a hybrid network planning tool (whose design is of course out of the scope of this document) should in fact proceed as follows for the great lines • First compute the geometrical hybrid SFN zones taking into account the possible timing advance: roughly the zones where differential delay is below or over 150 % (TBC) of the GI • Then compute terrestrial only coverage with a given C/N threshold, and consider it as terrestrial only coverage • Then compute inside the geometrical SFN zone the possible gain in LOS of the satellite signal, by using next formula. In this case the QoS could be determines by using maps like in section 11.7.4, The reulst will give inputs that will give inputs for possible reduction of sites or power. C C C 0 ,1×  0 ,1×    = 10 × Log (10  Sat + 10  Ter ) in dB. N N  N  eq • Outside the geometrical zone, the tool can compute the impact on satellite C/N , and then the possible degradations I some areas, In that case the terrestrial C/N is seen as a C/I for the satellite signal,a nd the resulting C/N is now computed by C C C − 0 ,1×    N  Sat − 0 ,1×     = −10 × Log (10 + 10  N Ter ) in dB;  N eq 11.8.4 Interference limited zones in multibeam hybrid networks 11.8.4.1 Introduction In mutibeam satellites, there is the possibility of large overlaps between adjacent beams (see for example clause 4, figure 4.2). In satellite-only systems, a suitable intra-system C/I level is maintained by segmenting the overall bandwidth in colours and reusing each colour only in the beams where there is a good antenna isolation. However, in hybrid terrestrial/satellite systems, all the colours are used in each beam (either by the satellite or by the terrestrial component).Therefore, the terrestrial transmitter in a beam may be interfered by the satellite signal of the adjacent beam when it reuses the frequency of that adjacent beam. In the following, we will define the interference limited zone as the coverage area that would be served if the interference component was not present. In this analysis, the source of the interference is the satellite signal from the adjacent beams. The situation is depicted schematically in the following figure. ETSI 306 ETSI TS 102 584 V1.1.2 (2010-XY) Figure 11.19: Interference limited zone in Country A caused by satellite beam over country B This kind of interference limited zone impacts the local and the common content as follows: Case (a), for the Local content only: the interference limited zone is the fringe coverage area of the transmitter carrying the Local content that would be covered if the interfering satellite signal were removed. Case (b), for the Common content in an MFN (and only MFN): the fringe area of the transmitter carrying the Common content where the wanted satellite signal is not usable (e.g. blocked) and that would be covered if the interfering satellite signal were removed. In both cases, the result is a reduction of the cell radius covered by the CGC. The new cell radius can be calculated considering an increase in the noise floor (or equivalently a degradation of the receiver G/T) caused by the interference signal arriving from the satellite. The following link budget exemplifies the approach for a satellite EIRP of 63dBW. 11.8.4.2 Reception conditions B1 and B2: Indoor reception In the case of indoor reception, the satellite signal is heavily attenuated by the building penetration loss. Hereafter, the table shows the cell reduction caused by the satellite signal experienced in reception conditions B1 (light indoor) and B2 (deep indoor), for different environments: ETSI 307 ETSI TS 102 584 V1.1.2 (2010-XY) Table 11.38: Link budgets results in reception conditions B1 & B2 DVB-SSP in S band INDOOR NO INTERFERENCE INDOOR WITH INTERFERENCE Terrestrial link budget Dense Urban Urban Sub-urban Rural Dense Urban Urban Sub-urban Rural Radio Interface Parameters Unit Value Value Value Value Value Value Value Value Channel bandwidth MHz 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 Frequency MHz 2182.50 2183 2183 2183 2182.50 2183 2183 2183 Mode 2048.00 2048.00 2048.00 2048.00 2048.00 2048.00 2048.00 2048.00 Radio interface mod code QPSK1/3 QPSK1/3 QPSK1/3 QPSK1/3 QPSK1/3 QPSK1/3 QPSK1/3 QPSK1/3 Antenna type TRI TRI TRI TRI TRI TRI TRI TRI 1/Guard Interval 8 8 8 8 8 8 8 8 Guard interval 0.125 0.125 0.125 0.125 0.125 0.125 0.125 0.125 Total number of sub-carriers 1705.00 1705.00 1705.00 1705.00 1705.00 1705.00 1705.00 1705.00 Nber of data sub-carriers 1512.00 1512.00 1512.00 1512.00 1512.00 1512.00 1512.00 1512.00 Elementary period us 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 Tu duration us 358.40 358.40 358.40 358.40 358.40 358.40 358.40 358.40 GI duration us 44.80 44.80 44.80 44.80 44.80 44.80 44.80 44.80 Ts duration us 403.20 403.20 403.20 403.20 403.20 403.20 403.20 403.20 Sub-carrier spacing kHz 2.79 2.79 2.79 2.79 2.79 2.79 2.79 2.79 Symbol rate ksymb/s 2.480 2.480 2.480 2.480 2.480 2.480 2.480 2.480 Useful bandwidth occupancy MHz 4.75 4.75 4.75 4.75 4.75 4.75 4.75 4.75 Useful data rate at MPEG2-TS interface Mbit/s 2.50 2.50 2.50 2.50 0.00 0.00 0.00 0.00 Spectrum efficiency 0.50 0.50 0.50 0.50 0.00 0.00 0.00 0.00 Transmitting end Tx Tx Tx Tx Tx Tx Tx Tx Power amplifier per carrier and sector W 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 Tx Power at antenna input dBm 40.8 40.8 40.8 40.8 40.8 40.8 40.8 40.8 cable loss dB 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 Diplexer loss dB 0.7 0.7 0.7 0.7 0.7 0.7 0.7 0.7 Tx antenna gain dBi 18.0 18.0 18.0 18.0 18.0 18.0 18.0 18.0 EIRP dBm 55.1 55.1 55.1 55.1 55.1 55.1 55.1 55.1 EIRP W 323.0 323.0 323.0 323.0 323.0 323.0 323.0 323.0 Intereference source Satellite received power (LOS) dBm -1000 -1000 -1000 -1000 -107 -107 -107 -107 Building Penetration loss 18.0 16.0 14.0 12.0 18.0 16.0 14.0 12.0 Interference From satellite -1018.0 -1016.0 -1014.0 -1012.0 -125.0 -123.0 -121.0 -119.0 Receiving end Rx Rx Rx Rx Rx Rx Rx Rx Rx antenna gain dBi -3.0 -3.0 -3.0 -3.0 -3.0 -3.0 -3.0 -3.0 Polarization mismatch dB 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Noise figure dB 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 Antenna temperature K 290.0 290.0 290.0 290.0 290.0 290.0 290.0 290.0 Ambiant temperature K 290.0 290.0 290.0 290.0 290.0 290.0 290.0 290.0 kT dBm/Hz -174.0 -174.0 -174.0 -174.0 -174.0 -174.0 -174.0 -174.0 Equivalent Rx band dBm.Hz 66.8 66.8 66.8 66.8 66.8 66.8 66.8 66.8 G/T dB/K -32.1 -32.1 -32.1 -32.1 -32.1 -32.1 -32.1 -32.1 Rx noise floor dBm -102.7 -102.7 -102.7 -102.7 -102.7 -102.7 -102.6 -102.6 Required C/N dB 2.8 2.8 2.8 2.8 2.8 2.8 2.8 2.8 Rx sensitivity dBm -99.9 -99.9 -99.9 -99.9 -99.9 -99.9 -99.8 -99.8 Minimum Rx level at antenna dBm -96.9 -96.9 -96.9 -96.9 -96.9 -96.9 -96.8 -96.8 Minimum Signal Level for Network Planning dBm -72.01 -74.01 -71.31 -73.31 -71.98 -73.97 -71.24 -73.21 System Gain dB 152.0 152.0 152.0 152.0 152.0 152.0 151.9 151.9 Margins Average Hand Loss dB 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Average building penetration loss dB 18.0 16.0 14.0 12.0 18.0 16.0 14.0 12.0 Target level of signal penetration B2 B2 B1 B1 B2 B2 B1 B1 Std shadowing outdoor dB 8.00 8.00 8.00 8.00 8.00 8.00 8.00 8.00 Std shadowing indoor dB 6.00 6.00 6.00 6.00 6.00 6.00 6.00 6.00 Std dev of Fading Margin dB 10.00 10.00 10.00 10.00 10.00 10.00 10.00 10.00 Propagation constant 3.5 3.4 3.5 3.4 3.5 3.4 3.5 3.4 Shadow fading Margin - whole cell dB 11.6 11.6 11.6 11.6 11.6 11.6 11.6 11.6 Coverage Probability - whole Cell 95.0% 94.9% 95.0% 94.9% 95.0% 94.9% 95.0% 94.9% Coverage Probability - edge of cell % 87.7 87.7 87.7 87.7 87.7 87.7 87.7 87.7 SFN network gain dB 4.7 4.7 0.0 0.0 4.7 4.7 0.0 0.0 Rx gain (antenna diversity) dB 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 MAPL dB 127.1 129.1 126.4 128.4 127.1 129.1 126.3 128.3 Cell range computation Cost231-HATA model Cell range km 0.593 1.107 0.899 2.376 0.592 1.104 0.895 2.359 Surface km2 0.69 2.39 1.58 11.00 0.68 2.38 1.56 10.86 As expected, the impact of the satellite interference is negligible in the case of indoor propagation. ETSI 308 ETSI TS 102 584 V1.1.2 (2010-XY) 11.8.4.3 Reception conditions A: Outdoor pedestrian In the case of outdoor reception, the satellite signal is not impaired by the building penetration loss and the cell radius calculated considering the interference from the satellite signal homogenously present in the cell is around 8% for the handsets as seen in the following table . Table 11.39: Link budgets results in reception condition A DVB-SSP in S band OUTDOOR NO INTERFERENCE OUTDOOR WITH INTERFERENCE Terrestrial link budget Dense Urban Urban Sub-urban Rural Dense Urban Urban Sub-urban Rural Radio Interface Parameters Unit Value Value Value Value Value Value Value Value Channel bandwidth MHz 5.00 5.00 5.00 5.00 5.00 5.00 5.00 5.00 Frequency MHz 2182.50 2183 2183 2183 2182.50 2183 2183 2183 Mode 2048.00 2048.00 2048.00 2048.00 2048.00 2048.00 2048.00 2048.00 Radio interface mod code QPSK1/3 QPSK1/3 QPSK1/3 QPSK1/3 QPSK1/3 QPSK1/3 QPSK1/3 QPSK1/3 Antenna type TRI TRI TRI TRI TRI TRI TRI TRI 1/Guard Interval 8 8 8 8 8 8 8 8 Guard interval 0.125 0.125 0.125 0.125 0.125 0.125 0.125 0.125 Total number of sub-carriers 1705.00 1705.00 1705.00 1705.00 1705.00 1705.00 1705.00 1705.00 Nber of data sub-carriers 1512.00 1512.00 1512.00 1512.00 1512.00 1512.00 1512.00 1512.00 Elementary period us 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 Tu duration us 358.40 358.40 358.40 358.40 358.40 358.40 358.40 358.40 GI duration us 44.80 44.80 44.80 44.80 44.80 44.80 44.80 44.80 Ts duration us 403.20 403.20 403.20 403.20 403.20 403.20 403.20 403.20 Sub-carrier spacing kHz 2.79 2.79 2.79 2.79 2.79 2.79 2.79 2.79 Symbol rate ksymb/s 2.480 2.480 2.480 2.480 2.480 2.480 2.480 2.480 Useful bandwidth occupancy MHz 4.75 4.75 4.75 4.75 4.75 4.75 4.75 4.75 Useful data rate at MPEG2-TS interface Mbit/s 2.50 2.50 2.50 2.50 0.00 0.00 0.00 0.00 Spectrum efficiency 0.50 0.50 0.50 0.50 0.00 0.00 0.00 0.00 Transmitting end Tx Tx Tx Tx Tx Tx Tx Tx Power amplifier per carrier and sector W 12.0 12.0 12.0 12.0 12.0 12.0 12.0 12.0 Tx Power at antenna input dBm 40.8 40.8 40.8 40.8 40.8 40.8 40.8 40.8 cable loss dB 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 Diplexer loss dB 0.7 0.7 0.7 0.7 0.7 0.7 0.7 0.7 Tx antenna gain dBi 18.0 18.0 18.0 18.0 18.0 18.0 18.0 18.0 EIRP dBm 55.1 55.1 55.1 55.1 55.1 55.1 55.1 55.1 EIRP W 323.0 323.0 323.0 323.0 323.0 323.0 323.0 323.0 Intereference source Satellite received power (LOS) dBm -1000 -1000 -1000 -1000 -107 -107 -107 -107 Building Penetration loss 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Interference From satellite -1000.0 -1000.0 -1000.0 -1000.0 -107.0 -107.0 -107.0 -107.0 Receiving end Rx Rx Rx Rx Rx Rx Rx Rx Rx antenna gain dBi -3.0 -3.0 -3.0 -3.0 -3.0 -3.0 -3.0 -3.0 Polarization mismatch dB 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Noise figure dB 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 Antenna temperature K 290.0 290.0 290.0 290.0 290.0 290.0 290.0 290.0 Ambiant temperature K 290.0 290.0 290.0 290.0 290.0 290.0 290.0 290.0 kT dBm/Hz -174.0 -174.0 -174.0 -174.0 -174.0 -174.0 -174.0 -174.0 Equivalent Rx band dBm.Hz 66.8 66.8 66.8 66.8 66.8 66.8 66.8 66.8 G/T dB/K -32.1 -32.1 -32.1 -32.1 -32.1 -32.1 -32.1 -32.1 Rx noise floor dBm -102.7 -102.7 -102.7 -102.7 -101.3 -101.3 -101.3 -101.3 Required C/N dB 2.8 2.8 2.8 2.8 2.8 2.8 2.8 2.8 Rx sensitivity dBm -99.9 -99.9 -99.9 -99.9 -98.5 -98.5 -98.5 -98.5 Minimum Rx level at antenna dBm -96.9 -96.9 -96.9 -96.9 -95.5 -95.5 -95.5 -95.5 Minimum Signal Level for Network Planning dBm -92.91 -92.91 -88.21 -88.21 -91.53 -91.53 -86.83 -86.83 System Gain dB 152.0 152.0 152.0 152.0 150.6 150.6 150.6 150.6 Margins Average Hand Loss dB 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Average building penetration loss dB 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Target level of signal penetration A A A A A A A A Std shadowing outdoor dB 8.00 8.00 8.00 8.00 8.00 8.00 8.00 8.00 Std shadowing indoor dB 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Std dev of Fading Margin dB 8.00 8.00 8.00 8.00 8.00 8.00 8.00 8.00 Propagation constant 3.5 3.4 3.5 3.4 3.5 3.4 3.5 3.4 Shadow fading Margin - whole cell dB 8.7 8.7 8.7 8.7 8.7 8.7 8.7 8.7 Coverage Probability - whole Cell 95.0% 94.9% 95.0% 94.9% 95.0% 94.9% 95.0% 94.9% Coverage Probability - edge of cell % 86.2 86.2 86.2 86.2 86.2 86.2 86.2 86.2 SFN network gain dB 4.7 4.7 0.0 0.0 4.7 4.7 0.0 0.0 Rx gain (antenna diversity) dB 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 MAPL dB 148.0 148.0 143.3 143.3 146.6 146.6 141.9 141.9 Cell range computation Cost231-HATA model Cell range km 2.327 3.970 2.713 6.503 2.127 3.618 2.480 5.926 Surface km2 10.56 30.74 14.36 82.45 8.82 25.53 12.00 68.48 ETSI 309 ETSI TS 102 584 V1.1.2 (2010-XY) It is worth to notice that the above link budgets consider the satellite always in LOS. However, the probability of the satellite signal being shadowed or blocked depends on the environment. Therefore, for the same probability of coverage, the cell reduction caused by the satellite will be less significant in the dense urban and urban case than in the suburban and rural scenario. The link budgets above shall be considered as worst case calculations for all scenarios. The impact of the satellite signal can also be evaluated by assessing the loss of coverage probability caused by the satellite interference at constant cell radius. For the cell radius considered above, probability (in the whole coverage) goes down from 95% to 93.5% due to the satellite interference. 11.8.4.3 Conclusions When the terrestrial CGC is dimensioned for indoor reception, the link budgets show that the impact of the satellite signal in the cell coverage is negligible. If we call this indoor coverage cell radius Ro, for distances below Ro of the transmitter, the impact of satellite reception is also negligible for outdoor reception conditions (obtained cell radius is always larger than Ro even in the presence of interference). When only outdoor reception in considered, the cell radius of the terrestrial CGC will be reduced by the satellite interference particularly in sub-urban and rural areas. The impact of this cell reduction is relatively small for handsets at 95% of coverage probability. Cell radius could be kept constant at the expense of loosing 1.5% of coverage probability in the whole cell. More accurate characterization of the interference limited zones will require a description of the statistics of the satellite signal in the area of interest including the satellite antenna gain over that area. A.1 Geostationary Satellite Payload characteristics (S-band) A.1.1 Payload architectures The satellite payload distortions are of two types: linear distortions caused by the filters amplitude and group delay and the nonlinear distortions (see note) caused by TWTA(s) operated near its saturation point. The general analysis is complicated by the fact that linear and nonlinear distorting elements are typically cascaded. Nevertheless, the linear distortions are of second order compared to the nonlinear distortions. The target EIRP of a mobile TV satellite in S-band is in the order of 63 dBW to 72 dBW, with the 68 dBW being a possible limit between "medium" and "high" power satellites broadcasting to handheld devices. For comparison, the EIRP of satellites broadcasting to fixed reception in the Ku-band are in the region of 50 dBW to 55 dBW. Antenna and power amplification architectures play an equally central role whose final objective is the achievement of the target EIRP and the desired coverage areas. Antenna gain is upper-bounded by the theoretical limit relevant to the solid angular view of the coverage area. EIRP requirements and antenna performance limit lead in most cases to a demanding RF power that exceeds the generation capabilities of single amplifiers available for space applications and "power combining" must be adopted. Two main payload classes can be distinguished depending on the power amplification and combining architecture: "Linear" and "Non-linear" classes. For a single-beam satellite (or multibeam satellites that do not require flexible power-sharing between beams), the required EIRP is typically obtained by using parallel TWTAs through network or polarization combining. Each TWTA amplifies only one signal (but several TWTAs amplify the same signal). The TWTA can be driven close to saturation if needed. The satellite is then said to belong to the "non-linear" class, providing a power advantage for the TDM waveform, which does not require the same TWTA back-off as the OFDM waveform. When there is a need to reallocate available power to the different beams after the satellite is launched, a different amplification architecture is adopted. In this case, the amplification is performed through a low-power BFN followed by a shared TWTA network exploiting a stack of multiport amplifiers (MPA) connected to the transmit antenna feeds. Each TWTA now amplifies a weighted sum of several signals. This architecture implements spatial power combining that provides the flexible power allocation to the various beams. This flexibility is achieved at the expense of operating the TWTAs in multicarrier mode (i.e. with sufficient back-off) and therefore the power advantage of TDM over OFDM is diminished. ETSI 310 ETSI TS 102 584 V1.1.2 (2010-XY) NOTE: The multiplicative phase noise impact is discussed in Appendix. A.11.1.2. A.1.2 TWTA characteristics and Operating Point Optimization The satellite on-board platform DC power limitation calls for an TWTA(s) operation as close as possible to the maximum TWTA saturated power. At system level the TWTA is typically characterized by a maximum (saturated) power and three key performance characteristics: the AM/AM, AM/PM and DC to RF conversion efficiency curves. From a system perspective the quantity to minimize is the total payload degradation DTOT in dB given by: NL AWGN E  E  DTOT (IBO) =  s  (IBO) −  s  + OBO(IBO)-10log η (IBO)     N 0  req  N 0  req NL AWGN E  E  being  s  and  s  expressed in dB the average symbol energy over noise density required to achieve the  N 0  req  N 0  req target Frame Error Rate (FER) in the nonlinear and linear channel respectively, IBO, OBO represent the TWTA input and output backoff respectively in dB, while η represents the TWTA DC to RF conversion efficiency at the current drive level. It is important that the OBO and η are computed with the real DVB-SH signal at the selected drive level (IBO). It is apparent that an accurate payload loss assessment and TWTA operating optimisation involve quite complex physical layer simulations and accurate system modelling. Intuitively the IBO optimisation will result in the best trade- off between the loss of TWTA efficiency represented by the term OBO and -10log η (IBO)  which are growing   NL AWGN E  E  with IBO and the channel distortion losses represented by the term (in dB)  s  (IBO) −  s  which will  N 0  req  N 0  req reduce with IBO. For a DVB-SH type of system requiring large RF power (in the order of hundreds of Watts per beam) typically Travelling Wave Tubes (TWTA) are adopted. Figure A1.1Erreur ! Source du renvoi introuvable. shows typical AM/AM, and AM/PM characteristics for an S-band TWTA.                    ETSI 311 ETSI TS 102 584 V1.1.2 (2010-XY)                                     Figure A1.1: S-band TWTA typical AM/AM, and AM/PM and efficiency characteristics Over nonlinear channels the DVB-SH signal constellation, when observed at the demodulator symbol matched filter decimated output, is affected by two main kinds of impairment: 9) The constellation centroids1 warping due to the AM/AM and AM/PM TWTA nonlinear characteristic. 1 By centroid we consider the compilation of received constellation cluster center of mass conditioned to each constellation point. ETSI 312 ETSI TS 102 584 V1.1.2 (2010-XY) 10) The clustering effect due to the inter-symbol interference (ISI) experienced at the demodulator matched filter output. The warping phenomenon is responsible for the reduction of the distance among 16QAM points (SH-A) or 16APSK rings (SH-B) (AM/AM compression) as well as a differential phase rotation among them (AM/PM differential phase). The ISI causing the constellation clustering is related to the demodulator SRRC filter mismatch on the received signal due to the combination of the signal band limiting introducing memory in the channel, the IMUX filter linear distortion, the TWTA memory less nonlinearity, the OMUX linear filter distortions, resulting in a nonlinear channel with memory. These effects can be easily understood recalling the TWTA AM/AM and AM/PM characteristic of .Figure A1.1. Clearly the warping effect has no impact on single ring constellations such as QPSK and 8PSK2 .Degradations for 16QAM and 16APSK may be compensated in the receiver using pilots. A.1.3 Performance over nonlinear channels A.1.3.1 Simulations The reference demodulator as defined in Appendix A.11 has been used to derive via simulation the reported performances. In particular, the demodulator implements two-dimensional pilot-aided channel estimation, therefore, the impact of imperfect channel estimation is included. Tested physical layer configuration is QPSK-1/3 for both OFDM and TDM. The optimum IBO/OBO is derived via simulation as C/(N+I) link margin maximisation, according to the definitions in A.1.2. However, the total payload degradation DTOT was evaluated neglecting the TWTA DC to RF conversion efficiency. The formulae hereafter hold: NL AWGN E  E  DTOT (IBO) =  s  (IBO) −  s  + OBO(IBO)  N 0  req  N 0  req C+I C LossesIBO, ESR = − N IBO , ESR N ESR A.1.3.1.1 Single HPA per beam Figure A1.2 shows the total degradation DTOT as function of IBO when an AWGN channel is considered.The optimum IBO (OBO) has been found to be 2,3 dB (1,8 dB) for the SH-A configuration and 0,3 dB (0,7 dB) for the SH-B configuration. It is worth noticing that the DTOT curve is almost flat around the IBOOPT, resulting in a reduced degradation when a sub- optimum IBO is chosen. Also, the OBO variation is within 0,5 dB when the IBO is kept below 3 dB. Thus, IBO/OBO point could be adapted according to the design of the payload with little degradation in terms of end-to end performance. 2 The phase and amplitude PSK distortion will be compensated for by the demodulator AGC and phase recovery sub-systems. ETSI 313 ETSI TS 102 584 V1.1.2 (2010-XY) Target BER = 10-5 Target BER = 10-5 OFDM TDM 2.0 5.0 4.0 1.5 3.0 dB 1.0 dB 2.0 0.5 1.0 0.0 0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 IBO (dB) IBO (dB) OBO TOTAL DEG Loss OBO TOTAL DEG LOSS Figure A1.2: IBO/OBO optimization for single HPA per beam, AWGN channel: a) SH-A (a) and b) SH-B Figure A1.3 shows the in-band and out-of-band characteristics when a SH-A waveform is considered. The out-of-band signal emission characteristic in Figure A1.3b) has been evaluated through Power Spectrum Density (PSD).The in-band C/I values shown in Figure A1.3a), have been derived with a per FFT-bin correlation method. This method is applicable to any complex QAM signal and it is independent from the payload architecture. It requires alignment of transmitted and received symbol data sequences for a Data Aided Maximum Likelihood Estimation (DA- MLE). Two cases are reported: worst C/I case when the HPA is driven at saturation and when driven at optimum IBO as per Figure A1.2. In the latter case, the minimum C/I value (FFT center) decreases from the 9 dB when IBO= 0 dB to roughly 7.5 dB; it is clear that external FFT-bins suffer for less interference, then they are characterized by improved C/I. C/I per carrier Transmitted Power in-band and out-of-band 16 0 -5 12 -10 C/I(dB) -15 C/I(dB) -20 8 -25 -30 4 Carriers -35 IBO = 0.0dB IBO = 0.0dB IBO = 2.3 dB Figure A1.3: SH-A In-band and out-of-band characteristics: a) C/I per FFT-bin and b) out-of-band emission A.1.3.1.2 Multiport Amplifier This clause refers to the multiport amplifier (MPA) payload configuration as described in Clause A1.1. In detail an 8 by 8 MPA (multiple multiplex per HPA, as per Figure A1.4) followed by high power BFN has been considered. No mismatch has been introduced and the HPAs have been modeled assuming they have the same AM/AM and AM/PM characteristics shown in Figure A1.1.Erreur ! Source du renvoi introuvable. Also, the same input power to all amplifiers after the input hybrids has been considered. Performances have been verified for different level of MPA load, up to 3 active inputs over 8 available ports. The evaluation of the total payload degradation DTOT always refer to a single DVB-SH signal input, defined as target signal. ETSI 314 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A1.4: Multi-Port Amplifier configuration Figure A1.5 shows the DTOT as function of IBO when SH-A configuration in AWGN channel is considered. The optimum IBO (OBO) has been found to be 2,8 dB (1,9 dB) with 1 active input and 2,3 dB (1,8 dB) with 3 ports. It appears that SH-A OFDM MPA DTOT performance are practically the same as for single HPA. On the contrary, in case SH-B TDM configuration would be considered the DTOT performance would be degraded in case of MPA exploitation being the HPAs composing the MPA operating with multicarrier input even if the individual input ports are driven by a quasi-constant envelope TDM signal. In case of MPA the DTOT for TDM and OFDM is almost identical. It is worth noticing that the DTOT curve is almost flat around the IBOOPT point, resulting in a reduced degradation when a different IBO is chosen. Also, the OBO variation is within 0,5dB when the IBO is kept below 3 dB. Thus, IBO /OBO point could be adapted according to the design of the payload with little degradation in terms of end-to end performance. Target BER = 10-5 Target BER = 10-5 OFDM OFDM 6.0 6.0 5.0 5.0 4.0 4.0 dB 3.0 dB 3.0 2.0 2.0 1.0 1.0 0.0 0.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 IBO (dB) IBO (dB) OBO TOTAL DEG Loss OBO TOTAL DEG Loss Figure A1.5: IBO/OBO optimization for SH-A with different MPA loads: (a) 1 SH-A carrier and (b) 3 SH-A carriers A.1.3.3 Single HPA per beam – Laboratory Measurements The measurements reported in this clause have been obtained for a SH-A QPSK-1/3 configuration in AWGN channel with no co-channel interferer and for different IBO values. The target figure of merit is 95% ESR5 fulfillment. The values of OBO|IBO used for the calculation of the Csat/N are obtained by simulation, whereas (C+I)/N is measured with real equipment. Csat is defined as the relative satellite power at amplifier’s saturation point needed to close the link budget, thus: Csat C+I = + OBO IBO N IBO N IBO ETSI 315 ETSI TS 102 584 V1.1.2 (2010-XY) Non linearity results Non linearity results 120 120 100 100 AWGN AWGN gaussien gaussien 80 0 dB 80 0 dB ESR5 (%) 1 dB ESR 5% 1 dB 60 2 dB 60 2 dB 3 dB 3 dB 40 4 dB 4 dB 5 dB 40 5 dB 20 20 0 -1 -0,5 0 0,5 1 1,5 0 (C+I)/N (dB) -1 0 1 2 3 4 Csat/N Figure A1.6: ESR5 fulfillment performance for QPSK coding rate 1/3 in AWGN channel versus: a) (C+I)/N and b) Csat/N Non linearity losses at ESR 95% target 3,5 3 2,5 2 Total Losses [dB] Losses 1,5 OBO 1 0,5 0 0 1 2 3 4 5 IBO [dB] Figure A1.7: Measured Total Degradation as a function of the IBO Laboratory measurement results reported in Figure A1.6 and in Figure A1.7 confirm the simulation findings that a large possible IBO working zone with a reduced degradation when the IBO is kept between 0 and 3 dB (corresponding to a working zone of 1.4 o 1.68 dB for OBO). Moreover, IBO /OBO point could be adapted according to the design of the payload. ETSI 316 ETSI TS 102 584 V1.1.2 (2010-XY) A.2 Void A.3 Simulation Conditions A.3.1 Simulations with ideal channel estimator in demodulator A.3.1.1 System parameters The first focus is to optimize the interleaving scheme according to the channel impairments. The imperfect channel estimator in the demodulator is neglected at this stage. To limit the number of cases, the system parameters are those considered relevant to hybrid S-Band systems that are the first to make use of DVB-SH. Waveform spectrum efficiencies resulting in bitrates at MPEG-TS level ranging between 2,2 Mbps and 2,8 Mbps, for a 5 MHz channelization, are considered. The number of services chosen ranges between 8 and 11. A reference value of roughly 10 s has been considered for the end-to-end delay introduced by the DVB-SH processing. This allowance is to be shared between the physical layer and the LL-FEC, Turbo FEC delays being negligible (one Turbo FEC block is in the order of several msec). Different configurations selected are shown in tables A3.1 to A3.4. Direct comparison of power efficiency is possible when the bit rate is strictly the same, otherwise the bit rate should be kept in mind when comparing different cases. The interest to increase the modulation order while keeping the net bit rate is to take advantage of lower Turbo code rates that can bring higher channel erasure resilience. Also different sharing between PL and LL-FEC for class 1 receivers for the same overall spectral efficiency is investigated. System definitions for OFDM and TDM are reported in Tables A3.5 and A3.6 respectively. Finally, note that TDM and OFDM configurations are designed so that they can be synchronous in an SH-B system, so that code combining can be performed. Table A3.1: OFDM Reference configurations for Class 1 receivers (5 MHz) Parameter/Case name 16QAM_1/4_S 16QAM_2/7_S QPSK1/2_S QPSK2/3_S FFT Mode 2K+GI 1/4 2K+GI ¼ 2K+GI 1/4 2K+GI 1/4 Modulation 16 QAM 16 QAM QPSK QPSK PHY FEC rate 1/4 2/7 1/2 2/3 LL-FEC rate (recommended) 2/3 7/12 2/3 1/2 OFDM Symbols/ coded FEC 8,00 7,00 8,00 6,00 Services 8 8 8 8 Bit rate/service (at TS level) 279,8 kbps 273,6 kbps 279,8 kbps 277,7 kbps MPEG TS total bit rate 3,357 Mbps 3,752 Mbps 3,357 Mbps 4,443 Mbps Early interleaver duration 0 ms 0 ms 0 ms 0 ms Late interleaver duration 211 ms 211 ms 211 ms 211 ms ETSI 317 ETSI TS 102 584 V1.1.2 (2010-XY) Table A3,2: OFDM Reference configurations for Class 2 receivers (5 MHz) Parameter/Case name QPSK1/3_U QPSK_1/3_UL 16QAM_1/5_U 16QAM_1/5_UL FFT Mode 2K+GI 1/4 2K+GI 1/4 2K+GI 1/4 2K+GI 1/4 Modulation QPSK QPSK 16 QAM 16 QAM PHY FEC rate 1/3 1/3 1/5 1/5 LL-FEC rate (recommended) 1 1 1 1 OFDM Symbols/ coded FEC 12 12 10 10 # Services 8 8 9 9 Bit rate/service (at TS level) 277,7 kbps 277,7 kbps 296,2 kbps 296,2 kbps MPEG TS Total bit rate 2.222 Mbps 2.222 Mbps 2.666 Mbps 2.666 Mbps Early interleaver duration 11 000 ms 10 000 ms 11 000 ms 10 000 ms Late interleaver duration 0 ms 215 ms 0 ms 215 ms Table A.3.3: TDM Reference configurations for Class 1 receivers, compatible with OFDM QPSK (GI 1/4) Parameter/Case name T-8PSK1/3_S T-QPSK1/2_S Bandwidth (including R.O.) 4,89 MHz 4,89 MHz Modulation 8PSK QPSK PHY FEC rate 1/3 1/2 UL FEC rate (recommended) 2/3 2/3 CW per SH-FRAME 79 79 Services 9 9 Bit rate/service (at TS level) 288,9 kbps 288,9 kbps MPEG TS Total bit rate 3.900 Mbps 3.900 Mbps Early interleaver duration 0 ms 0 ms Late interleaver duration 190 ms 180 ms Table A3.4: TDM Reference configurations parameters for Class 2 receivers Parameter/Case name T-8PSK2/9_U T-8PSK2/9_UL T-QPSK1/3_U T-QPSK1/3_UL Bandwidth (including R.O.) 4,89 MHz 4,89 MHz 4,89 MHz 4,89 MHz Modulation 8PSK 8PSK QPSK QPSK PHY FEC rate 2/9 2/9 1/2 1/3 LL-FEC rate (recommended) Not used Not used Not used Not used CW per SH-FRAME 52 52 52 52 Services 9 9 9 9 Bit rate/Service (at TS level) 285,2 kbps 285,2 kbps 285,2 kbps 285,2 kbps MPEG TS Total bit rate 2.567 Mbps 2.567 Mbps 2.567 Mbps 2.567 Mbps Early interleaver duration 11 000 ms 10 000 ms 11 000 ms 10 000 ms Late interleaver duration 0 ms 180 ms 0 ms 180 ms ETSI 318 ETSI TS 102 584 V1.1.2 (2010-XY) Table A3.5: System definitions for OFDM simulations Definition System Parameter Comment Bandwidth definition Noise bandwidth 5-MHz Signal bandwidth Limited to modulated carriers (data + TPS + pilots) Pilots insertion Pilots as defined by DVB-T boosted by 16/9 Total Signal Power Total power of the received signal Normalized to 1 All Channels (TU6 inclusive) It Affects the Es/N0 calculation Definition of C/N ratio of useful power to noise power in the used bandwidth (= non-zero subcarriers) Definition of S/N According to "Bandwidth definition" S C (Nc−U + Nc−TPS + Nc−Pilot ) and "Total Signal Power" = . N N NFFT Definition of Es/N0 E S− U S N FFT ES-U accounts for the useful = . energy per symbol N0 N (N c − U + N c − TPS + 16 / 9.N c − Pilot ) N_FFT is the total number of carriers (FFT-size) Nc-U is the number of useful carrier Nc-TPS is the number of TPS Nc-Pilots is the number of pilots Table A3.6: System definitions for TDM simulations Definition System Parameter Comment Bandwidth definition Noise bandwidth = Symbol rate Square-root raised cosine Signal bandwidth = [Symbol rate]*1,15 Roll-off 15 % is proposed Pilots insertion 80 symbols every 1 008 data symbols Higher than DVB-S2: 36/1 440 Total Signal Power Total power of the received signal Normalized to 1 All Channels (TU6 inclusive) It Affects the Es/N0 calculation Definition of C/N ratio of useful power to noise power in the used bandwidth (=pilot power is taken into account) Definition of Es/N0 Ratio of Symbol energy to noise density To be compared with FEC simulations without pilot symbols Ratio C/N and Es/N0 C ES Pilots have the same power as data = symbols. So the power is conserved. N N0 A.3.1.2 Physical layer FEC and De-mapper configuration Table A3.7 details the algorithm and parameters chosen for the simulated decoder and de-mapper. Other implementations may use different methods and parameters that could lead to slightly different results. Table A3.7: Physical layer FEC configuration Criteria Simulation Parameters Comment Code Turbo Code 3GGP2 3GPP2 C.S0002-D, Version 1.0, Date: February 13, 2004 Information Block size 12 282 bits Decoder engine LOG-MAP Implementation Number of iterations 8 Constant De-mapper engine LOG-MAP as "Decoder engine Implementation implementation" - It Affects the LLR calculation ETSI 319 ETSI TS 102 584 V1.1.2 (2010-XY) A.3.2 Void A.3.3 Demodulator state machine principle A.3.3.1 States representation of a demodulator To emulate the demodulator behaviour in fading channels with deep and long fades, a state machine model was defined, based on the fact that demodulation takes place after channel estimation and that channel estimation is an iterative process that requires some time to converge. The overall behaviour is modelled with 4 states, representing the following effects: • loss of demodulator synchronization because of received C/N falls below operational threshold. Consequently transitions to coasting mode and after some timeout to the re-acquisition mode; • when the demodulator is operating below threshold, it feeds erased bits (neutral values) to the Turbo decoder; • demodulator re-acquisition time; • slicing impact on demodulator operation. The demodulator threshold is represented with a single value, independently of the channel behaviour, while in real implementation, it will vary according to mobile channel condition (user speed, type of mobile channel). A reference receiver with a single antenna reception has been considered for the definition of the demodulator threshold. In the model, the threshold margin is calculated as distance between the estimated C/N of the channel and the nominal demodulator threshold, both with Rayleigh, Rice channels. In the case of the 3-state LMS model, the margin is computed relative to the LOS receive C/N. All receivers of a multiplexing frame (simulations use a fixed pattern of services distribution) are simulated and the states of the receivers are used to determine whether the LLR estimates of all bits in a received symbol must be erased. As indicated in figure A3.1, it should be noted that all states have a transition to the idle state when the end of a burst occurs. While the green states are driven by the modulator behaviour, the red state represents the time slicing mechanism.                                                        Figure A3.1: State machine representing the demodulator process The transitions parameters are: • C/N : C/N estimated value at the receiver input; ETSI 320 ETSI TS 102 584 V1.1.2 (2010-XY) • Th : receiver C/N Threshold (for staying locked not for successful frame decoding); • D : duration of a state (origin of the transition); • Dlock : maximum coasting state duration for the receiver operating under noise; • TBurst : burst duration (assumed constant); • CTSlicing: time interval between slices; • Dacq : demodulator acquisition time (here assumed constant while in reality it will be a random variable). The start transition stands for the condition required to wake up from the power saving mode. This is dependent on the time slicing mechanism. For example, in the current implementation of DVB-H the start condition (CTSlicing) is as follows: CTSlicing = T>(Delta_T- Delta_T_Margin) Where Delta_T stands for the start of the next burst as it is indicated by the header of all clauses of the burst. Delta_T_Margin is the preparation time including some margin for the jitter. It is assumed that Delta_T is known as long as the demodulator has been locked for a certain period of the previous burst. However, if the demodulator has been unlocked or with C/N below the minimum required for frames detection over the whole duration of the previous burst this Delta_T remains unknown. A mechanism to overcome this limitation must then be implemented (i.e. establish a fixed time between bursts). In any case, the CTSlicing condition is set in accordance with the time slicing strategy. In this way, this method is also suitable for testing the time slicing performance. In simulations, it is assumed that a fixed and regular time-slicing is used for all programs in a multiplex and therefore, receivers go from idle state to acquisition state at a time corresponding to the time-slice start minus Dacq. To represent faithfully the demodulator estimator, it is necessary to consider that a given number of symbols are affected by the channel state to lead to a channel estimator state transition. This number of symbols is Nestimator. Impact of the demodulator behaviour in terms of performances will be show in clause A.12.2.5. IDLE state (0) The idle state represents the power saving mode. The idle state is the starting mode of the state machine. It is also entered each time a time slice ends. Transition to the unlocked state occurs when a start of slice is anticipated. The start of the slice information is exploited to switch on the demodulator in advance to allow fast acquisition. This time advance is supposed to be Dacq. DEMOD unlocked state (3) The demodulator is trying to acquire the carrier. Neutral value LLRs are fed to the Turbo decoder. If the channel estimates (A), averaged over Nestimator symbols goes below the threshold during the acquisition time, the acquisition is restarted until a correct margin is reached for the whole duration of the acquisition time. When the channel estimates are above threshold over the whole Dacq time, the transition to the locked state occurs. DEMOD locked state (1) Locked state occurs when the C/N estimate is above the demodulator threshold Th. Actual LLRs are delivered to the decoder. This state lasts until averaged channel estimates are above the required threshold margin. If the averaged estimated C/N falls below the margin, the demodulator switches to the coasting state. DEMOD coasting state (2) ETSI 321 ETSI TS 102 584 V1.1.2 (2010-XY) This state persists as long as the channel C/N estimates averaged over Nestimator symbols are under the demodulator threshold Th. Neutral values of LLR are fed to the Turbo decoder. Each time the demodulator enters this state, a timer starts for a duration of Dlock. If the estimated demodulator C/N averaged over Nestimator symbols remains under the threshold Th longer then Dlock, then the demodulator moves to the acquisition mode (3). Instead if after Dlock the averaged C/N estimate is above the demodulator threshold Th, then a transition to the locked mode (1) occurs and the timer is reset. A.3.3.2 OFDM simulations The demodulator state machine model must be activated to provide results on the LMS channel in OFDM mode. It accounts for the degradations introduced by the non-instantaneous channel synchronization in the receiver. As far as channel estimation and symbol phase synchronization are concerned, the ideal channel estimator remains, providing ideal phase and amplitude recovery to any demodulated symbol. The demodulator state machine parameters are recalled in Table A3.8.. Table A3.8: Demodulator state machine parameters definitions Parameters Definition Bandwidth Bandwidth of the OFDM signal, used to compute the number of subcarriers per symbol. Mode OFDM mode value indicating the symbol size (1K, 2K, 4K, 8K). Constellation Number of bits per subcarrier. Because input estimates are assumed to be on one subcarrier and output is supposed to correspond to LLR (bit by bit), the state machine generates constellation x number of subcarriers states per channel sample. Punc_Pattern_ID Puncturing pattern ID providing the coding rate. TPS_CommonMultiplier TPS parameters specifying the interleaver configuration.. TPS_NLateTaps TPS_NSlices TPS_SliceDistance TPS_NonLateIncrement MuxFrameProfile Profile of the repetition pattern where the services are repeated on a regular pattern, each service being allocated a fixed number of code words during that period. The number of services is deduced from that profile. ReceiverMarginThreshold_Db Margin between LOS channel C/N level and demodulator threshold in dB. C/N=-3,5 dB is selected as the receiver threshold value. LockDurationInSymbols Number of OFDM symbols during which the demodulator remains locked while the signal averaged values are below threshold. This gives the duration of the "coasting" mode. 68 symbols is selected for this duration value. AcquisitionDurationInSymbols Number of OFDM symbols required to switch to locked mode after the demodulator is started or after it has left the locked mode. 68 symbols is selected for this duration value. LengthOfAveragingInSymbols Number of OFDM symbols over which the demodulator is supposed to average the received channel estimates to determine its own state. 12 symbols is selected for this duration value. The set of demodulator state machine values corresponds to typical performance of existing DVB-H receivers. Simulation results without activation of the state machine are also provided as a benchmark. ETSI 322 ETSI TS 102 584 V1.1.2 (2010-XY) A.3.3.3 TDM simulations The same state machine model and parameters are used TDM simulations. The rationale is that, like for OFDM case, the TDM demodulators will rely on the "data aided" information from the pilots that are present in the TDM frame (equivalent to an OFDM symbol). Therefore, the acquisition and channel estimation algorithms will provide similar performances. This hypothesis is possibly slightly pessimistic for TDM which do not require the two dimensional (time/frequency) channel estimation of OFDM. A.3.3.4 Analysis of acquisition time for LMS channels The statistical analysis uses the symbols generated by the LMS channel model and feeds them into the receivers state machines model. Then, the number of occurrences of states are counted. ETSI 323 ETSI TS 102 584 V1.1.2 (2010-XY) State machine Erasures 60% 50% Percentage of erasures 40% 30% 20% 10% ITS 50 Km/h -6 dB ITS 3 Km/h -6 dB 0% ITS 50 Km/h -10 dB 272 204 ITS 3 Km/h -10 dB 136 68 (0) Acquisition tim e (OFDM symbols) State machine Erasures 25% 20% Percentage of erasures 15% 10% 5% SU 50 Km/h -6 dB SU 3 Km/h -6 dB 0% SU 50 Km/h -10 dB 272 204 SU 3 Km/h -10 dB 136 68 (0) Acquisition time (OFDM sym bols) Figure A3.2: Erasure rate for ITS and suburban channels Simulation results reported in figure A3.2show that the LMS-ITS channel generates much more erasures than the LMS- SU. Also, having the acquisition time in the order of 68 OFDM symbols (one DVB-SH frame) does not increase significantly the percentage of erasures with respect to the ideal case (instantaneous acquisition). Note that those results also depend on the averaging duration which can even decrease the number of erasures compared to the ideal acquisition case. ETSI 324 ETSI TS 102 584 V1.1.2 (2010-XY) A.3.4 MPE-IFEC Simulation conditions A.3.4.1 Definitions and interfaces The following interfaces are defined and represented in figure A3.3: • on sender side: - E1 represents the IP flow coming into the IP encapsulator; these IP flows are of variable bit rates since they represent individual services; - E2 represents the TS interface at the output of the IP encapsulator; the TS bit rate is fixed and is a function of the waveform parameters; - E3 is the output of the DVB-SH modulator; • on the receiver side: - R3 is the reception of the DVB-SH signal, at the input of the SH baseband decoder; - R2 is the output of the SH baseband receiver delivering a TS stream at continuous bit rate after all physical deinterleaving and decoding processes; - R1 is the output of the SH link layer receiver delivering IP flows at variable bit rates after all MPE-IFEC deinterleaving and decoding processes; - finally R0 is the output of the H264 video buffer ready for rendering. Broadcast network E3 (DVB-SH interface) R3 (DVB-SH interface) Bit rate DVB-SH modulator Physical interleaver buffer R2 (TS interface) E2 (TS interface) Compound constant bit rate Time IP Encapsulator MPE-IFEC buffer Individual Bit rate R1 (IP interface) variable bit rate RTP pre-decoder E1 (IP interface) buffer Time R0 (RTP interface) Encoder Encoder H264 player Encoder Figure A3.3: receiver interfaces All simulation results have been measured at interface R1. Some of the processes in the receiver may introduce jitter during transitions like zapping. The following performance results have been produced assuming a jitter-free interface, so assuming a constant delay in the physical and link layer buffers. Performance during transitions like zapping at non-jitter free interfaces are not scope of this implementation guidelines release. Performance including RTP buffer are found in the IPDC over DVB-SH implementation guidelines. However some aspects of zapping time improvements techniques, without performance figures, can be found in clause 6.2.5.2 Zapping time performance. A.3.4.2 Decoder architecture A simple decoder architecture is presented in figure A3.4. It implements the following features: ETSI 325 ETSI TS 102 584 V1.1.2 (2010-XY) • event driven approach: - time outs for the reception of a time slice burst with simple timing heuristics based on delta-t and max_burst_durations; - timing are based on ADST: one ADST is output at a deterministic time; this enables to output an ADST even in the case of a "starvation" due to a long erasure event; • support of erased bursts detection: - when erased bursts are detected, they are sequentially mapped over the ADT to avoid having ADT with outdated columns when decoding is requested; • "brute force approach" for the decoding: - the decoding is initiated every time a new information concerning the Encoding Matrix (ADT and FDT) is detected within a specific range; - this enables to output the ADST based on output time without requiring any new decoding since, when the ADST needs to be output, all the relevant information has already been processed by previous "brute force decoding"; the ADST is the best one with received information; - this enables also to support the decoding in different configurations by profiling the range of EM to be decoded; one typical configuration is to decode only at the jitter-free interface (no more information will be received so it is time to decode); another one is to decode by anticipation, especially during the zapping period without having received all parity; - this also enables to support a D>0; • typical configurations: - anticipated decoding:  Min_ADT = Min( ifdt_index(k,S-1); adt_index(k-D, 0));  Max_ADT = Min( ifdt_index(k,0); adt_index(k-D,B-1)); - without anticipated decoding:  Min_ADT = Min( ifdt_index(k,S-1); adt_index(k-D, 0));  Max_ADT = Min( ifdt_index(k,0); adt_index(k-D,0)). This decoder can be used for providing performance at the jitter free interface by using the non anticipated decoding range and D=0. • Min_ADT = Min(ifdt_index(k,S-1); adt_index(k, 0)) [M] = Min ((k – S– 2 + M)[M], k[M]) • Max_ADT = Min(ifdt_index(k,0); adt_index(k,0)) [M] = Min ( (k-1+M)[M], k[M] ) ETSI 326 ETSI TS 102 584 V1.1.2 (2010-XY) iFEC decoder overview event Section ReceiveSection Init event BurstEnd event call Map ADST k on ADTs Wait ProcessNewBurst event TimeoutBurst Map ADST k on ADTs end iteration Iterate over j=0..C-1 HandleADSTOutput event OutputADST Insert ADST k column j in ADT(adt_index(k,j),adt_column(k,j)) if ADT(k,j) is unreliable event BurstEnd or event TimeoutBurst Fill ADT adt_index(k,B-1), columns [floor(C/B)..K-1] with unreliable zeros ProcessNewBurst Create ADST(?,Tcur) with MPE sections of temp buffer Store ADST(?,Tcur) in ADSTs buffer event Section @Tsec Generate event OutputADST(Tcur) at Tcur+Tmax ReceiveSection Store section in temp buffer No MPE section with table_boundary presents ? Section of new burst ? (Tsec > Tprev_sec+dTprev_sec) Yes No Zeros padding of current ADST after MPE section frame_boundary Yes Create event BurstTimeout @ Tsec+max_burst_duration No Last section of burst ? MPE iFEC present ? No frame_boundary ? Yes Yes CBN=Read burst number Create event BurstEnd @ Tsec Set burst number to ADST(CBN-D,Tcur) Erase previous event BurstTimeout Map MPE iFEC sections on iFDTs Tprev_sec=Tsec Call Map current ADST CBN-D to ADTs dTprev_sec=dTsec No CBN-LBN >1 ? Missing bursts ? Yes event call ProcessMatrixDecode Call ProcessMissingADST ProcessMatrixDecode End of iteration Iterate over all M matrices m=min_ADT..max_ADT Zeros padding of ADSTs CBN-D-1-j[jmax] according to prev_burst_size Matrix m Call ProcessMatrixDecode(min_ADT,max_ADT) Nb Col Errors <= N ? No Yes LBN=CBN RS decoding of matrix m Yes Erase sections in temp buffer Update ADSTs with corrected columns event Call ProcessMissingBurst event OutputADST(T) ProcessMissingBurst HandleADSTOuput Nb_missing_ADSTs=CBN-LBN-1 Find ADST in ADSTs buffer associated to time T Yes >M Reset M matrices Start at first reliable byte No to find a first IP packet End of iteration Iterate i=1.. nb_missing_ADSTs Deliver this IP packet Create ADST LBN-D+i as unreliable Store ADST LBN-D+i in ADSTs buffer Generate eventOutputADST(Ttheoric) Yes Do we have Move to the next more reliable reliable byte data after ? Call Map ADST LBN-D+i to ADTs No Figure A3.4: simple MPE-IFEC decoder architecture ETSI 327 ETSI TS 102 584 V1.1.2 (2010-XY) A.3.4.3 Other parameters The following parameters and options have been sought during simulation: • IP size: set to 1 000 bytes; • encapsulation: MPE; • section packing: on; • errors processing: when a TS of a section is lost: the full MPE section is considered as lost; • the MPE-IFEC makes usage of TS level dumps which have a constant size at TP level; this fixed number of TP gives a capacity that is used to convey the IP flows using MPE sections and the RS parity using the MPE-IFEC. If the TS capacity is not sufficient to transport an IP packet or an MPE-IFEC section, the corresponding TS are set to null and not used. A.4 Void A.5 Void A.6 Void A.7 Propagation channels It should be remarked that Rayleigh and Rice channel models definitions are also used in the context of DVB-T/H but they refer to two different contexts: • The stationary Rice/Rayleigh channel models defined in EN 300 744 [28], clause B. The models represent stationary reception with antennas on the roof of a house (model "F1") or reception of signal from very high power repeater (as typical for DVB-T networks) with small antennas (model "P1"). The models are also (improperly) called RICE and RAYLEIGH in the context of DVB-T/H measurement. In reality the models represent a particular realization of TDL with stochastic fading process and TDL complex taps are fixed. These models play a minor role for the evaluation of DVB-SH and have been only used in an intermediate phase of the project; ETSI 328 ETSI TS 102 584 V1.1.2 (2010-XY) • The uncorrelated Rice/Rayleigh channel models used in clauses 7.2.2.6.2 and 7.2.2.6.3 of the present document. The performance of the system depends highly on the speed and the interleaver length. For the evaluation of the FEC decoding a simple channel model representing an ideal interleaving (upper bound in performance compared to finite length interleaving) has been selected. A symbol-by-symbol independent Rayleigh amplitude sample distribution is used to represent typical terrestrial FEC operating conditions. In a similar way a symbol-by-symbol independent Rice amplitude sample distribution is used to represent typical satellite FEC operating conditions. A.7.1 Terrestrial Channels A.7.1.1 Channel Model Definitions f0 Maximum Doppler Frequency: fD = v ⋅ being v the mobile speed, f0 the carrier frequency and c the c speed of light (c=3 108 m/s). 1  f  Jakes (Rayleigh) Doppler Profile: S( f )= rect   π fD 1− ( f fD ) 2  2 fD   f2  Gaussian Doppler Profile: G ( f ; σ ) = exp  − 2   2δ  For the terrestrial channel the state machine is not activated as it is assumed that the receiver is always in lock. Indeed, only the short term fading (TU6 model) is simulated, not the macro attenuation due to user large displacement within the cell (lognormal shadowing). For this channel, 10 to 20 minutes duration were used for statistics. A.7.1.1.1 Reference channels: AWGN and Rayleigh Before measuring the system performance in more complex mobile channels it is important to derive the system performance under two reference channel cases: AWGN and flat Rayleigh fading. The first one will allow deriving the reference and ultimate system performance allowing easy comparison with DVB-SH IG reference values. The flat Rayleigh fading introduces the channel amplitude and phase variability allowing again equipment under test easy verification against IG reference performances. These models do not depend on terminal speed / Doppler. A.7.1.1.2 Single transmitter: Classical TU6 The TU6 channel is widely used for terrestrial mobile networks. It is composed of a tapped delay line (TDL) with 6 taps as described in Table A7.1. To each tap is associated a delay and an average attenuation power on top of which a unit power Rayleigh complex fading process is applied. Each TDL tap Rayleigh process is independently generated and is characterized by correlated fading with a programmable power spectral density (PSD). A Rayleigh (Jakes) Doppler spectrum has been associated to the TU6 channel model. Table A7.1 shows the TU6 reference channel parameters. Table A7.1: TU6 channel model parameters. ETSI 329 ETSI TS 102 584 V1.1.2 (2010-XY) A.7.1.1. 3 Single transmitter: TU6 plus lognormal shadowing As previously discussed for DVB-SH, it is interesting to extend the classical TU6 by adding a lognormal shadowing on top. It is suggested to have a common lognormal shadowing process multiplying the TU6 output with zero mean and standard deviation 8 dB. The actual range of C/N in dB is to be defined as an output of the simulation exercise. The lognormal process coherence distance is 20 m [34]. A.7.1.1.3 SFN MBRAI model 1 for weak long echoes As first SFN model, it is proposed to reuse the one proposed in the document EICTA MBRAI document [i.34], p. 26-27 for SFN with weak long echoes. The channel profile parameters are described in Table A7.2 Where C/Nmode represents the C/N required in AWGN for the selected waveform configuration. We remark that the second SFN transmitter delay is not absolute but related to the selected OFDM Guard Interval (GI). This assumes that the GI is correctly sized for the SFN geometry and waveform parameters. This is a potential limitation of the model that will be overcome by the SFN Model 4. As for TU6 the fading PSD is Rayleigh distributed with the same relation with the maximum Doppler fD. Table A7.2: Mobile SFN synchronization test channel for weak long echoes. A.7.1.1.4 SFN MBRAI model 2 for strong long echoes This second SFN model is extracted from the document: EICTA MBRAI [Erreur ! Source du renvoi introuvable.] p. 26-27 and corresponds to a SFN with strong long echoes. The channel profile parameters are summarized in Table 3. ETSI 330 ETSI TS 102 584 V1.1.2 (2010-XY) Table A7.3: Mobile SFN synchronization test channel for long echoes. A.7.1.1.5 SFN MBRAI model 3 for strong short echoes3 The third SFN model is extracted from the document: EICTA MBRAI document [Erreur ! Source du renvoi introuvable.] p. 28 and corresponds to SFN with strong short echoes. The channel profile parameters are summarized in Table 4. Table A7.4: Mobile SFN synchronization test channel for strong short echoes. A.7.1.1.6 Portable Indoor and Outdoor channel model The PI & PO model here summarised is extracted from the document: EICTA MBRAI [Erreur ! Source du renvoi introuvable.], p. 26-27. The portable indoor (PI) and portable outdoor channel models have been developed by the Wing-TV project for describing the slowly moving hand held reception indoors and outdoors. The channel models are based on measurements in DVB-H Single Frequency Networks and have paths from two different transmitter locations. Definitions of the taps for the channels are given in sub-table 8 and sub-table 9 of Table A7.6. The indicated Doppler frequency of 1,5 Hz is corresponding to a 3 km/h velocity at mid UHF. It is suggested to scale up the original PI & PO Doppler spectra (UHF) to be representative of S-band so the fD value shall be increased to 6 Hz at 2.2 GHz. The Doppler spectra of various taps are defined in Table A7.5. 3 This model is typically less critical than SFN model 2. Therefore derivation of results is optional. ETSI 331 ETSI TS 102 584 V1.1.2 (2010-XY) Table A7.5: Doppler spectra of PI/PO model various taps Table A7.6: PI/PO channel configuration A.7.2 Satellite Channels A.7.2.1 Propagation channels A.7.2.1.1 AWGN Channel This simple reference channel is considered to be very useful to see the overall demodulator performance and allows easy comparison with ideal performance. A.7.2.1.2 Rician channel Model Rice channel is another easy model to see the demodulator performance in dynamic conditions and allow an easy comparison with simulation results. It is proposed to test the Rice factor K=5 dB at speeds of 3, 50 and 120 kmph. Additional mobile speed of 80 and 170 kmph will be required. ETSI 332 ETSI TS 102 584 V1.1.2 (2010-XY) A.7.2.1.3 Land Mobile Satellite Channel Model The following LMS propagation channel environments have been selected for laboratory tests: Satellite URBAN N/A SUBURBAN 3-state LMS-SU (50 Km/h) 3-state LMS-SU (3 Km/h) RURAL 3-state LMS-ITS (50 Km/h) Quasi-stationary Ricean, K=5 dB; 7 dB;10 dB The C/N and C/I levels to be used are derived from link budgets (see clause 11), with terminal characteristics given in clause 10. The used values are listed in clause A.9. The S-band parameters for the 3-state LMS channels are obtained from [35] and [36] and are dependent to the elevation angle. In addition to the 40 degrees elevation angle used for IGv1 results for which SU and ITS parameters are available (in [37]), 30 (in [38]) and 50 degrees (in [39]) elevation parameters have been added. For these two new satellite elevations, data is only available for the SU environment. Furthermore, a close look at the 50 degrees model parameters revealed possible inaccuracy of the model parameters due to the limited data set available. So results for this elevation shall be taken with caution. For the LMS channels the simulations were performed with and without receiver state machine. The parameters for the LMS channels are obtained from [40] and [41] and are specified in Tables A7.7, A7.8 and A7.9. The applicability of the selected values is not claimed to be universal but only as a starting point to differentiate between waveform configurations. For a discussion on the applicability of such models, refer to clause 4. From experiments, it was found that the simulation length when the 3-state Markov chain is activated should be at least 1 hour of simulated time for 50 kmph and at least 3 hours for 3 kmph. ETSI 333 ETSI TS 102 584 V1.1.2 (2010-XY) Table A7.7: LMS model states based on measurements parameters for 30° elevation State 1: LOS State 2: Shadowing State 3: Deep shadow Environment MP MP MP α (dB) ψ (dB) α (dB) ψ (dB) α (dB) ψ (dB) (dB) (dB) (dB) Open (*) Suburban -0.5 1.0 -15.0 -4.7 1.5 -19.0 -7.0 3.0 -20.0 Intermediate Tree-Shadow Heavy Tree- Shadow (2) Environment [P] [W] dcorr (m) Lframe (m) Ltrans (m) Open (2) 0.9531 0.0350 0.0119 0.7467 4.0 (1) Suburban 0.1891 0.6198 0.1911 0.1511 2.3 (1) 4.0 (1) 2.2 (1) 0.0631 0.3065 0.6304 0.1022 4.0 (1) Intermediate Tree-Shadow Heavy Tree- Shadow (2) Note 1 : These values have been extrapolated since they are not given in.[Fontan2] Note 2 : Not simulated, for information only Note 3: LMS model for 30° Elevation obtained from L- Band models. Simulations with this model may be inaccurate. Table A7.8: LMS model states based on measurements parameters for 40° elevation State 1: LOS State 2: Shadowing State 3: Deep shadow Environment α (dB) ψ (dB) MP (dB) α (dB) ψ (dB) MP (dB) α (dB) ψ (dB) MP (dB) Open (*) 0,1 0,37 -22,0 -1,0 0,5 -22,0 -2,25 0,13 -21,2 Suburban -1,0 0,5 -13,0 -3,7 0,98 -12,2 -15,0 5,9 -13,0 Intermediate -0,4 1,5 -13,2 -8,2 3,9 -12,7 -17,0 3,14 -10,0 Tree-Shadow Heavy Tree- - - - -10,1 2,25 -10,0 -19,0 4,0 -10,0 ETSI 334 ETSI TS 102 584 V1.1.2 (2010-XY) Shadow (2) Environment [P] [W] dcorr (m) Lframe (m) Ltrans (m) 0,9530 0,0431 0,0039 0,5 8,9 Open (2) 0,0515 0,9347 0,0138 0,375 2,5 7,5 12,4 0,0334 0,0238 0,9428 0,125 4,0 (1) 0,8177 0,1715 0,0108 0,4545 5,2 Suburban 0,1544 0,7997 0,0459 0,4545 1,7 3,7 2,2 0,1400 0,1433 0,7167 0,091 3,0 (1) 0,7193 0,1865 0,0942 0,3929 6,3 Intermediate 0,1848 0,7269 0,0883 0,3571 1,5 6,3 2,6 Tree-Shadow 0,1771 0,0971 0,7258 0,25 4,5 0,7792 0,0452 0,1756 0 - Heavy Tree- 0 0,9259 0,0741 0,5 1,7 4,8 3,5 Shadow (2) 0 0,0741 0,9259 0,5 4,5 NOTE 1: These values have been extrapolated since they are not given in [17]. NOTE 2: Not simulated, for information only. Table A7.9: LMS model states based on measurements parameters for 50° elevation State 1: LOS State 2: Shadowing State 3: Deep shadow Environment MP MP MP α (dB) ψ (dB) α (dB) ψ (dB) α (dB) ψ (dB) (dB) (dB) (dB) Open (*) Suburban -0.5 1.0 -17.0 -6.5 2.5 -17.0 -14.0 2.5 -20.0 Intermediate Tree-Shadow Heavy Tree- Shadow (2) ETSI 335 ETSI TS 102 584 V1.1.2 (2010-XY) Environment [P] [W] dcorr (m) Lframe (m) Ltrans (m) Open (2) 0.7498 0.2462 0.0040 0.1626 4.0 (1) Suburban 0.0479 0.9160 0.0361 0.7642 2.3 (1) 4.0 (1) 2.2 (1) 0.0554 0.3296 0.6150 0.0732 4.0 (1) Intermediate Tree-Shadow Heavy Tree- Shadow (2) Note 1 : These values have been extrapolated since they are not given in.[Fontan2] Note 2 : Not simulated, for information only Note 3: LMS model for 50° Elevation obtained from L- Band models. Simulations with this model may be inaccurate. The parameters values are: • α : Average value of the attenuation on the LOS link for a state; • ψ : Sandard deviation of the attenuation on the LOS link for a state; • [P] : Probability of occurrence of a transition (3x3 matrix); • [W] : Total probability of having a given state; • dcorr: Correlation distance of the channel; • LFrame: Minimum state frame length as defined in [17]; • LTrans: Transition region length as defined in [16]. A.7.2.1.4 Satellite quasi stationary channel For portable satellite reception it was considered important to also analyse the case of LOS with multipath but very slow user movement (e.g. 0,5 kmph or less). This channel is called quasi-stationary satellite. The results of this analysis are exploited in clause 11.7.2.2. The channel is modelled as AWGN over the FEC block size with a block-by-block slight C/N change due to the very slow fading. Hence, we can analytically compute for a given LOS C/N the probability to be below the demodulator threshold (see clause 11). Recalling the PDF of the power for a Ricean faded signal: 1   s 2 + d   s d    2  o   pD ( d ) = exp  −  I   2σ    σ  2 2 2σ     where d is the LOS signal power; ETSI 336 ETSI TS 102 584 V1.1.2 (2010-XY) σ2 the variance of the in-phase and quadrature multipath fading components; and I o ( ⋅) is the modified zero-th order Bessel function. The carrier to multipath ratio is defined as the ratio between the power of the signal complex envelope LOS component s and the power of the multipath (random) power component 2σ 2 : A.8 Evaluation criteria A.8.1 Introduction In case of long time interleaving and block fading, BER and WER alone are not representative of actual perceived quality. A long mute may be better than repetitive short error events. To take this consideration into account, we define new criteria in section A.8.3 after having defined the logging capabilities for both field trial and laboratory environments in section A.8.2. A.8.2 Laboratory and field trial logging These logging capabilities are common to terrestrial and satellite contexts. A.8.2.1 Overview A typical architecture of a laboratory and field trial measurement configuration is presented hereunder in Figure A8.6. The only difference between both configurations is the (x,y,z) coordinates are fixed in the laboratory case. (1s mean) Field strength / RF level P- (1s mean) C/(N+I) NB2 (mean) MER ESR5 (Mean) Sprectrum BER (1s) Lock status (AGC, carrier, TPS) PEC TS- Noise power PER Rx ESC TS Radio Demod Turbo IP- IPE Mod TS Sce MRC Decod Demod ES stream TS- Tx NB1 Post processing F- FEC ESR5 FER GPS position Time stamp Post MPE-IFEC Mi- Mi-FEC ESR5 processing Mi-FER Decod GPS IP stream Time-stamp Post-processing Post-processing IP- ESR5 IP- Measure storage traffic flow rce Mandatory: this would be delivered on each case Optional / nice to have: this could be delivered on a case by case basis video Video player Not supported: this will not be delivered in any case ETSI 337 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A8.6: field trial measurement configuration NB1: requires at least one MPE-FEC section per burst NB2: could be used in replacement of FEC/FER and F-ESR5 ETSI 338 ETSI TS 102 584 V1.1.2 (2010-XY) A.8.2.2 Physical layer logging capabilities The physical layer loggings are listed in Table A8.10. The dumps shall be aligned with the GPS time stamp. This can be achieved, for instance, by creating a log file, which reports for each line the time stamp, the GPS position, and the pointer to the last written byte in each of the dumps. Parameter Time Used for Derived from Required Status, resolution preparation/Setup implemented in characterization version Position 1 second correlation between the recorded dumps GPS Mandatory and the position of the terminal Time stamp 1 second correlation between the recorded dumps GPS, with timing Mandatory and the position of the terminal interpolation C/(N+I) 1 second Alternative to field strength; allows Demodulator Mandatory estimation (average over simple correlation between received at least 10 signal level and output of the physical samples) layer decoder; verification of the channel status, esp. specific environments under scope of test Lock status 1 second detect eventual signal losses, useful for Demodulator Mandatory (reporting detecting unexpected behaviours of the eventual lock terminal losses during the past may be used to evaluate eventual gaps in second) the recorded dumps Field strength 1 second Raw data for coverage analysis and time Band power AGC and antenna Desirable (average over series generation measurement calibration at least 10 NB: 2 values(or samples) AGC values more) when diversity is activated Noise power 1 second Estimate C/N for low field strength Band power Desirable level (average over measurement in other at least 10 band samples) BER (channel Average over Cross-check of demodulator Analyze pilot fields or cBER versus C/N Desirable BER) at approx. 100 performance padding for different fading demodulator ms channels This can be output measured with pre- configured modulator input Strength of On user Identify critical interferer AGC value Spectrum mask Nice to have interference request in case of un- Wideband power level allowed interference expected behaviour Spectrum analysis MER Stationary Signal quality analysis for stationary Offline analysis of Nice to have only reception recorded I/Q data Format may differ (MER instead of I&Q) Spectrum One snapshot Fading statistics vs. frequency axis Offline analysis of Nice to have per second recorded I/Q data Format may differ (FFT in {freq; power} coordinates Interference Upon user Identify critical interferer Spectrum analyzer Nice to have level request Table A8.10: physical layer logging capabilities ETSI 339 ETSI TS 102 584 V1.1.2 (2010-XY) A.8.2.3 Link layer logging capabilities The upper layer logging capabilities are listed in Table A8.11. Also here, the results shall be aligned with the GPS time stamp. Parameter Time Used for Derived from Required Status, resolution preparation/ implemented Setup in version characterization Burst Error Burst ~1 Permits the analysis in terms of Depends: Depends on the case Mandatory Indicator second ESR5(20) service availability - after iFEC decoding (see MiFER/MiFEC) - after TC for Class-2 (see PER) Physical layer MPEG2 Evaluate WER and ESR5 Output of turbo Special internal to Desirable (at frames/ packet performance decoder MPEG2 TS least for some MPEG-TS signalling for sessions) error series Complete analysis of the performance detection of gaps if after PHY (turbo) decoding. greater than MPEG2 CC Map different MPE-iFEC profiles on the TS Requires additional signalling for detecting gaps (outages) FEC/FER Frame=burst~ Frame Error Count / Frame Error Monitoring of the Requires at least 1 Desirable 1 second Rate: measure after turbo-decoding at MPE-FEC frame (so MPE-FEC section the level of MPE-FEC frame at link level) per burst because based on MPE-FEC Gives an indication of the burst error frame indicator MiFER/MiFE Frame=burst~ MPE iFec Frame Error Ratio after Logging at the output Needs link layer Desirable for R 1 second MPE-iFEC decoding of the MPE-IFEC MPE-IFEC class 1 decoder configurations Can be used for ESR5 computation PER/PEC MPEG2 Packet error Rate measured after information derived Special internal to Desirable packet Turbo decoding; from the MPEG2 TS TS_erroneous_flag signalling for can be used to compute different set thanks to the detection of gaps if MPE-iFEC schemes via post- CRC16 computed greater than MPEG2 processing over each ST CC ADT columns Each ADT Helps in reproducing the behaviour of MPE-iFEC decoder Nice-to-have status flags column different MPE-iFEC techniques, input/output I/O sharing the same parameters (B, S, T) of the encapsulated configuration (same results can be achieved by Meas. No. 1, but with more effort) MPE/-iFEC Section level Permits a refined analysis of the MPE-iFEC decoder Continuity issues to Nice to have sections status recorded results logs manager Full received IP level Permits eventual off-line play-out of Output of Nice to have (decapsulated) the video MPE/MPE-iFEC payload, at IP decapsulation level Permits ESR5 computation at IP level based on comparison between IP input / IP output Table A8.11: physical layer logging capabilities ETSI 340 ETSI TS 102 584 V1.1.2 (2010-XY) A.8.3 Performance metrics A.8.3.1 Introduction To define the performance metrics listed in Figure A8.6, we zoon on the receiver side on Figure 5.1.                                                Figure A8.7: performance measurement points in the receiver ETSI 341 ETSI TS 102 584 V1.1.2 (2010-XY) A.8.3.2 Output of the turbo decoder (R3) The TS packet status is derived from the CRC16 in the turbo decode that sets the status of the transport_error_indicator bit in the TS header. From this information we can derive: - The T_PER (Transport Packet Error Rate), - The TP_ESC (TP Erroneous second count), - The TP_ESR (TP Erroneous second ratio) A.8.3.3 After PID filtering (R2) Measurements at the R2 reference point provide: - The FC (MPE Frame counter), - The FEC (Frame Error Count), - The FER (Frame Error Rate). Depending on the receiver state (Active/Locked or Active/Not Locked), the receiver may receive the complete set, a partial set, or none of the transmitted MPE sections in a time-slice burst. MPE sections contain information on the next time slice burst used for the management of the receiver wake-up times, and that allow the detection of lost MPE frames provided that the receiver implements appropriate mechanisms. The MPE Frame Counter (the FC) is incremented each time a time slice burst is expected. The FEC counts the number of erroneous MPE-Frames, i.e., MPE frames - Either containing at least one MPE section for which the CRC-32 check fails, - Or lost MPE-frames. The FER is simply expressed as FEC/FC, i.e., the ratio of erroneous MPE-frames. Additionally, measurements at the R2 reference point provide in case MPE-IFEC is used: - The MiFEC (MPE-iFEC Frame Error Count), and - The MiFER (MPE-iFEC Frame Error Rate). The MPE-iFEC Frame Error Counter (the MiFEC) is incremented each time an ADST table (or equivalently the IP datagram payload of a Time Slice Burst output by the MPE-iFEC decoder) is detected as erroneous. This decision relies on the RS decoder ability for errors correction (i.e., the capacity of the RS in terms of errors detection/correction). The MPE-iFEC frame Error Rate (the MiFER) is simply expressed as MiFEC/FC, i.e., the ratio of erroneous ADST tables generated by the RS decoder. A.8.3.4 After MPE decapsulation and MPE-IFEC decoding (R1) Measurement at the IP level includes the IP PER (packet error rate). It is usually computed either by post processing comparison between sent and received files (a methodology that can be applied to any kind of traffic, even those encrypted) or by using packets with a specific header having a sequential number that enables to detect lost packets. A.8.3.5 ESR5(20) metrics A.8.3.5.1 Generic definition The ESR5(20) definition is recalled hereafter: “ESR5(20) is the ratio of windows for which ESR5 is fulfilled, over the total number of windows ESR5 criterion is fulfilled when in a time interval of 20 seconds there is at most one second in error.” The ESR5(20) evaluation formula is: ETSI 342 ETSI TS 102 584 V1.1.2 (2010-XY)  max [ 0; n_of_erroneous_seconds( w) − 1]  Nw  ceil  20  ESR5(20) = 1 − w =1   Nw where we assumed an individual observation window size of 20 seconds, N w represents the number of observation windows and n_of_erroneous_seconds(w) is the number of erroneous seconds in the window w, the ceil function round the specified number up, and return the smallest number that is greater than or equal to the specified number. The previous formula needs to be adapted to the type of measurements possible within the receiver: - It is very important that ESR5(20) statistics are obtained over a long enough measurement time in particular at low mobile speed. - The N w 20 s windows used for the measurement may be disjoint or overlapped. The important is that enough statistics are accumulated to achieve reliable ESR5(20) estimate over the channel under consideration. - The above equation can be easily generalised to different window lengths by simply replacing 20 to the actual window length in seconds and the -1 at the nominator by -5*window_length / 100. A.8.3.5.2 F-ESR5(20) and Mi-ESR5(20) According to Figure 5.1, the ESR5(20)measurement point is: - R1 for class 1 receivers in a DVB-SH-A network configured with MPE-iFEC protection, - R2 for class 2 and class 1 receivers in a DVB-SH network configured without MPE-iFEC protection. So, a second is erroneous if it contains at least: - One erroneous MPE-iFEC ADST (class 1 with MPE-iFEC), therefore called Mi_ESR5(20), or - One erroneous MPE frame (class2 without MPE-iFEC/ class 1 without MPE-iFEC), therefore called F_ESR5(20). The F_ESR5(20) and Mi_ESR5(20) evaluations are performed by a post processing tool that takes into account the Time Slice Burst repetition period. A.8.3.5.3 IP_ESR5(20) Sometimes, an additional measurement point is required, providing finer analysis and debug means. The erroneous second decision is based on the comparison of the transmitted IP datagrams carrying the traffic of one given service with the comparison of the IP datagrams output at the R1 interface. It allows the detection of: - IP datagrams with bad checksum, - Duplicated IP datagrams, - Lost IP datagrams, - IP datagrams with erroneous payload. The ESR5(20) evaluated from the erroneous second decision based on comparison on the transmitted IP datagrams with the IP datagrams at the R1 interface in Figure 5.1 is called the IP_ESR5(20). A.8.3.6 Synthesis The following Table A8.12 gives the list of performance criteria. ETSI 343 ETSI TS 102 584 V1.1.2 (2010-XY) Acronym Full name Measurement point (name and interface) T_PER Transport Packet Error Rate Turbo-decoder output R3 TP_ESC TP Erroneous second count Turbo-decoder output R3 TP_ESR TP Erroneous second ratio Turbo-decoder output R3 FC MPE Frame counter In MPE decapsulation R2 FEC Frame Error Count In MPE decapsulation R2 FER Frame Error Rate In MPE decapsulation R2 MiFEC MPE-iFEC Frame Error Count In MPE-IFEC decoder R2 MiFER MPE-iFEC Frame Error Rate In MPE-IFEC decoder R2 IP PER IP Packet Error Rate Output of the receiver R3 F_ESR5(20) Frame Erroneous Second ratio 20 seconds In MPE decapsulation R2 Mi_ESR5(20) Mi-Frame Erroneous Second ratio 20 seconds In MPE-IFEC decoder R2 IP_ESR5(20) IP Erroneous second ratio Output of the receiver R1 Table A8.12: performance measurements criteria synthesis For a receiver nominal behaviour (after complete receiver validation tests), the IP_ESR5(20) and the F_ESR5(20) (no MPE-iFEC only) or Mi_ESR5(20) should give the same results. A.9 Summary of simulated cases Table Table A9.9-A9.3 hereafter provides all the simulation cases with the associated parameters. The satellite C/N and C/I are in line with the link budgets detailed in clause 11 (rationale with link-budgets is given at the end of this clause). In the Tables A9.4-A9.6 for the satellite case C/N represents the LOS C/N (no channel fading/shadowing) while C/I represents the co-channel useful signal over interference and uplink C/N contribution. Note that satellite link C/I and C/N are provided separately as the interference is suppose to fade together with the signal and is simulated like an extra noise source modulated by the same fade as the useful signal. Table A9.9 hereafter reports the list of simulated case in TU6 channel. The spectral efficiency value is given for information and corresponds to the case where MPE-IFEC code rate is equal to 1. The Demodulator State Machine is always OFF. Table A9.9: List of simulated cases in TU6 channel ETSI 344 ETSI TS 102 584 V1.1.2 (2010-XY) Overall Waveform Speed C/I C/N ID Efficiency Additional parameters Comment configuration (Kmph) [bpsHz]  Demodulator State Machine OFF  All possible code rates and  Mobile fading bandwidth for a range O_16QAM_xx_S 3-50 20 dB [2:18] 1 0.8-2.67 Table A10.6 modulations are tested from 3 to 600 Hz. O_QPSK_xx_S 3-50 20 dB [-2:14] 2 0.4-1.33 Table A10.8 O_16QAM_xx_UL 3-50 20 dB [2:18] 3 0.8-2.67 Table A10.17, Table A10.18 Table A10.19, O_QPSK_xx_UL 3-50 20 dB [-2:14] 4 0.4-1.33 Table A10.20 Table A9.10and A9.11 hereafter report the simulated cases to test different interleaver configurations. This is done for OFDM (QPSK and 16QAM) and TDM (only QPSK and 8PSK, no 16APSK). Only a subset of all possible code rates are tested for each selected modulation. And in each case two types of interleaver are tested (uniform late and uniform long). The Demodulator State Machine is always ON. Table A9.10: List of simulated TDM cases in LMS channel Overall Waveform Speed Efficien Additional C/I C/N ID Comment configuration (Kmph) cy parameters [bpsHz] Table A10.39, T_QPSK_1/3_U_xx 1  Demodulator State Machine ON  Interleaver duration ranging from 500ms up to 11 seconds IFEC CR=1 12.3 12.0 Table A10.48, T_QPSK_1/3_UL_xx 2 Table A10.49, 0.67 IFEC CR=1 Table A10.43, T_8PSK_2/9_U_xx 3 LMS-SU - LMS-ITS; Elevation angle: 40 IFEC CR=1 11.8 12.5 Table A10.54, T_8PSK_2/9_UL_xx 4 IFEC CR=1 Speed 3-50 Kmph Table A10.40, T_QPSK_1/2_U_xx 5 IFEC CR=1 12.3 12.0 Table A10.49, T_QPSK_1/2_UL_xx 6 1.0 IFEC CR=1 Table A10.44, T_8PSK_1/3_U_xx 7 IFEC CR=1 11.8 12.5 Table A10.55, T_8PSK_1/3_UL_xx 8 IFEC CR=1 Table A10.45, T_8PSK_2/5_U_xx 9 1.2 IFEC CR=1 11.8 12.5 Table A10.56, T_8PSK_2/5_UL_xx 10 IFEC CR=1 ETSI 345 ETSI TS 102 584 V1.1.2 (2010-XY) Table A9.41, T_QPSK_2/3_U_xx 11 IFEC CR=1 12.3 12.0 1.33 Table A10.51, T_QPSK_2/3_UL_xx 12 IFEC CR=1 Table A9.11: List of simulated OFDM cases in LMS channel Overall Waveform C/N C/I Efficien Additional Channel ID Comment configuration [dB] [dB] cy parameters [bpsHz] O_QPSK_xx_U_10s  Demodulator State  All coding Rates Machine ON Table A10.30, 1 LMS-SU - LMS-ITS; Elevation angle: 40 IFEC CR=1 Speed 50 12.3 12.0 0.67 O_QPSK_xx_UL_10s Table A9.19, 2 Table A9.20, IFEC CR=1 ETSI 346 ETSI TS 102 584 V1.1.2 (2010-XY) Rationale between values used in Simulation and values given in clause 11 link-budgets C/N are provided in link bugets from clause A.11. As LMS simulations are lossless, C/N provided in link budgets are reduced from the amount of implementations losses identified in clause 10 (as well as in link budgets of clause 11). Thus the following C/N are used in the LMS simulations: Vehicular terminal: Table A9.12 Waveform Sh-A Sh-A Sh-B Sh-B Modulation QPSK 16QAM QPSK 8PSK Link budget LOS C/N (dB) 12,3 12,3 12,8 12,8 Implementation losses (dB) 1,1 1,5 0,5 1 C/N for simulations 11,2 10,8 12,3 11,8 Handheld (2b) terminal: Table A9.13 Waveform Sh-A Sh-A Sh-B Sh-B Modulation QPSK 16QAM QPSK 8PSK Link budget LOS C/N (dB) 6,2 6,2 6,7(see note) 6,7(see note) Implementation losses (dB) 1,1 1,5 0,5 1 C/N for simulations 5,1 4,7 6,2 5,7 NOTE: Link budget for SH-B and handheld terminal is not provided in clause 11. However C/N may be estimated by similarly to vehicular terminal, where SH-B C/N is 0,5 dB over SH-A C/N. The same methodology is used for the calculation of the C/I. In link budgets, two C/I contributions are identified. Thus the following C/I are used in the LMS simulations: Table A9.14 Waveform Sh-A Sh-A Sh-B Sh-B Modulation QPSK 16QAM QPSK 8PSK Link budget uplink C/I (dB) 19,5 19,5 20 20 Link budget Satellite C/I (dB) 14 14 14 14 Link budget total C/I (dB) 13 13 13 13 Implementation losses (dB) 1,1 1,5 0,5 1 C/I for simulations 11,9 11,5 12,5 12 ETSI 347 ETSI TS 102 584 V1.1.2 (2010-XY) A.10 Configurations description This clause provides the detailed reference configurations used in other places in the document (simulation, laboratory test, field test). As a summary, we precise for all the listed configurations, where it is being used. Some configurations are actually not tested in this guidelines, however they are listed for sake of completeness. Section Title More def Tables SIM LAB FIELD OFDM class 1 16QAM variable code rate classical A10.6 X X X variable code rate legacy A10.7 X OFDM class 1 QPSK variable code rate classical A10.8 X X X A.10.3 OFDM class 2 OFDM class 2 uniform late 16QAM variable duration classical A10.11, A10.13, A10.15 X variable duration optimized A10.12, A10.14, A10.16 fixed duration 10s variable code rate classical A10.17 X X fixed duration 10s variable code rate optimized A10.18 X X OFDM class 2 uniform late QPSK variable duration classical A10.19 variable duration optimized A10.20 X fixed duration 10s variable code rate classical A10.21 X fixed duration 10s variable code rate optimized A10.22 X X OFDM class 2 uniform long OFDM class 2 uniform long 16QAM variable duration classical A10.23, A10.24, A10.25 fixed duration 10s variable code rate A10.26 X X OFDM class 2 uniform long QPSK variable duration classical A10.27, A10.28, A10.29 X X X fixed duration 10s variable code rate A10.30 X A.10.4 TDM class 1 TDM class 1 T-QPSK A10.32 TDM class 1 T-8PSK classical A10.33 X legacy A10.34 TDM class 1 T-16APSK classical A10.35 legacy A10.36 A.10.5 TDM class 2 TDM class 2 uniform long T-QPSK variable delay A10.39, A10.40, A10.41 X fixed duration 10s variable code rate A10.42 X X TDM class 2 uniform long T-8PSK variable delay A10.43, A10.44, A10.45 X fixed duration 10s variable code rate A10.46 X TDM class 2 uniform long T-16APSK fixed duration 10s variable code rate A10.47 X X TDM class 2 uniform late T-QPSK variable delay classical A10.48 X variable delay optimized A10.49 X fixed duration 10s variable code rate classical A10.50, A10.51 fixed duration 10s variable code rate optimized A10.52, A10.53 TDM class 2 uniform late T-8PSK variable delay classical A10.54, A10.55, A10.56 X fixed duration 10s variable code rate A10.57 X TDM class 2 uniform late T-16APSK fixed duration 10s variable code rate A10.58 X TDM 8PSK1/3 uniform long variable delay (16QAM) A10.59 X Table A10.0: Usage of listed configuration throughout the document The way the cases are presented in the table is detailed in section A.10.1. ETSI 348 ETSI TS 102 584 V1.1.2 (2010-XY) A.10.1 Introduction In the implementation guideline, only short reference should be made so that complete description of the configurations are factorized in this Clause. This 2-tables structure is exemplified below. By this way, waveform configurations are mutualised in the clause with all their details and can be reused several times in the body of the document in different circumstances. For instance, in the Figure A10.8, we can see that the body table makes a reference to the physical configuration detailed in the 16QAM short interleaver table in the clause A.10. The corresponding case in the table is the number 2 (code rate 2/9). Since the IFEC code rate selected in the body table is 1, the corresponding bit rate at IP level is 304 kbps, the corresponding TP are 240/208/0/24/8…    Figure A10.8: organization logic of the table configuration Let us have a look at the waveform tables presented in the Clause. For each case presented in a column, there are a number of parameters given line by line but organized in 3 groups: - Group 1: Physical parameters groups: see Table A.1 below; this table basically lists physical parameters as can be given by the SH_delivery_system descriptor. There is no special information to be given here. The resulting interleaver configuration is represented by Figure A10.9 giving individual tap length from the receiver perspective (tap 0 is always fixed to 0). ETSI 349 ETSI TS 102 584 V1.1.2 (2010-XY) Interleaver 1 length (IU) 300 250 200 150 100 50 0 0 5 10 15 Tap number 20 25 30 35 40 45 Figure A10.9: example of resulting physical interleaver Parameter name Description Comment Bandwidth (MHz) ∈{1.7;5;6;7;8} - FFT Mode ∈{1;2;4; 8}*1024 - Modulation ∈{O-QPSK;16QAM} for OFDM - ∈{T-QPSK;8PSK; 16APSK} for TDM PHY FEC rate ∈{1/5;2/9;1/4;2/7;1/3;2/5;1/2;2/3} - Interleaver configuration The 5 parameters of the interleaver - (common_multiplier/nof_l configuration ate_taps/ nof_slices/slice_distance/n on_late_increment) Minimum / late / maximum The minimum, late and maximum durations Minimum value definition could be interleaver duration (ms) of the physical interleaver with the modified to reflect the duration above configuration: needed for receiving a number of parity bits ≥ 1 (fast zapping - minimum value is usually 0 as conditions) given by the smallest tap (0) - late value is given when there is a late part in the interleaver - maximum value is proportional to largest tap value MPEG TS total bit rate Bit rate at the MPEG2 level Takes into account SH framing with EFRAMES overhead and padding Table A10.1: physical parameters description - Group 2: TS layer parameters group: see Table A10.2 below; this table gives the structure of the time-slice services, structure that is to be fixed before any IP service is inserted. ETSI 350 ETSI TS 102 584 V1.1.2 (2010-XY) Parameter name Description Comment Services Number of time-slice services must be ≥3 Repetition interval (ms) Repetition interval between services, must be fixed value it has to be chosen as a multiple of SH frame duration so that simulation cases are simplified. Burst duration (ms) Average burst duration: Max burst duration can be estimated as the double of this value. repetiton_interval burst_duration = services OFDM/TDM symbols/ Number of symbols necessary to This value is useful check the coded FEC carry the encoded word number of coded word, and therefore the useful bandwidth, required to carry the data. TP per service Number of MPEG2 TS packets The total number of TS can be used available in each time-slice in 4 different ways: burst. There are 5 values given: - PSI TP: TP used to carry the PSI signalling, based on the PSI - total number of TP ratio; - data part - DATA TP: carry MPE - ifec part sections (and IP packets); - safety part - IFEC TP: carry MPE- - psi part. IFEC sections; TPtotal = TPdata + TPIFEC - Safety TP: carry null TS packets as a way to cope for bit rate + TPsafety + TPPSI computation errors. This is derived from the Safety factor that can be fixed to 90%. Note that section packing is used only in each category (MPE alone, MPE-IFEC alone) but not between MPE and MPE- IFEC In case several IFEC configurations are possible, we give extreme possible computations MPEG2 per service bit rate Bit rate at the MPEG2 level for the The bit rate is done at MPEG2 level (kbps at TS level) individual service and the so that we include all different categories of TP: overhead like MPE and MPE- IFEC headers and MPEG2 MPEG2_bitrate kbps = TP *188 * 8 headers. repetition_intervalms The following values are given: In case several IFEC code rates are - total capacity bit rate possible, we give extreme - data capacity bit rate possible computations. - ifec capacity bit rate - safety capacity bit rate - pas capacity bit rate Table A10.2: TS layer parameters description The resulting structure in terms of time-slice services with their duration is given in Figure A10.10. The repetition interval has to be chosen so that it corresponds to a multiple of SH frame duration for ease of simulation purpose (for lab and field trial cases, this is not a requirement). This representation is only an average configuration since boundaries between services may vary in the case of VBR traffic. As can be seen in Figure A10.10, the maximum number of TP (here 240) results directly from the division of the TS total capacity during a repetition interval (as given by the multiplication of the radio capacity from radio configuration and the repetition interval from the TS configuration) by the number of service. This total number of TP has to be allocated to different budgets: ETSI 351 ETSI TS 102 584 V1.1.2 (2010-XY) - PSI ratio: a fraction of the TS capacity has to be allocated to the PSI signalling. This value can be typically set to 1/32 for lab and field environments. In simulation environments, the value is null. In the example we use 1/32 of 240, 8 TP. - “safety TP”: to avoid exceeding capacity of the link due to rounding errors in bit rate computation, a safety factor is usually fixed. A typical value of 90% for lab and field cases is taken meaning 10% of the capacity is useless. In the example, 10% of 240-8=232, 24 TP are considered as “safety TP”. - The remainder of PSI and safety can be used to transport DATA and IFEC: 208 TP. In our case, we have an IFEC code rate of 2/3, meaning we have 138 and 70 TP for, respectively, DATA and IFEC sections. - In reality, real IP traffic may be lower than this value. Therefore the actual capacity based on “real” IP traffic but not on maximum TP capacity is also shown as values inferior to the maximum.                   Figure A10.10: example of TS layer resulting configuration - Group 3: Link layer parameters groups: see Table A10.3 below; this table gives only the potential range of MPE-IFEC code rates and the resulting bit rates computed at MPEG2 level for the MPE sections (excluding the MPE-IFEC and excluding the safety TP). Therefore the bit rates given at TS level also exclude IFEC and safety TP and give the actual maximum throughput. It is acknowledged that any detailed MPE-IFEC configuration being done at the link layer must be provided in the complete configuration table that will reference those physical layer configurations. Therefore you will not find any precise MPE-IFEC parameters such as B, S, D, number of rows… in these tables. Parameter name Description Comment LL-FEC rate min / Range of possible Min: best protection; recommended: typical value; max: recommended code rate w/o any IFEC protection, best throughput / max values Bitrate/service (kbps Resulting bit rates at The bit rate is done at IP level, therefore is must be lower at IP level) IP level than the data bit rate computed at MPEG2 layer by a factor of 184 1000 for IP packet of size * ≈ 0.96 188 1017 1000 bytes Table A10.3: link layer parameters description As a final reminder, we give below main parameters values for different configurations. ETSI 352 ETSI TS 102 584 V1.1.2 (2010-XY) Configuration Name Value Explanation Simulation PSI ratio Epsilon (~0) We don’t take into account PSI and safety at the simulation level: we can ensure Safety 100% perfect CBR traffic filling perfectly the burst GI non SFN OFDM ¼ For legacy purposes sat GI OFDM terrestrial ¼ For legacy purposes Lab & Field trial PSI ratio 1/32 When using real-life equipment we cannot any more assume perfect data rates Safety 90% filling exactly the capacity. To compare with simulation, we must consider the total capacity bit rate at TS level that must be the same. GI non SFN OFDM 1/32 To optimize satellite throughput sat GI OFDM terrestrial 1/8 To optimize terrestrial throughput Table A10.4: typical configurations parameters values For simplification purposes, all GI in the configuration are fixed to ¼. However, according to real-life lab & field experiments, different values may be selected like 1/8 and 1/32. The reader will have to adjust the bitrate values accordingly. A.10.2 OFDM class 1 In the IG release 1, we used class 1 configurations that exceeded the class 1 memory. Therefore we provide both class 1 compliant configurations that becomes the new reference case mandated for all lab and trial configurations and, for legacy purposes and comparison with IG release 1, the “old class 1 configuration” tables called “class 1 legacy”. Modulation Code rate Duration Table 16QAM Variable (1/5 to 2/3) Short (official) Table A10.6 Variable (1/5 to 2/3) Short (legacy) Table A10.7 QPSK Variable (1/5 to 2/3) short Table A10.8 Table A10.5: OFDM class 1 physical layer configuration summary ETSI 353 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X - - X X X X Parameter/Case name 16QAM_1/5_S 16QAM_2/9_S 16QAM_1/4_S 16QAM_2/7_S 16QAM_1/3_S 16QAM_2/5_S 16QAM_1/2_S 16QAM_2/3_S PHYSICAL CONFIGURATION Bandwidth (MHz) 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM PHY FEC rate 1/5 2/9 ¼ 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 nof_slices/slice_distance/non_late_increment ) Min / late / Max interleaver duration (ms) 0 / 105 / 105 0 / 105 / 105 0 / 105 / 105 0 / 105 / 105 0 / 105 / 105 0 / 105 / 105 0 / 105 / 105 0 / 105 / 105 MPEG TS total bit rate 2,666 2,962 3,357 3,752 4,443 5,332 6,714 8,887 TS LAYER CONFIGURATION  8 8 8 8 8 8 8 8 Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 122 122 122 122 122 122 122 122 OFDM Symbols/ coded FEC 10 9 8 7 6 5 4 3 216/94/94/21/7 240/104/104/24/8 272/118/118/27/9 304/132/132/30/10 360/156/157/35/12 432/188/188/42/14 544/237/237/53/17 720/313/314/70/23 total/data/ifec/safety/psi TP per time slice 216/125/63/21/7 240/138/70/24/8 272/157/79/27/9 304/176/88/30/10 360/208/105/35/12 432/250/126/42/14 544/316/158/53/17 720/418/209/70/23 216/188/0/21/7 240/208/0/24/8 272/236/0/27/9 304/264/0/30/10 360/313/0/35/12 432/376/0/42/14 544/474/0/53/17 720/627/0/70/23 333/145/145/32/11 370/160/160/37/12 420/182/182/42/14 469/204/204/46/15 555/241/242/54/19 666/290/290/65/22 839/366/366/82/26 1111/483/484/108/35 total/data/ifec/safety/psi kbps@TP per time 333/193/97/32/11 370/213/108/37/12 420/242/122/42/14 469/272/136/46/15 555/321/162/54/19 666/386/194/65/22 839/488/244/82/26 1111/645/322/108/35 slice 333/290//32/11 370/321//37/12 420/364//42/14 469/407//46/15 555/483//54/19 666/580//65/22 839/731//82/26 1111/967//108/35 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66/ 1 0.5 / 0.66/ 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 Bitrate/service (kbps at IP level) 140 / 180 / 280 148 / 197 / 304 173 / 230 / 345 189 / 255 / 386 230 / 304 / 460 280 / 370 / 559 345 / 468 / 698 460 / 616 / 928 Table A10.6: OFDM 16QAM class 1 configuration ETSI 354 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X - - X X X X Parameter/Case name 16QAM_1/5_S 16QAM_2/9_S 16QAM_1/4_S 16QAM_2/7_S 16QAM_1/3_S 16QAM_2/5_S 16QAM_1/2_S 16QAM_2/3_S PHYSICAL CONFIGURATION Bandwidth (MHz) 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 10/48/1/0/0 10/48/1/0/0 10/48/1/0/0 10/48/1/0/0 10/48/1/0/0 10/48/1/0/0 10/48/1/0/0 10/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 211 / 211 0 / 211 / 211 0 / 211 / 211 0 / 211 / 211 0 / 211 / 211 0 / 211 / 211 0 / 211 / 211 0 / 211 / 211 MPEG TS total bit rate 2,666 2,962 3,357 3,752 4,443 5,332 6,714 8,887 TS LAYER CONFIGURATION  8 8 8 8 8 8 8 8 Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 122 122 122 122 122 122 122 122 OFDM Symbols/ coded FEC 10 9 8 7 6 5 4 3 216/94/94/21/7 240/104/104/24/8 272/118/118/27/9 304/132/132/30/10 360/156/157/35/12 432/188/188/42/14 544/237/237/53/17 720/313/314/70/23 total/data/ifec/safety/psi TP per time slice 216/125/63/21/7 240/138/70/24/8 272/157/79/27/9 304/176/88/30/10 360/208/105/35/12 432/250/126/42/14 544/316/158/53/17 720/418/209/70/23 216/188/0/21/7 240/208/0/24/8 272/236/0/27/9 304/264/0/30/10 360/313/0/35/12 432/376/0/42/14 544/474/0/53/17 720/627/0/70/23 333/145/145/32/11 370/160/160/37/12 420/182/182/42/14 469/204/204/46/15 555/241/242/54/19 666/290/290/65/22 839/366/366/82/26 1111/483/484/108/35 total/data/ifec/safety/psi kbps@TP per time 333/193/97/32/11 370/213/108/37/12 420/242/122/42/14 469/272/136/46/15 555/321/162/54/19 666/386/194/65/22 839/488/244/82/26 1111/645/322/108/35 slice 333/290//32/11 370/321//37/12 420/364//42/14 469/407//46/15 555/483//54/19 666/580//65/22 839/731//82/26 1111/967//108/35 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66/ 1 0.5 / 0.66/ 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 Bitrate/service (kbps at IP level) 140 / 180 / 280 148 / 197 / 304 173 / 230 / 345 189 / 255 / 386 230 / 304 / 460 280 / 370 / 559 345 / 468 / 698 460 / 616 / 928 Table A10.7: OFDM 16QAM class 1 legacy configuration ETSI 355 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X - - X X - - Parameter/Case name QPSK_1/5_S QPSK_2/9_S QPSK_1/4_S QPSK_2/7_S QPSK_1/3_S QPSK_2/5_S QPSK_1/2_S QPSK_2/3_S PHYSICAL CONFIGURATION Bandwidth (MHz) 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation QPSK QPSK QPSK QPSK QPSK QPSK QPSK QPSK PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 211 / 211 0 / 211 0 / 211 0 / 211 0 / 211 0 / 211 0 / 211 0 / 211 MPEG TS total bit rate (Mbps) 1,333 1,481 1,679 1,876 2,222 2,666 3,357 4,443 TS LAYER CONFIGURATION  8 8 8 8 8 8 8 8 Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 122 122 122 122 122 122 122 122 OFDM Symbols/ coded FEC 20 18 16 14 12 10 8 6 108/46/47/11/4 120/52/52/12/4 136/58/59/14/5 152/66/66/15/5 180/78/78/18/6 216/94/94/21/7 272/118/118/27/9 360/156/157/35/12 total/data/ifec/safety/psi TP per time slice 108/62/31/11/4 120/69/35/12/4 136/78/39/14/5 152/88/44/15/5 180/104/52/18/6 216/125/63/21/7 272/157/79/27/9 360/208/105/35/12 108/93/0/11/4 120/104/0/12/4 136/117/0/14/5 152/132/0/15/5 180/156/0/18/6 216/188/0/21/7 272/236/0/27/9 360/313/0/35/12 333/145/145/32/11 167/71/73/17/6 185/80/80/19/6 210/89/91/22/8 235/102/102/23/8 278/120/120/28/9 420/182/182/42/14 555/241/242/54/19 total/data/ifec/safety/psi kbps@TP per time 333/193/97/32/11 167/96/48/17/6 185/106/54/19/6 210/120/60/22/8 235/136/68/23/8 278/160/80/28/9 420/242/122/42/14 555/321/162/54/19 slice 333/290//32/11 167/143//17/6 185/160//19/6 210/181//22/8 235/204//23/8 278/241//28/9 420/364//42/14 555/483//54/19 333 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66/ 1 0.5 / 0.66/ 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 Bitrate/service (kbps at IP level) 66 / 91 / 132 74 / 99 / 148 83 / 115 / 173 91 / 124 / 189 115 / 148 / 230 140 / 181 / 280 173 / 230 / 345 230 / 304 / 460 Table A10.8: OFDM QPSK class 1 configuration ETSI 356 ETSI TS 102 584 V1.1.2 (2010-XY) A.10.3 OFDM class 2 In order to synchronize the long physical interleaver with the slice structure, some rules must be followed. Interleaver depth = (nof_slices-1)*sh_distance+last_slice_depth Nof_slices Nof_slices-1 3 2 1 sh_distance sh_distance sh_distance sh_distance Last_slice_depth Late part SH_distance There is synchronization between physical interleaver and MAC layer when: ∈Ν repetition_interval ∃ an integer ' n' such that SH_distance = n * repetition_interval Since ∃ an integer ' k' such that SH_distance = k * SH_duration k ∃ (k; n ) such that repetition_interval = SH_duration * n Branch delay from receiver perspective ! Figure A10.11: correspondence between SH physical interleaver and TS configuration When such synchronization is performed between link and physical layers, this is signaled in one of the informative lines. In fact the only requirement is that SH distance is an exact multiple of repetition interval; therefore repetition interval is also a multiple of SH duration. The following logic is therefore thought while fixing the interleaver parameters: 1) search a physical interleaver configuration matching expected duration => this leads to a specific SH_distance value 2) fix the number of time slices so that individual slice can accommodate an individual IP flow 3) Define repetition_interval as an integer fraction of SH_distance best approaching a1 second interval. 4) Compute the burst duration as the division of repetition_interval by the number of time slices. ETSI 357 ETSI TS 102 584 V1.1.2 (2010-XY)                                      Figure A10.12: heuristic for repetition_interval determination                                Figure A10.13: heuristic example in the case of 16QAM ¼_U_10s For uniform-late, two types of modulations are investigated, QPSK and 16QAM. For each modulation we give either {fixed code rate; variable duration} or {variable code rate; fixed duration} configurations. In addition, some investigations on the size of the late part have led to consider different configurations. These are called “optimized UL configuration”. They do not replace legacy ones but are new configurations added for the sake of referencing, in particular by some field trials. ETSI 358 ETSI TS 102 584 V1.1.2 (2010-XY) Modulation Code rate Duration Table 16QAM 1/5 Variable (0.5 to 10 seconds) Table A10.11 Optimized interleaver Table A10.12 Variable (5 to 10 seconds) ¼ Variable (0.5 to 10 seconds) Table A10.13 Optimized interleaver Table A10.14 Variable (5 to 10 seconds) 1/3 Variable (0.5 to 10 seconds) Table A10.15 Optimized interleaver Table A10.16 Variable (5 to 10 seconds) Variable (1/5 to 2/5) 10 seconds Uniform late Table A10.17 Optimized interleaver Table A10.18 10 seconds Uniform late CR up to 1/3 only QPSK Variable (1/5 to 2/5) 10 seconds Uniform late Table A10.19 Optmized Table A10.20 10 seconds Uniform late CR up to 1/3 only 1/3 Variable (0.5 to 10 seconds) Table A10.21 Optimized Table A10.22 Variable (5 to 10 seconds) Table A10.9: OFDM class 2 uniform late physical layer configuration summary For uniform-long, two types of modulations are investigated, QPSK and 16QAM. For each modulation we give either {fixed code rate; variable duration} or {variable code rate; fixed duration} configurations. Modulation Code rate Duration Table 16QAM 1/5 Variable (0.5 to 10 seconds) Table A10.23 ¼ Variable (0.5 to 10 seconds) Table A10.24 1/3 Variable (0.5 to 10 seconds) Table A10.25 Variable (1/5 to 2/3) 10 seconds Uniform late Table A10.26 QPSK 1/3 Variable (0.5 to 10 seconds) Table A10.27 ½ Variable (0.5 to 10 seconds) Table A10.28 2/3 Variable (0.5 to 10 seconds) Table A10.29 Variable (1/5 to 2/3) 10 seconds Uniform late Table A10.30 Table A10.10: OFDM class 2 uniform long physical layer configuration summary ETSI 359 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 - X X X X Parameter/Case name 16QAM_1/5_UL_9.9s 16QAM_1/5_UL_5s 16QAM_1/5_UL_2.5s 16QAM_1/5_UL_1s 16QAM_1/5_UL_0.5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 5 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM PHY FEC rate 1/5  1/5 1/5 1/5 Interleaver configuration (common_multiplier/nof_late_taps/ 20/24/9/10/12 20/24/9/5/12 18/24/4/5/12 9/24/2/1/10 10/24/2/1/4 nof_slices/slice_distance/non_late_increment) Min / late / max interleaver duration (ms) 0 / 9964 0 / 5089 0 / 2505 0 / 1049 0 / 534 MPEG TS total bit rate 2,666 2,666 2,666 2,666 2,666 TS LAYER CONFIGURATION  9 9 9 9 9  YES NO NO NO NO Repetition interval (ms) 1218,6 1218,6 1218,6 1218,6 1218,6 Burst duration (ms) 135 135 135 135 135 OFDM Symbols/ coded FEC 10 10 10 10 10 total/data/ifec/safety/psi TP per time slice 240/208/0/24/8 240/208/0/24/8 240/208/0/24/8 240/208/0/24/8 240/208/0/24/8 total/data/ifec/safety/psi kbps@TP per time slice 296/257//30/10 296/257//30/10 296/257//30/10 296/257//30/10 296/257//30/10 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 243 243 243 243 243 Table A10.11: OFDM 16QAM1/5 Uniform Late variable duration ETSI 360 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 new to IG rel 1 - X Parameter/Case name 16QAM_1/5_UL_9.9s 16QAM_1/5_UL_5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM PHY FEC rate 1/5  Interleaver configuration (common_multiplier/nof_late_taps/ 10/12/10/9/4 10/12/7/7/2 nof_slices/slice_distance/non_late_increment) Min / late / max interleaver duration (ms) 0 / 49 / 9924 0 / 49 / 5163 MPEG TS total bit rate 2,666 2,666 TS LAYER CONFIGURATION  9 9  YES NO Repetition interval (ms) 1097 1219 Burst duration (ms) 122 135 OFDM Symbols/ coded FEC 10 10 total/data/ifec/safety/psi TP per time slice 216/125/63/21/7 240/138/70/24/8 total/data/ifec/safety/psi kbps@TP per time slice 296/258//29/10 296/257//30/10 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 Max bitrate/service (kbps at IP level) 243 243 Table A10.12: OFDM 16QAM1/5 optimized Uniform Late variable duration ETSI 361 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name 16QAM_1/4_UL_9.9s 16QAM_1/4_UL_5s 16QAM_1/4_UL_2.5s 16QAM_1/4_UL_1s 16QAM_1/4_UL_0.5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 5 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM PHY FEC rate 1/4  1/4 1/4 1/4 Interleaver configuration (common_multiplier/nof_late_taps/ 20/24/9/10/12 20/24/9/5/12 18/24/4/5/12 9/24/2/1/10 10/24/2/1/4 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 206 / 9964 0 / 206 / 9964 0 / 206 / 9964 0 / 206 / 9964 0 / 206 / 9964 MPEG TS total bit rate 3,357 3,357 3,357 3,357 3,357 TS LAYER CONFIGURATION  9 9 9 9 9  YES NO NO NO NO Repetition interval (ms) 1218,6 1218,6 1218,6 1218,6 1218,6 Burst duration (ms) 135 135 135 135 135 OFDM Symbols/ coded FEC 8 8 8 8 8 total/data/ifec/safety/psi TP per time slice 302/262/0/30/10 302/262/0/30/10 302/262/0/30/10 302/262/0/30/10 302/262/0/30/10 total/data/ifec/safety/psi kbps@TP per time sliceTP 373/323//37/12 373/323//37/12 373/323//37/12 373/323//37/12 373/323//37/12 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 309 309 309 309 309 Table A10.13: OFDM 16QAM1/4 Uniform Late variable duration ETSI 362 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 new to IG rel 1 X X Parameter/Case name 16QAM_1/4_UL_9.9s 16QAM_1/4_UL_5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM PHY FEC rate 1/4 1/4 Interleaver configuration (common_multiplier/nof_late_taps/ 10/16/9/10/6 10/16/9/5/10 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 67 / 9829 0 / 67 / 4955 MPEG TS total bit rate 3,357 3,357 TS LAYER CONFIGURATION Services 9 9 Synchronization LINK/PHY YES NO Repetition interval (ms) 1218,6 1218,6 Burst duration (ms) 135 135 OFDM Symbols/ coded FEC 8 8 total/data/ifec/safety/psi TP per time slice 302/262/0/30/10 302/262/0/30/10 total/data/ifec/safety/psi kbps@TP per time slice 373/323//37/12 373/323//37/12 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 309 309 Table A10.14: OFDM 16QAM1/4 optimized Uniform Late variable duration ETSI 363 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name 16QAM_1/3_UL_9.9s 16QAM_1/3_UL_5s 16QAM_1/3_UL_2.5s 16QAM_1/3_UL_1s 16QAM_1/3_UL_0.5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 5 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM PHY FEC rate 1/3  1/3 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 20/24/9/10/12 20/24/9/5/12 18/24/4/5/12 9/24/2/1/10 10/24/2/1/4 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 206 / 9964 0 / 5089 0 / 2505 0 / 1049 0 / 534 MPEG TS total bit rate 4,443 4,443 4,443 4,443 4,443 TS LAYER CONFIGURATION  9 9 9 9 9  YES NO NO NO NO Repetition interval (ms) 1218,6 1218,6 1218,6 1218,6 1218,6 Burst duration (ms) 135 135 135 135 135 OFDM Symbols/ coded FEC 6 6 6 6 6 total/data/ifec/safety/psi TP per time slice 400/348/0/39/13 400/348/0/39/13 400/348/0/39/13 400/348/0/39/13 400/348/0/39/13 total/data/ifec/safety/psi kbps@TP per time slice 494/430//48/16 494/430//48/16 494/430//48/16 494/430//48/16 494/430//48/16 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 408 408 408 408 408 Table A10.15: OFDM 16QAM1/3 Uniform Late variable duration ETSI 364 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 new to IG rel 1 X X Parameter/Case name 16QAM_1/3_UL_9.9s 16QAM_1/3_UL_5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM PHY FEC rate 1/3  Interleaver configuration (common_multiplier/nof_late_taps/ 10/18/11/8/9 10/18/6/8/4 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 76 / 9829 0 / 76/ 4964 MPEG TS total bit rate 4,443 4,443 TS LAYER CONFIGURATION  9 9  NO NO Repetition interval (ms) 1218,6 1218,6 Burst duration (ms) 135 135 OFDM Symbols/ coded FEC 6 6 total/data/ifec/safety/psi TP per time slice 400/348/0/39/13 400/348/0/39/13 total/data/ifec/safety/psi kbps@TP per time slice 494/430//48/16 494/430//48/16 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 408 408 Table A10.16: OFDM 16QAM1/3 optimized Uniform Late variable duration ETSI 365 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 new to IG rel 1 - X X X X X 16QAM_1/5_ Parameter/Case name 16QAM_2/9_UL 16QAM_1/4_UL 16QAM_2/7_UL 16QAM_1/3_UL 16QAM_2/5_UL UL PHYSICAL CONFIGURATION Bandwidth (MHz) 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 Interleaver configuration (common_multiplier/nof_late_taps/ 20/24/9/10/12 20/24/9/10/12 20/24/9/10/12 20/24/9/10/12 20/24/9/10/12 20/24/9/10/12 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 206 / 9964 0 / 206 / 9964 0 / 206 / 9964 0 / 206 / 9964 0 / 206 / 9964 0 / 9964 MPEG TS total bit rate 2,666 2,962 3,357 3,752 4,443 5,332 TS LAYER CONFIGURATION Services 9 9 9 9 9 9 Synchronization PHY / LINK YES YES YES YES YES YES Repetition interval (ms) 1218,6 1219 1219 1219 1219 1219 Burst duration (ms) 135 135 135 135 135 135 OFDM Symbols/ coded FEC 10 9 8 7 6 5 total/data/ifec/safety/psi TP per time slice 240/208/0/24/8 266/231/0/26/9 302/262/0/30/10 337/293/0/33/11 400/348/0/39/13 480 total/data/ifec/safety/psi kbps@TP per time slice 296/257//30/10 328/285//32/11 373/323//37/12 416/362//41/14 494/430//48/16 592 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 243 270 309 348 408 480 Table A10.17: OFDM 16QAM variable code rate Uniform Late 10 seconds NB: code rates ½ and 2/3 are not included since the UL configuration is not useful because late part code rate would be ≥ 1. ETSI 366 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 - X X X X 16QAM_1/5_ Parameter/Case name 16QAM_2/9_UL 16QAM_1/4_UL 16QAM_2/7_UL 16QAM_1/3_UL UL PHYSICAL CONFIGURATION Bandwidth (MHz) 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM PHY FEC rate 1/5 2/9 1/4 2/7 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 10/12/10/9/4 10/15/12/7/8 10/16/9/10/6 10/16/9/10/6 10/18/11/8/9 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 49 / 9924 0 / 63 / 9455 0 / 67 / 9829 0 / 67 / 9829 0 / 76 / 9829 MPEG TS total bit rate 2,666 2,962 3,357 3,752 4,443 TS LAYER CONFIGURATION  9 9 9 9 9  NO NO YES YES NO Repetition interval (ms) 1218,6 1219 1219 1219 1219 Burst duration (ms) 135 135 135 135 135 OFDM Symbols/ coded FEC 10 9 8 7 6 total/data/ifec/safety/psi TP per time slice 240/208/0/24/8 266/231/0/26/9 302/262/0/30/10 337/293/0/33/11 400/348/0/39/13 total/data/ifec/safety/psi kbps@TP per time slice 296/257//30/10 328/285//32/11 373/323//37/12 416/362//41/14 494/430//48/16 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 243 270 309 348 408 Table A10.18: OFDM 16QAM variable code rate optimized Uniform Late 10 seconds NB: code rates 2/5, ½ and 2/3 are not included since the UL configuration is not useful because late part code rate would be ≥ 1. ETSI 367 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 new to IG rel 1 X X X X - X Parameter/Case name QPSK_1/5_UL QPSK_2/9_UL QPSK_1/4_UL QPSK_2/7_UL QPSK_1/3_UL QPSK_2/5_UL PHYSICAL CONFIGURATION Bandwidth (MHz) 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation QPSK QPSK QPSK QPSK QPSK QPSK PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 Interleaver configuration (common_multiplier/nof_late_taps/ 10/24/9/5/12 10/24/9/5/12 10/24/9/5/12 10/24/9/5/12 10/24/9/5/12 10/24/9/5/12 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 206 / 9964 0 / 206 / 9964 0 / 206 / 9964 0 / 206 / 9964 0 / 206 / 9964 0 / 206 / 9964 MPEG TS total bit rate (Mbps) 1,333 1,481 1,679 1,876 2,222 2,666 TS LAYER CONFIGURATION  8 8 8 8 8 8  NO NO NO NO NO NO Repetition interval (ms) 1219 1219 1219 1219 1219 1219 Burst duration (ms) 152 152 152 152 152 152 OFDM Symbols/ coded FEC 20 18 16 14 12 10 total/data/ifec/safety/psi TP per time slice 135/117/0/13/5 150/130/0/15/5 170/147/0/17/6 190/165/0/19/6 225/195/0/22/8 270/234/0/27/9 total/data/ifec/safety/psi kbps@TP per time 167/144//16/6 185/160//19/6 210/181//21/7 235/204//23/7 278/241//27/10 333/289//33/11 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 138 185 171 191 230 276 Table A10.19: OFDM QPSK variable code rate Uniform Late 10 seconds NB: code rates ½ and 2/3 are not included since the UL configuration is not useful because late part code rate would be ≥ 1. ETSI 368 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X - Parameter/Case name QPSK_1/5_UL QPSK_2/9_UL QPSK_1/4_UL QPSK_2/7_UL QPSK_1/3_UL PHYSICAL CONFIGURATION Bandwidth (MHz) 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation QPSK QPSK QPSK QPSK QPSK PHY FEC rate 1/5 2/9 1/4 2/7 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 5/12/10/5/4 5/15/12/4/7 5/16/9/5/6 5/16/9/5/6 5/18/11/4/6 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 49 / 11021 0 / 63 / 10786 0 / 67 / 9829 0 / 67 / 9829 0 / 76 / 9802 MPEG TS total bit rate (Mbps) 1,333 1,481 1,679 1,876 2,222 TS LAYER CONFIGURATION  8 8 8 8 8  NO NO NO NO NO Repetition interval (ms) 1219 1219 1219 1219 1219 Burst duration (ms) 152 152 152 152 152 OFDM Symbols/ coded FEC 20 18 16 14 12 total/data/ifec/safety/psi TP per time slice 135/117/0/13/5 150/130/0/15/5 170/147/0/17/6 190/165/0/19/6 225/195/0/22/8 total/data/ifec/safety/psi kbps@TP per time 167/144//16/6 185/160//19/6 210/181//21/7 235/204//23/7 278/241//27/10 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 138 185 171 191 230 Table A10.20: OFDM QPSK variable code rate optimized Uniform Late 10 seconds ETSI 369 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 - X X X X Parameter/Case name QPSK_1/3_UL_9.9s QPSK_1/3_UL_5.1s QPSK_1/3_UL_2.1s QPSK_1/3_UL_1s QPSK_1/3_UL_0.5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 5 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation QPSK QPSK QPSK QPSK QPSK PHY FEC rate 1/3 1/3 1/3 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 10/24/9/5/12 10/24/3/8/12 10/24/9/1/10 23/48/1/0/0 12/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 206 / 9964 0 / 206 / 5082 0 / 206 / 2129 0 / 969 / 969 0 / 505 / 505 MPEG TS total bit rate (Mbps) 2,222 2,222 2,222 2,222 2,222 TS LAYER CONFIGURATION Services 8 8 8 8 8 Synchronization PHY/LINK NO NO NO NO NO Repetition interval (ms) 1219 1219 1219 1219 1219 Burst duration (ms) 152 152 152 152 152 OFDM Symbols/ coded FEC 12 12 12 12 12 total/data/ifec/safety/psi TP per time slice 225/195/0/22/8 225/195/0/22/8 225/195/0/22/8 225/195/0/22/8 225/195/0/22/8 total/data/ifec/safety/psi kbps@TP per time 278/241//27/10 278/241//27/10 278/241//27/10 278/241//27/10 278/241//27/10 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 230 230 230 230 230 Table A10.21: OFDM QPSK1/3 Uniform Late variable duration ETSI 370 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 new to IG rel 1 - X Parameter/Case name QPSK_1/3_UL_9.9s QPSK_1/3_UL_5.1s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ Modulation QPSK QPSK PHY FEC rate 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 5/18/11/4/6 5/18/6/4/4 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 76 / 9802 0 / 76 / 4964 MPEG TS total bit rate (Mbps) 2,222 2,222 TS LAYER CONFIGURATION Services 8 8 Synchronization PHY/LINK NO NO Repetition interval (ms) 1219 1219 Burst duration (ms) 152 152 OFDM Symbols/ coded FEC 12 12 total/data/ifec/safety/psi TP per time slice 225/195/0/22/8 225/195/0/22/8 total/data/ifec/safety/psi kbps@TP per time 278/241//27/10 278/241//27/10 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 230 230 Table A10.22: OFDM QPSK1/3 optimized Uniform Late variable duration ETSI 371 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 - X X X X Parameter/Case name 16QAM_1/5_U_11s 16QAM_1/5_U_5s 16QAM_1/5_U_2.5s 16QAM_1/5_U_1s 16QAM_1/5_U_0.5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 5 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM PHY FEC rate 1/5 1/5 1/5 1/5 1/5 Interleaver configuration (common_multiplier/nof_late_taps/ 40/0/12/8/4 40/0/6/8/4 10/0/4/6/7 48/48/1/0/0 24/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 0 / 10938 0 / 0 / 5376 0 / 0 / 2538 0 / 0 / 1011 0 / 505 / 505 MPEG TS total bit rate 2,666 2,666 2,666 2,666 2,666 TS LAYER CONFIGURATION  9 9 9 9 9 Synchronization PHY/LINK YES YES NO YES YES Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 OFDM Symbols/ coded FEC 10 10 10 10 10 total/data/ifec/safety/psi TP per time slice 192/167/0/19/6 192/167/0/19/6 192/167/0/19/6 192/167/0/19/6 192/167/0/19/6 total/data/ifec/safety/psi kbps@TP per time 296/258//29/9 296/258//29/9 296/258//29/9 296/258//29/9 296/258//29/9 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 247 247 247 247 247 Table A10.23: OFDM 16QAM1/5 Uniform Long variable duration ETSI 372 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name 16QAM_1/4_U_11s 16QAM_1/4_U_5s 16QAM_1/4_U_2.5s 16QAM_1/4_U_1s 16QAM_1/4_U_0.5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 5 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM PHY FEC rate 1/4  1/4 1/4 ¼ Interleaver configuration (common_multiplier/nof_late_taps/ 40/0/12/8/4 40/0/6/8/4 10/0/4/6/7 48/48/1/0/0 24/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 0 / 10938 0 / 0 / 5376 0 / 0 / 2538 0 / 1011 / 1011 0 / 505 / 505 MPEG TS total bit rate 3,357 3,357 3,357 3,357 3,357 TS LAYER CONFIGURATION  9 9 9 9 9  YES YES NO YES YES Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 OFDM Symbols/ coded FEC 8 8 8 8 8 total/data/ifec/safety/psi TP per time slice 241/209/0/24/8 241/209/0/24/8 241/209/0/24/8 241/209/0/24/8 241/209/0/24/8 total/data/ifec/safety/psi kbps@TP per time slice 372/322//37/12 372/322//37/12 372/322//37/12 372/322//37/12 372/322//37/12 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 304 304 304 304 304 Table A10.24: OFDM 16QAM1/4 Uniform Long variable duration ETSI 373 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name 16QAM_1/3_U_9.9s 16QAM_1/3_U_5s 16QAM_1/3_U_2.5s 16QAM_1/3_U_1s 16QAM_1/3_U_0.5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 5 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16 QAM 16 QAM 16 QAM 16 QAM 16 QAM PHY FEC rate 1/3  1/3 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 40/0/12/8/4 40/0/6/8/4 10/0/4/6/7 48/48/1/0/0 24/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 0 / 10938 0 / 0 / 5376 0 / 0 / 2538 0 / 1011 / 1011 0 / 505 / 505 MPEG TS total bit rate 4,443 4,443 4,443 4,443 4,443 TS LAYER CONFIGURATION  9 9 9 9 9  YES YES NO YES YES Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 OFDM Symbols/ coded FEC 6 6 6 6 6 total/data/ifec/safety/psi TP per time slice 320/186/93/31/10 320/186/93/31/10 320/186/93/31/10 320/186/93/31/10 320/186/93/31/10 total/data/ifec/safety/psi kbps@TP per time slice 494/287/143/48/15 494/287/143/48/15 494/287/143/48/15 494/287/143/48/15 494/287/143/48/15 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 411 411 411 411 411 Tab le A10.25: OFDM 16QAM1/3 Uniform Long variable duration ETSI 374 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 - X X X X X X X Parameter/Case name 16QAM_1/5_U 16QAM _2/9_U 16QAM _1/4_U 16QAM _2/7_U 16QAM _1/3_U 16QAM _2/5_U 16QAM _1/2_U 16QAM _2/3_U PHYSICAL CONFIGURATION Bandwidth (MHz) 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation 16QAM 16QAM 16QAM 16QAM 16QAM 16QAM 16QAM 16QAM PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 40/0/12/8/4 40/0/12/8/4 40/0/12/8/4 40/0/12/8/4 40/0/12/8/4 40/0/12/8/4 40/0/12/8/4 40/0/12/8/4 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 MPEG TS total bit rate (Mbps) 2,666 2,962 3,357 3,752 4,443 5,332 6,714 8,887 TS LAYER CONFIGURATION  9 9 9 9 9 9 9 9 Synchronization PHY/LINK YES YES YES YES YES YES YES YES Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 108 108 108 OFDM Symbols/ coded FEC 10 9 8 7 6 5 4 3 total/data/ifec/safety/psi TP per time slice 192/167/0/19/6 213/185/0/21/7 241/209/0/24/8 270/234/0/27/9 320/279/0/31/10 384/334/0/38/12 483/420/0/47/16 640/558/0/62/20 total/data/ifec/safety/psi kbps@TP per time 296/258//29/9 329/285//32/11 372/322//37/12 417/361//42/14 494/430//48/15 592/515//59/19 745/648//73/25 987/861//96/31 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 247 271 304 345 411 493 616 821 Table A10.26: OFDM 16QAM variable code rate Uniform Long 10 seconds ETSI 375 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 - X X X X Parameter/Case name QPSK_1/3_U_11s QPSK_1/3_U_5.5s QPSK_1/3_U_2.3s QPSK_1/3_U_1s QPSK_1/3_U_0.5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 5 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation QPSK QPSK QPSK QPSK QPSK PHY FEC rate 1/3  1/3 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 40/0/12/4/2 40/0/12/2/2 10/0/4/3/2 24/48/1/0/0 12/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 /10938 0 / 0 / 5577 0 / 0 / 2391 0 / 1011 / 1011 0 / 505 /505 MPEG TS total bit rate (Mbps) 2,222 2,222 2,222 2,222 2,222 TS LAYER CONFIGURATION  8 8 8 8 8 Synchronization LINK PHY YES NO NO YES YES Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 122 122 122 122 122 OFDM Symbols/ coded FEC 12 12 12 12 12 total/data/ifec/safety/psi TP per time slice 180/156/0/18/6 180/156/0/18/6 180/156/0/18/6 180/156/0/18/6 180/156/0/18/6 total/data/ifec/safety/psi kbps@TP per time 278/241//28/9 278/241//28/9 278/241//28/9 278/241//28/9 278/241//28/9 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 230 230 230 230 230 Table A10.27: OFDM QPSK1/3 Uniform Long variable duration ETSI 376 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name QPSK_1/2_U_9.9s QPSK_1/2_U_5.1s QPSK_1/2_U_2.1s QPSK_1/2_U_1s QPSK_1/2_U_0.5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 5 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation QPSK QPSK QPSK QPSK QPSK PHY FEC rate 1/2 1/2 1/2 1/2  Interleaver configuration (common_multiplier/nof_late_taps/ 40/0/12/4/2 40/0/12/2/2 10/0/4/3/2 24/48/1/0/0 12/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10938 0 / 0 / 5577 0 / 0 / 2391 0 / 1011 / 1011 0 / 505 / 505 MPEG TS total bit rate (Mbps) 3,357 3,357 3,357 3,357 3,357 TS LAYER CONFIGURATION  8 8 8 8 8  YES NO NO YES YES Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 122 122 122 122 122 OFDM Symbols/ coded FEC 8 8 8 8 8 total/data/ifec/safety/psi TP per time slice 272/236/0/27/9 272/236/0/27/9 272/236/0/27/9 272/236/0/27/9 272/236/0/27/9 total/data/ifec/safety/psi kbps@TP per time 420/364//42/14 420/364//42/14 420/364//42/14 420/364//42/14 420/364//42/14 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 345 345 345 345 345 Table A10.28: OFDM QPSK1/2 Uniform Long variable duration ETSI 377 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name QPSK_2/3_U_9.9s QPSK_2/3_U_5.1s QPSK_2/3_U_2.1s QPSK_2/3_U_1s QPSK_2/3_U_0.5s PHYSICAL CONFIGURATION Bandwidth (MHz) 5 5 5 5 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation QPSK QPSK QPSK QPSK QPSK PHY FEC rate 2/3 2/3 2/3 2/3  Interleaver configuration (common_multiplier/nof_late_taps/ 40/0/12/4/2 40/0/12/2/2 10/0/4/3/2 24/48/1/0/0 12/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10938 0 / 0 / 5577 0 / 0 / 2391 0 / 1011 / 1011 0 / 505 / 505 MPEG TS total bit rate (Mbps) 4,443 4,443 4,443 4,443 4,443 TS LAYER CONFIGURATION  8 8 8 8 8  YES NO NO YES YES Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 122 122 122 122 122 OFDM Symbols/ coded FEC 6 6 6 6 6 total/data/ifec/safety/psi TP per time slice 360/313/0/35/12 360/313/0/35/12 360/313/0/35/12 360/313/0/35/12 360/313/0/35/12 total/data/ifec/safety/psi kbps@TP per time 555/483//54/19 555/483//54/19 555/483//54/19 555/483//54/19 555/483//54/19 slice MPEG2 per service bit rate (kbps at TS level) 555 555 555 555 555 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 460 460 460 460 460 Table A10.29: OFDM QPSK2/3 Uniform Long variable duration ETSI 378 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X X X - X X X Parameter/Case name QPSK_1/5_U QPSK_2/9_U QPSK_1/4_U QPSK_2/7_U QPSK_1/3_U QPSK_2/5_U QPSK_1/2_U QPSK_2/3_U PHYSICAL CONFIGURATION Bandwidth (MHz) 5 FFT Mode 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ 2K+GI ¼ Modulation QPSK QPSK QPSK QPSK QPSK QPSK QPSK QPSK PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 0 / 0 / 10938 MPEG TS total bit rate (Mbps) 1,333 1,481 1,679 1,876 2,222 2,666 3,357 4,443 TS LAYER CONFIGURATION  8 8 8 8 8 8 8 8 Synchronization PHY/LINK YES YES YES YES YES YES YES YES Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 122 122 122 122 122 122 122 122 OFDM Symbols/ coded FEC 20 18 16 14 12 10 8 6 total/data/ifec/safety/psi kbps@TP per time 108/93/0/11/4 120/104/0/12/4 136/117/0/14/5 152/132/0/15/5 180/156/0/18/6 216/188/0/21/7 272/236/0/27/9 360/313/0/35/12 slice MPEG2 per service bit rate (kbps at TS level) 167/143//17/6 185/160//19/6 210/181//22/8 235/204//23/8 278/241//28/9 333/290//32/11 420/364//42/14 555/483//54/19 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 132 148 173 189 230 280 345 460 Table A10.30: OFDM QPSK variable code rate Uniform Long 10 seconds ETSI 379 ETSI TS 102 584 V1.1.2 (2010-XY) A.10.3 TDM class 1 We have the same issue as for OFDM regarding the memory available for the class 1 with two types of interleaver, the official one respecting the class 1 memory and the legacy one corresponding to the implementation guidelines version 1. Please refer to OFDM class 1 explanation. Modulation Code rate Duration Table T-QPSK Variable (1/5 to 2/3) Short (official) Table A10.32 8PSK Variable (1/5 to 2/3) Short (official) Table A10.33 Short (legacy) Table A10.34 16APSK Variable (1/5 to 2/3) Short (official) Table A10.35 Short (legacy) Table A10.36 Table A10.31: TDM class 1 physical layer configuration summary ETSI 380 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X X X X X - X Parameter/Case name T_QPSK_1/5_S T_QPSK_2/9_S T_QPSK_1/4_S T_QPSK_2/7_S T_QPSK_1/3_S T_QPSK_2/5_S T_QPSK_1/2_S T_QPSK_2/3_S PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; (TDM bandwidth, TDM_modulation; roll- 5MHz, QPSK ; 0,15 5MHz, QPSK ; 0,15 0,15 0,15 0,15 0,15 0,15 0,15 off) PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 nof_slices/slice_distance/non_late_increment ) Min / Late / Max interleaver duration (ms) 0 / 185 / 185 0 / 182 / 182 0 / 184 / 184 0 / 182 / 182 0 / 184 / 184 0 / 182 / 182 0 / 181 / 181 0 / 185 / 185 MPEG TS total bit rate (Mbps) 1,530 1,728 1,925 2,222 2,567 3,110 3,900 5,184 TS LAYER CONFIGURATION  8 8 8 8 8 8 8 8 Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 122 122 122 122 122 122 122 122 TDM Symbols/ coded FEC 30240 27216 24192 21168 18144 15120 12096 9072 124/54/54/12/4 140/60/61/14/5 156/67/68/16/5 180/78/78/18/6 208/90/90/21/7 252/109/110/25/8 316/137/138/31/10 420/182/183/41/14 total/data/ifec/safety/psi kbps@TP per time 124/72/36/12/4 140/80/41/14/5 156/90/45/16/5 180/104/52/18/618 208/120/60/21/7 252/146/73/25/8 316/183/92/31/10 420/243/122/41/14 slice 124/108/0/12/4 140/121/0/14/5 156/135/0/16/5 0/156/0/18/6 208/180/0/21/7 252/219/0/25/8 316/275/0/31/10 420/365/0/41/14 191/83/83/19/6 216/93/94/22/8 241/103/105/25/8 278/120/120/28/9 321/139/139/32/11 389/168/170/39/12 488/211/213/48/15 648/281/282/63/22 MPEG2 per service bit rate (kbps at TS level) 191/111/56/19/6 216/123/63/22/8 241/139/69/25/8 278/160/80/28/9 321/185/93/32/11 389/225/113/39/12 488/282/142/48/15 648/375/188/63/22 191/167//19/6 216/187//22/8 241/208//25/8 278/241//28/9 321/278//32/11 389/338//39/12 488/424//48/15 648/563//63/22 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 05 / 0.66 / 1 05 / 0.66 / 1 05 / 0.66 / 1 05 / 0.66 / 1 05 / 0.66 / 1 05 / 0.66 / 1 05 / 0.66 / 1 05 / 0.66 / 1 Bitrate/service (kbps at IP level) 74/ 107 / 156 83 / 115 / 173 99 / 132 / 197 115 / 148 / 230 132 / 173 / 263 156 / 214 / 321 197 / 271 / 403 263 / 353 / 542 Table A10.32: T-QPSK short interleaver ETSI 381 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X X X X X X- X Parameter/Case name T_8PSK_1/5_S T_8PSK_2/9_S T_8PSK_1/4_S T_8PSK_2/7_S T_8PSK_1/3_S T_8PSK_2/5_S T_8PSK_1/2_S T_8PSK_2/3_S PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 0,15 5MHz, QPSK ; 0,15 (TDM bandwidth, TDM_modulation; roll-off) 0,15 0,15 0,15 0,15 0,15 0,15 PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 122/ 122 0 / 122/ 122 0 / 121 / 121 0 / 122 / 122 0 / 121 / 121 0 / 121 / 121 0 / 121 / 121 0 / 121 / 121 MPEG TS total bit rate (Mbps) 2,320 2,567 2,913 3,308 3,900 4,690 5,826 7,800 TS LAYER CONFIGURATION  9 9 9 9 9 9 9 9 Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 108 108 108 TDM Symbols/ coded FEC 20160 18144 16128 14112 12096 10080 8064 6048 167/72/72/17/6 184/80/80/18/6 209/90/91/21/7 238/103/104/23/8 280/121/122/28/9 337/146/147/33/11 419/182/182/41/14 561/244/244/55/18 total/data/ifec/safety/psi kbps@TP per time 167/96/48/17/6 184/106/54/18/6 209/120/61/21/7 238/138/69/23/8 280/162/81/28/9 337/195/98/33/11 419/242/122/41/14 561/325/163/55/18 slice 167/144/0/17/6 184/160/0/18/6 209/181/0/21/7 238/207/0/23/8 280/243/0/28/9 337/293/0/33/11 419/364/0/41/14 561/488/0/55/18 258/111/111/26/9 284/123/123/28/9 322/139/140/32/11 367/159/160/35/12 432/187/188/43/14 520/225/227/51/17 646/281/281/63/22 866/376/376/85/28 MPEG2 per service bit rate (kbps at TS level) 258/148/74/26/9 284/164/83/28/9 322/185/94/32/11 367/213/106/35/12 432/250/125/43/14 520/301/151/51/17 646/373/188/63/22 866/501/251/85/28 258/222//26/9 284/247//28/9 322/279//32/11 367/319//35/12 432/375//43/14 520/452//51/17 646/562//63/22 866/753//85/28 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 Bitrate/service (kbps at IP level) 107 / 140 / 214 0.5 / 0.66 / 1 132 / 173 / 263 148 / 197 / 304 173 / 238 / 353 214 / 288 / 435 263 / 353 / 534 362 / 476 / 723 Table A10.33: T-8PSK short interleaver ETSI 382 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X X X -1 X X- X Parameter/Case name T_8PSK_1/5_S T_8PSK_2/9_S T_8PSK_1/4_S T_8PSK_2/7_S T_8PSK_1/3_S T_8PSK_2/5_S T_QPSK_1/2_S T_QPSK_2/3_S PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 5MHz, QPSK ; 0,15 (TDM bandwidth, TDM_modulation; roll-off) 0,15 0,15 0,15 0,15 0,15 0,15 0,15 PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 9/48/1/0/0 9/48/1/0/0 9/48/1/0/0 9/48/1/0/0 9/48/1/0/0 9/48/1/0/0 9/48/1/0/0 9/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 219 / 219 0 / 220 / 220 0 / 218 / 218 0 / 220 / 200 0 / 217 / 217 0 / 217 / 217 0 / 218 / 218 0 / 217 / 217 MPEG TS total bit rate (Mbps) 2,320 2,567 2,913 3,308 3,900 4,690 5,826 7,800 TS LAYER CONFIGURATION  9 9 9 9 9 9 9 9 Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 108 108 108 TDM Symbols/ coded FEC 20160 18144 16128 14112 12096 10080 8064 6048 167/72/72/17/6 184/80/80/18/6 209/90/91/21/7 238/103/104/23/8 280/121/122/28/9 337/146/147/33/11 419/182/182/41/14 561/244/244/55/18 total/data/ifec/safety/psi kbps@TP per time 167/96/48/17/6 184/106/54/18/6 209/120/61/21/7 238/138/69/23/8 280/162/81/28/9 337/195/98/33/11 419/242/122/41/14 561/325/163/55/18 slice 167/144/0/17/6 184/160/0/18/6 209/181/0/21/7 238/207/0/23/8 280/243/0/28/9 337/293/0/33/11 419/364/0/41/14 561/488/0/55/18 258/111/111/26/9 284/123/123/28/9 322/139/140/32/11 367/159/160/35/12 432/187/188/43/14 520/225/227/51/17 646/281/281/63/22 866/376/376/85/28 MPEG2 per service bit rate (kbps at TS level) 258/148/74/26/9 284/164/83/28/9 322/185/94/32/11 367/213/106/35/12 432/250/125/43/14 520/301/151/51/17 646/373/188/63/22 866/501/251/85/28 258/222//26/9 284/247//28/9 322/279//32/11 367/319//35/12 432/375//43/14 520/452//51/17 646/562//63/22 866/753//85/28 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 Bitrate/service (kbps at IP level) 107 / 140 / 214 0.5 / 0.66 / 1 132 / 173 / 263 148 / 197 / 304 173 / 238 / 353 214 / 288 / 435 263 / 353 / 534 362 / 476 / 723 Table A10.34: T-8PSK short interleaver legacy 1 This configuration differs slightly from the IGrel1 one which was (8:48;1;0;0); however its status is set to be already covered by IGrel1. ETSI 383 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X X X X X X- X Parameter/Case name 16APSK_1/5_S 16APSK_2/9_S 16APSK_1/4_S 16APSK_2/7_S 16APSK_1/3_S 16APSK_2/5_S 16APSK_1/2_S 16APSK_2/5_S PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; (TDM bandwidth, TDM_modulation; roll-off) 0,15 0,15 0,15 0,15 0,15 0,15 0,15 0,15 PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 5/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 91/ 91 0 / 91 / 91 0 / 91 / 91 0 / 91 0 / 91 0 / 91 / 91 0 / 91 / 91 0 / 90 MPEG TS total bit rate (Mbps) 3,110 3,456 3,900 4,443 5,184 6,221 7,800 10,417 TS LAYER CONFIGURATION  12 12 12 12 12 12 12 12 Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 81 81 81 81 81 81 81 81 TDM Symbols/ coded FEC 15120 13608 12096 10584 9072 7560 6048 4536 168/72/73/17/6 186/81/81/18/6 210/91/91/21/7 240/104/104/24/8 280/121/122/28/9 336/146/146/33/11 421/183/183/41/14 562/244/245/55/18 total/data/ifec/safety/psi kbps@TP per time 168/96/49/17/6 186/108/54/18/6 210/121/61/21/7 240/138/70/24/8 280/162/81/28/9 336/194/98/33/11 421/244/122/41/14 562/326/163/55/18 slice 168/145/0/17/6 186/162/0/18/6 210/182/0/21/7 240/208/0/24/8 280/243/0/28/9 336/292/0/33/11 421/366/0/41/14 562/489/0/55/18 259/111/113/26/9 287/125/125/28/9 324/140/140/32/11 370/160/160/37/12 432/187/188/43/14 518/225/225/51/17 650/282/282/63/22 867/376/378/85/28 MPEG2 per service bit rate (kbps at TS level) 259/148/76/26/9 287/167/83/28/9 324/187/94/32/11 370/213/108/37/12 432/250/125/43/14 518/299/151/51/17 650/376/188/63/22 867/503/251/85/28 259/224//26/9 287/250//28/9 324/281//32/11 370/321//37/12 432/375//43/14 518/450//51/17 650/565//63/22 867/754//85/28 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 Bitrate/service (kbps at IP level) 107 / 140 / 214 115 / 156 / 238 132 / 173 / 263 148 / 197 / 304 173 / 238 / 353 214 / 288 / 427 271 / 362 / 542 362 / 476 / 723 Table A10.35: T-16PSK short interleaver legacy ETSI 384 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X X X X X X- X Parameter/Case name 16APSK_1/5_S 16APSK_2/9_S 16APSK_1/4_S 16APSK_2/7_S 16APSK_1/3_S 16APSK_2/5_S 16APSK_1/2_S 16APSK_2/5_S PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 5MHz, 16APSK ; 0,15 (TDM bandwidth, TDM_modulation; roll-off) 0,15 0,15 0,15 0,15 0,15 0,15 0,15 PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 12/48/1/0/0 12/48/1/0/0 12/48/1/0/0 12/48/1/0/0 12/48/1/0/0 12/48/1/0/0 12/48/1/0/0 12/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 218 0 / 218 0 / 217 0 / 218 0 / 218 0 / 218 0 / 217 0 / 217 MPEG TS total bit rate (Mbps) 3,110 3,456 3,900 4,443 5,184 6,221 7,800 10,417 TS LAYER CONFIGURATION  12 12 12 12 12 12 12 12 Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 81 81 81 81 81 81 81 81 TDM Symbols/ coded FEC 15120 13608 12096 10584 9072 7560 6048 4536 168/72/73/17/6 186/81/81/18/6 210/91/91/21/7 240/104/104/24/8 280/121/122/28/9 336/146/146/33/11 421/183/183/41/14 562/244/245/55/18 total/data/ifec/safety/psi kbps@TP per time 168/96/49/17/6 186/108/54/18/6 210/121/61/21/7 240/138/70/24/8 280/162/81/28/9 336/194/98/33/11 421/244/122/41/14 562/326/163/55/18 slice 168/145/0/17/6 186/162/0/18/6 210/182/0/21/7 240/208/0/24/8 280/243/0/28/9 336/292/0/33/11 421/366/0/41/14 562/489/0/55/18 259/111/113/26/9 287/125/125/28/9 324/140/140/32/11 370/160/160/37/12 432/187/188/43/14 518/225/225/51/17 650/282/282/63/22 867/376/378/85/28 MPEG2 per service bit rate (kbps at TS level) 259/148/76/26/9 287/167/83/28/9 324/187/94/32/11 370/213/108/37/12 432/250/125/43/14 518/299/151/51/17 650/376/188/63/22 867/503/251/85/28 259/224//26/9 287/250//28/9 324/281//32/11 370/321//37/12 432/375//43/14 518/450//51/17 650/565//63/22 867/754//85/28 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 0.5 / 0.66 / 1 Bitrate/service (kbps at IP level) 107 / 140 / 214 115 / 156 / 238 132 / 173 / 263 148 / 197 / 304 173 / 238 / 353 214 / 288 / 427 271 / 362 / 542 362 / 476 / 723 Table A10.36 : T-16APSK short interleaver legacy ETSI 385 ETSI TS 102 584 V1.1.2 (2010-XY) A.10.4 TDM class 2 For uniform-long, three types of modulations are investigated: T-QPSK, 8PSK and 16APSK. For each modulation we give either {fixed code rate; variable duration} or {variable code rate; fixed duration} configurations. In class 2 configurations, OFDM modulation choice has a deep impact on the the interleaver duration because the SH frame duration is doubled. Therefore, although most cases are given with O-QPSK modulation, some specific cases are also given with 16QAM modulation. Terrestrial Modulation Code rate Duration Table modulation O-QPSK T-QPSK 1/3 Variable (0.5 to 10 Table seconds) A10.39 ½ Variable (0.5 to 10 Table seconds) A10.40 2/3 Variable (0.5 to 10 Table seconds) A10.41 Variable (1/5 to 2/3) 10 seconds Uniform late Table A10.42 8PSK 2/9 Variable (0.5 to 10 Table seconds) A10.43 1/3 Variable (0.5 to 10 Table seconds) A10.44 2/5 Variable (0.5 to 10 Table seconds) A10.45 Variable (1/5 to 2/3) 10 seconds Uniform late Table A10.46 16APSK Variable (1/5 to 2/3) 10 seconds Uniform late Table A10.47 16QAM 8PSK 1/3 Variable (0.5 to 10 Table seconds) A10.57 Table A10.37 : TDM class 2 uniform long physical layer configuration summary For uniform-late, three types of modulations are investigated: T-QPSK, 8PSK and 16APSK. For each modulation we give either {fixed code rate; variable duration} or {variable code rate; fixed duration} configurations. ETSI 386 ETSI TS 102 584 V1.1.2 (2010-XY) Modulation Code rate Duration Table T-QPSK 1/3 Variable (0.5 to 10 seconds) Table A10.48 Variable (5 to 10 seconds) Table A10.49 Optimized interleaver Variable (1/5 to 2/5) 10 seconds Uniform late Table A10.50 10 seconds Uniform late Table A10.51 Optimized interleave only upto 1/3 8PSK 2/9 Variable (0.5 to 10 seconds) Table A10.52 Variable (5 to 10 seconds) missing Optimized interleaver 1/3 Variable (0.5 to 10 seconds) Table A10.53 Variable (5 to 10 seconds) Missing Optimized interleaver 2/5 Variable (0.5 to 10 seconds) Table A10.54 Variable (5 to 10 seconds) Missing Optimized interleaver Variable (1/5 to 2/5) 10 seconds Uniform late Table A10.55 10 seconds Uniform late Missing Optimized interleave only upto 1/3 16APSK Variable (1/5 to 2/3) 10 seconds Uniform late Table A10.56 10 seconds Uniform late Missing Optimized interleave only upto 1/3 Table A10.38 : TDM class 2 uniform late physical layer configuration summary ETSI 387 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 - X X X X Parameter/Case name T_QPSK_1/3_U_11s T_QPSK_1/3_U_5.5s T_QPSK_1/3_U_2.7s T_QPSK_1/3_U_1.2s T_QPSK_1/3_U_0.5s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 1/3 1/3 1/3 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/  40/0/12/2/2 20/0/12/1/2 10/0/6/1/1 5/0/3/1/1 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10911 0 / 0 / 5549 0 / 2775 0 / 1273 0 / 546 MPEG TS total bit rate (Mbps) 2,567 2,567 2,567 2,567 2,567 TS LAYER CONFIGURATION Services 9 9 9 9 9 PHY / MAC synchronization Yes No No No No Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 TDM Symbols/ coded FEC 18144 18144 18144 18144 18144 total/data/ifec/safety/psi kbps@TP per time 184/160/0/18/6 184/160/0/18/6 184/160/0/18/6 184/160/0/18/6 184/160/0/18/6 slice MPEG2 per service bit rate (kbps at TS level) 284/247//28/9 284/247//28/9 284/247//28/9 284/247//28/9 284/247//28/9 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 230 230 230 230 230 Table A10.39 : TDM QPSK1/3 Uniform Long variable delay (O-QPSK) ETSI 388 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name T_QPSK_1/2_U_11s T_QPSK_1/2_U_5s T_QPSK_1/2_U_2.7s T_QPSK_1/2_U_1.2.s T_QPSK_1/2_U_0.6s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 1/2 1/2 1/2 ½ 1/2 Interleaver configuration (common_multiplier/nof_late_taps/  20/0/12/2/2 10/0/12/1/2 5/0/6/1/2 3/0/3/1/2 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10816 0 / 0 / 5454 0 / 0 / 2727 0 / 0 / 1273 0 / 0 / 557 MPEG TS total bit rate (Mbps) 3,900 3,900 3,900 3,900 3,900 TS LAYER CONFIGURATION Services 9 9 9 9 9 PHY / MAC synchronization Yes No No No No Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 TDM Symbols/ coded FEC 12096 12096 12096 12096 12096 total/data/ifec/safety/psi kbps@TP per time 280/243/0/28/9 280/243/0/28/9 280/243/0/28/9 280/243/0/28/9 280/243/0/28/9 slice MPEG2 per service bit rate (kbps at TS level) 432/375//43/14 432/375//43/14 432/375//43/14 432/375//43/14 432/375//43/14 LINK LAYER CONFIGURATION  1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 353 353 353 353 353 Table A10.40 : TDM QPSK1/2 Uniform Long variable delay (O-QPSK) ETSI 389 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name T_QPSK_2/3_U_9.9s T_QPSK_2/3_U_4s T_QPSK_2/3_U_2s T_QPSK_2/3_U_1s T_QPSK_2/3_U_0.4s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 2/3 2/3 2/3 2/3 2/3 Interleaver configuration (common_multiplier/nof_late_taps/  20/0/12/2/2 10/0/12/1/2 5/0/6/1/2 3/0/3/1/2 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10816 0 / 0 / 5455 0 / 0 / 2727 0 / 0 / 1273 0 / 0 / 557 MPEG TS total bit rate (Mbps) 5184 5184 5184 5184 5184 TS LAYER CONFIGURATION Services 9 9 9 9 9 PHY / MAC synchronization YES NO NO NO NO Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 TDM Symbols/ coded FEC 9072 9072 9072 9072 9072 total/data/ifec/safety/psi kbps@TP per time 373/324/0/37/12 373/324/0/37/12 373/324/0/37/12 373/324/0/37/12 373/324/0/37/12 slice MPEG2 per service bit rate (kbps at TS level) 575/500//57/19 575/500//57/19 575/500//57/19 575/500//57/19 575/500//57/19 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 476 476 476 476 476 Table A10.41 : TDM QPSK2/3 Uniform Long variable delay (O-QPSK) ETSI 390 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 - X X X X Parameter/Case name T_8PSK_2/9_U_11s T_8PSK_2/9_U_6s T_8PSK_2/9_U_2.7s T_8PSK_2/9_U_1.2s T_8PSK_2/9_U_0.5s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 2/9 2/9 2/9 2/9 2/9 Interleaver configuration (common_multiplier/nof_late_taps/ 60/0/12/4/2 60/0/12/2/2 30/0/12/1/1 15/0/6/1/1 5/0/3/1/1 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10911 0 / 0 / 5549 0 / 0 / 2728 0 / 0 / 1273 0 / 0 / 526 MPEG TS total bit rate (Mbps) 2,567 2,567 2,567 2,567 2,567 TS LAYER CONFIGURATION  9 9 9 9 9 Synchronization LINK / PHY Yes Yes Yes Yes Yes Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 TDM Symbols/ coded FEC 18144 18144 18144 18144 18144 total/data/ifec/safety/psi kbps@TP per time 184/160/0/18/6 184/160/0/18/6 184/160/0/18/6 184/160/0/18/6 184/160/0/18/6 slice MPEG2 per service bit rate (kbps at TS level) 284/247//28/9 284/247//28/9 284/247//28/9 284/247//28/9 284/247//28/9 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 230 230 230 230 230 Table A10.42 : TDM 8PSK2/9 Uniform Long variable delay (O-QPSK) ETSI 391 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X X X - X X X Parameter/Case name T_QPSK_1/5_U T_QPSK_2/9_U T_QPSK_1/4_U T_QPSK_2/7_U T_QPSK_1/3_U T_QPSK_2/5_U T_QPSK1/2_U T_QPSK2/3_U PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, QPSK, 5MHz, QPSK, 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) 0,15 0,15 PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 40/0/12/4/2 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10912 0 / 0 / 10909 0 / 0 / 10911 0 / 0 / 10909 0 / 0 / 10911 0 / 0 / 10909 0 / 0 / 10908 0 / 0 / 10909 MPEG TS total bit rate (Mbps) 1,530 1,728 1,925 2,222 2,567 3,110 3,900 5,184 TS LAYER CONFIGURATION Services 9 9 9 9 9 9 9 9 PHY / MAC synchronization Yes Yes Yes Yes Yes Yes Yes Yes Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 108 108 108 TDM Symbols/ coded FEC 30240 27216 24192 21168 18144 15120 12096 9072 total/data/ifec/safety/psi kbps@TP per time 110/95/0/11/4 124/108/0/12/4 138/119/0/14/5 160/139/0/16/5 184/160/0/18/6 224/195/0/22/7 280/243/0/28/9 373/324/0/37/12 slice MPEG2 per service bit rate (kbps at TS level) 170/147//17/6 191/167//19/6 213/184//22/8 247/214//25/8 284/247//28/9 346/301//34/11 432/375//43/14 575/500//57/19 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 140 156 173 206 230 288 353 476 Table A10.43 : TDM QPSK variable code rate Uniform Long 10 seconds (O-QPSK) ETSI 392 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name T_8PSK_1/3_U_9.9s T_8PSK_1/3_U_4s T_8PSK_1/3_U_2s T_8PSK_1/3_U_1s T_8PSK_1/3_U_0.5s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 1/3 1/3 1/3 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 60/0/12/4/2 30/0/12/2/2 15/0/12/1/2 13/0/6/1/1 5/0/3/1/1 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10908 0 / 0 / 5454 0 / 0 / 2727 0 / 0 / 1265 0 / 0 / 526 MPEG TS total bit rate (Mbps) 3,900 3,900 3,900 3,900 3,900 TS LAYER CONFIGURATION  9 9 9 9 9 Synchronization LINK / PHY Yes Yes Yes Yes Yes Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 TDM Symbols/ coded FEC 12096 12096 12096 12096 12096 total/data/ifec/safety/psi kbps@TP per time 280/243/0/28/9 280/243/0/28/9 280/243/0/28/9 280/243/0/28/9 280/243/0/28/9 slice MPEG2 per service bit rate (kbps at TS level) 432/375//43/14 432/375//43/14 432/375//43/14 432/375//43/14 432/375//43/14 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 353 353 353 353 353 Table A10.44 : TDM 8PSK1/3 Uniform Long variable delay (O-QPSK) ETSI 393 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name T_8PSK_2/5_U_20s T_8PSK_2/5_U_16s T_8PSK_2/5_U_10s T_8PSK_2/5_U_4s T_8PSK_2/5_U_2s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 2/5 2/5 2/5 2/5 2/5 Interleaver configuration (common_multiplier/nof_late_taps/ 60/0/12/8/2 60/0/12/6/2 60/0/12/4/2 30/0/12/2/2 15/0/12/1/2 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 21631 0 / 0 / 16270 0 / 0 / 10908 0 / 0 / 5454 0 / 0 / 2727 MPEG TS total bit rate (Mbps) 4,690 4,690 4,690 4,690 4,690 TS LAYER CONFIGURATION  9 9 9 9 9 Synchronization LINK / PHY Yes No Yes No No Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 TDM Symbols/ coded FEC 10080 10080 10080 10080 10080 total/data/ifec/safety/psi kbps@TP per time 337/293/0/33/11 337/293/0/33/11 337/293/0/33/11 337/293/0/33/11 337/293/0/33/11 slice MPEG2 per service bit rate (kbps at TS level) 520/452//51/17 520/452//51/17 520/452//51/17 520/452//51/17 520/452//51/17 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 435 435 435 435 435 Table A10.45 : TDM 8PSK2/5 Uniform Long variable delay (O-QPSK) ETSI 394 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X - X X X X X X Parameter/Case name 8PSK_1/5_U 8PSK_2/9_U 8PSK_1/4_U 8PSK_2/7_U 8PSK_1/3_U 8PSK_2/5_U 8PSK1/2_U 8PSK2/3_U PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, 8PSK, 5MHz, 8PSK, 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) 0,15 0,15 PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 60/0/12/4/2 60/0/12/4/2 60/0/12/4/2 60/0/12/4/2 60/0/12/4/2 60/0/12/4/2 60/0/12/4/2 60/0/12/4/2 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 10910 0 / 0 / 10910 0 / 0 / 10910 0 / 0 / 10910 0 / 0 / 10910 0 / 0 / 10910 0 / 0 / 10910 0 / 0 / 10910 MPEG TS total bit rate (Mbps) 2,320 2,567 2,913 3,308 3,900 4,690 5,826 7,800 TS LAYER CONFIGURATION Services 9 9 9 9 9 9 9 9 Synchronization LINK / PHY Yes Yes Yes Yes Yes Yes Yes Yes Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 108 108 108 TDM Symbols/ coded FEC 20160 18144 16128 14112 12096 10080 8064 6048 total/data/ifec/safety/psi kbps@TP per time 167/144/0/17/6 184/160/0/18/6 209/181/0/21/7 238/207/0/23/8 280/243/0/28/9 337/293/0/33/11 419/364/0/41/14 561/488/0/55/18 slice MPEG2 per service bit rate (kbps at TS level) 258/222//26/9 284/247//28/9 322/279//32/11 367/319//35/12 432/375//43/14 520/452//51/17 646/562//63/22 866/753//85/28 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 214 230 263 304 353 435 534 723 Table A10.46 : TDM 8PSK variable code rate Uniform Long 10 seconds (O-QPSK) ETSI 395 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 7 8 new to IG rel 1 X X X X X X X X Parameter/Case name 16APSK_1/5_U 16APSK_2/9_U 16APSK_1/4_U 16APSK_2/7_U 16APSK_1/3_U 16APSK_2/5_U 16APSK1/2_U 16APSK2/3_U PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, 16APSK, 5MHz, 16APSK, 5MHz, 16APSK, 5MHz, 16APSK, 5MHz, 16APSK, 5MHz, 16APSK, 5MHz, 16APSK, 5MHz, 16APSK, (TDM bandwidth, TDM_modulation; roll-off) 0,15 0,15 0,15 0,15 0,15 0,15 0,15 0,15 PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 1/2 2/3 Interleaver configuration (common_multiplier/nof_late_taps/ 63/0/12/4/2 63/0/12/4/2 63/0/12/4/2 63/0/12/4/2 63/0/12/4/2 63/0/12/4/2 63/0/12/4/2 63/0/12/4/2 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 0 / 10870 0 / 0 / 10870 0 / 0 / 10869 0 / 0 / 10870 0 / 0 / 10870 0 / 0 / 10870 0 / 0 / 10869 0 / 0 / 10870 MPEG TS total bit rate (Mbps) 3,110 3,456 3,900 4,443 5,184 6,221 7,800 10,417 TS LAYER CONFIGURATION Services 9 9 9 9 9 9 9 9 Synchronization LINK / PHY Yes Yes Yes Yes Yes Yes Yes Yes Repetition interval (ms) 975 975 975 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 108 108 108 TDM Symbols/ coded FEC 15120 13608 12096 10584 9072 7560 6048 4536 total/data/ifec/safety/psi kbps@TP per time 224/195/0/22/7 248/216/0/24/8 280/243/0/28/9 320/279/0/31/10 373/324/0/37/12 448/390/0/44/14 561/488/0/55/18 750/653/0/73/24 slice 346/301//34/11 383/333//37/12 432/375//43/14 494/430//48/15 575/500//57/19 691/602//68/22 866/753//85/28 1157/1007//113/ MPEG2 per service bit rate (kbps at TS level) 37 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at TS level) 288 321 353 411 476 575 723 969 Table A10.47 : TDM 16APSK variable code rate Uniform Long 10 seconds (O-QPSK) ETSI 396 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 - X X X X Parameter/Case name T_QPSK_1/3_UL_9.9s T_QPSK_1/3_UL_4s T_QPSK_1/3_UL_2s T_QPSK_1/3_UL_1s T_QPSK_1/3_UL_0.4s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 1/3 1/3 1/3 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 10/24/9/5/12 10/24/9/2/6 10/24/9/1/3 10/24/5/1/3 12/48/1/0/0 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 180 / 9936 0 / 3993 0 / 1997 0 / 1092 0 / 441 MPEG TS total bit rate (Mbps) 2,567 2,567 2,567 2,567 2,567 TS LAYER CONFIGURATION  9 9 9 9 9  Yes No No No No Repetition interval (ms) 1219 1219 1219 1219 1219 Burst duration (ms) 135 135 135 135 135 CW per SH frame 52 52 52 52 52 total/data/ifec/safety/psi TP per time slice 231/200/0/23/8 231/200/0/23/8 231/200/0/23/8 231/200/0/23/8 231/200/0/23/8 total/data/ifec/safety/psi kbps@TP per time slice 285/247//28/10 285/247//28/10 285/247//28/10 285/247//28/10 285/247//28/10 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 237 237 237 237 237 Table A10.48 : TDM QPSK1/3 Uniform Late variable delay (O-QPSK) ETSI 397 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 new to IG rel 1 - X Parameter/Case name T_QPSK_1/3_UL_9.9s T_QPSK_1/3_UL_4s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 5/18/11/4/9 5/18/6/4/4 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 66 / 9819 0 / 66 / 4952 MPEG TS total bit rate (Mbps) 2,567 2,567 TS LAYER CONFIGURATION  9 9  Yes No Repetition interval (ms) 1219 1219 Burst duration (ms) 135 135 CW per SH frame 52 52 total/data/ifec/safety/psi TP per time slice 231/200/0/23/8 231/200/0/23/8 total/data/ifec/safety/psi kbps@TP per time slice 285/247//28/10 285/247//28/10 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 237 237 Table A10.49: TDM QPSK1/3 optimized Uniform Late variable delay (O-QPSK) ETSI 398 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 new to IG rel 1 X X X X - X Parameter/Case name T_QPSK_1/5_UL T_QPSK_2/9_UL T_QPSK_1/4_UL T_QPSK_2/7_UL T_QPSK_1/3_UL T_QPSK_2/5_UL PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate  2/9 1/4 2/7 1/3 2/5 Interleaver configuration (common_multiplier/nof_late_taps/ 10/24/9/5/12 10/24/9/5/12 10/24/9/5/12 10/24/9/5/12 10/24/9/5/12 10/24/9/5/12 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 9937 0 / 9934 0 / 9936 0 / 9934 0 / 9936 0 / 9934 MPEG TS total bit rate (Mbps) 1,530 1,728 1,925 2,222 2,567 3,110 TS LAYER CONFIGURATION  9 9 9 9 9 9  Yes Yes Yes Yes Yes Yes Repetition interval (ms) 1219 1219 1219 1219 1219 1219 Burst duration (ms) 135 135 135 135 135 135 CW per SH frame 31 35 39 45 52 63 total/data/ifec/safety/psi TP per time slice 137/118/0/14/5 155/135/0/15/5 173/150/0/17/6 200/173/0/20/7 231/200/0/23/8 280 total/data/ifec/safety/psi kbps@TP per time 169/146//17/6 191/167//19/6 214/185//21/7 247/214//25/9 285/247//28/10 346 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 138 158 178 204 237 346 Table A10.50 : TDM QPSK variable code rate Uniform Late 10 seconds (O-QPSK) ETSI 399 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X - Parameter/Case name T_QPSK_1/5_UL T_QPSK_2/9_UL T_QPSK_1/4_UL T_QPSK_2/7_UL T_QPSK_1/3_UL PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 5MHz, QPSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 1/5 2/9 1/4 2/7 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 5/12/10/5/4 5/15/12/4/7 5/16/9/5/6 5/16/9/5/6 5/18/11/4/6 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 43 / 11014 0 / 55/ 10777 0 / 59 / 9819 0 / 59 / 9818 0 / 66 / 9795 MPEG TS total bit rate (Mbps) 1,530 1,728 1,925 2,222 2,567 TS LAYER CONFIGURATION Services 9 9 9 9 9 Synchronization LINK / PHY Yes Yes Yes Yes Yes Repetition interval (ms) 1219 1219 1219 1219 1219 Burst duration (ms) 135 135 135 135 135 CW per SH frame 31 35 39 45 52 total/data/ifec/safety/psi TP per time slice 137/118/0/14/5 155/135/0/15/5 173/150/0/17/6 200/173/0/20/7 231/200/0/23/8 total/data/ifec/safety/psi kbps@TP per time 169/146//17/6 191/167//19/6 214/185//21/7 247/214//25/9 285/247//28/10 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 138 158 178 204 237 Table A10.51 : TDM QPSK variable code rate optimized Uniform Late 10 seconds (O-QPSK) ETSI 400 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 - X X X X Parameter/Case name T_8PSK_2/9_UL_9.9s T_8PSK_2/9_UL_6s T_8PSK_2/9_UL_4s T_8PSK_2/9_UL_1s T_8PSK_2/9_UL_0.5s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 2/9 2/9 2/9 2/9 2/9 Interleaver configuration (common_multiplier/nof_late_taps/ 15/24/9/5/12 15/24/9/3/12 8/24/9/2/6 5/24/5/1/3 3/24/3/1/1 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 180 / 9936 0 / 180 / 6037 0 / 96 / 3949 0 / 60 / 1014 0 / 36 / 505 MPEG TS total bit rate (Mbps) 2,567 2,567 2,567 2,567 2,567 TS LAYER CONFIGURATION  9 9 9 9 9 Synchronization LINK / PHY Yes No No No No Repetition interval (ms) 1219 1219 1219 1219 1219 Burst duration (ms) 135 135 135 135 135 TDM Symbols/ coded FEC 18144 18144 18144 18144 18144 total/data/ifec/safety/psi TP per time slice 231/200/0/23/8 231/200/0/23/8 231/200/0/23/8 231/200/0/23/8 231/200/0/23/8 total/data/ifec/safety/psi kbps@TP per time 285/247//28/10 285/247//28/10 285/247//28/10 285/247//28/10 285/247//28/10 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 237 237 237 237 237 Table A10.52 : TDM 8PSK2/9 Uniform Late variable delay (O-QPSK) ETSI 401 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name T_8PSK_1/3_UL_9.9s T_8PSK_1/3_UL_4s T_8PSK_1/3_UL_2s T_8PSK_1/3_UL_1s T_8PSK_1/3_UL_0.5s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 1/3 1/3 1/3 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 15/24/9/5/12 12/24/9/2/6 12/24/9/1/3 5/24/5/1/1 3/24/3/1/1 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 177 / 9934 0 / 142 / 3973 0 / 142 /1987 0 / 59 / 988 0 / 35 / 504 MPEG TS total bit rate (Mbps) 3,900 3,900 3,900 3,900 3,900 TS LAYER CONFIGURATION  9 9 9 9 9 Synchronization LINK / PHY Yes No No No No Repetition interval (ms) 1219 1219 1219 1219 1219 Burst duration (ms) 135 135 135 135 135 TDM Symbols/ coded FEC 12096 12096 12096 12096 12096 total/data/ifec/safety/psi TP per time slice 351/306/0/34/11 351/306/0/34/11 351/306/0/34/11 351/306/0/34/11 351/306/0/34/11 total/data/ifec/safety/psi kbps@TP per time 433/378//42/14 433/378//42/14 433/378//42/14 433/378//42/14 433/378//42/14 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 362 362 362 362 362 Table A10.53 : TDM 8PSK1/3 Uniform Late variable delay (O-QPSK) ETSI 402 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name T_8PSK_2/5_UL_20s T_8PSK_2/5_UL_16s T_8PSK_2/5_UL_10s T_8PSK_2/5_UL_4s T_8PSK_2/5_UL_2s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 2/5 2/5 2/5 2/5 2/5 Interleaver configuration (common_multiplier/nof_late_taps/ 15/24/9/10/12 15/24/9/8/12 15/24/9/5/12 15/24/9/2/6 10/24/9/1/4 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 177 / 19682 0 / 177 / 15782 0 / 177 / 9933 0 / 177 / 3992 0 / 118 / 1991 MPEG TS total bit rate (Mbps) 4,690 4,690 4,690 4,690 4,690 TS LAYER CONFIGURATION Services 9 9 9 9 9 Synchronization LINK / PHY Yes No No No No Repetition interval (ms) 1219 1219 1219 1219 1219 Burst duration (ms) 135 135 135 135 135 TDM Symbols/ coded FEC 10080 10080 10080 10080 10080 total/data/ifec/safety/psi TP per time slice 422/367/0/41/14 422/367/0/41/14 422/367/0/41/14 422/367/0/41/14 422/367/0/41/14 total/data/ifec/safety/psi kbps@TP per time 521/453//51/17 521/453//51/17 521/453//51/17 521/453//51/17 521/453//51/17 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 434,00 434,00 434,00 434,00 434,00 Table A10.54 : TDM 8PSK2/5 Uniform Late variable delay (O-QPSK) ETSI 403 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 new to IG rel 1 X - X X X X Parameter/Case name 8PSK_1/5_UL 8PSK_2/9_UL 8PSK_1/4_UL 8PSK_2/7_UL 8PSK_1/3_UL 8PSK_2/5_UL PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 Interleaver configuration (common_multiplier/nof_late_taps/ 15/24/9/5/12 15/24/9/5/12 15/24/9/5/12 15/24/9/5/12 15/24/9/5/12 15/24/9/5/12 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 179 / 9935 0 / 179 / 9935 0 / 179 / 9935 0 / 179 / 9935 0 / 179 / 9935 0 / 179 / 9935 MPEG TS total bit rate (Mbps) 2,320 2,567 2,913 3,308 3,900 4,690 TS LAYER CONFIGURATION Services 9 9 9 9 9 9 Synchronization LINK / PHY Yes Yes Yes Yes Yes Yes Repetition interval (ms) 1219 1219 1219 1219 1219 1219 Burst duration (ms) 135 135 135 135 135 135 TDM Symbols/ coded FEC 20160 18144 16128 14112 12096 10080 total/data/ifec/safety/psi TP per time slice 208/180/0/21/7 231/200/0/23/8 262/227/0/26/9 297/258/0/29/10 351/306/0/34/11 422/367/0/41/14 total/data/ifec/safety/psi kbps@TP per time 257/222//26/9 285/247//28/10 323/280//32/11 367/318//36/12 433/378//42/14 521/453//51/17 slice LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 211 237 270 302 362 434 Table A10.55 : TDM 8PSK variable code rate Uniform Late 10 seconds (O-QPSK) ETSI 404 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 6 new to IG rel 1 X X X X X X Parameter/Case name 16APSK_1/5_UL 16APSK_2/9_UL 16APSK_1/4_UL 16APSK_2/7_UL 16APSK_1/3_UL 16APSK_2/5_UL PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, 5MHz, GI=1/4, (BW, GI, OFDM modulation order) QPSK QPSK QPSK QPSK QPSK QPSK TDM modulation parameters 5MHz, 16APSK, 5MHz, 16APSK, 5MHz, 16APSK, 5MHz, 16APSK, 5MHz, 16APSK, 5MHz, 16APSK, (TDM bandwidth, TDM_modulation; roll-off) 0,15 0,15 0,15 0,15 0,15 0,15 PHY FEC rate 1/5 2/9 1/4 2/7 1/3 2/5 Interleaver configuration (common_multiplier/nof_late_taps/ 20/24/9/5/12 20/24/9/5/12 20/24/9/5/12 20/24/9/5/12 20/24/9/5/12 20/24/9/5/12 nof_slices/slice_distance/non_late_increment) Min / late / Max interleaver duration (ms) 0 / 178 / 9934 0 / 178 / 9936 0 / 177 / 9934 0 / 178 / 9936 0 / 178 / 9934 0 / 178 / 9933 MPEG TS total bit rate (Mbps) 3,110 3,456 3,900 4,443 5,184 6,221 TS LAYER CONFIGURATION Services 9 9 9 9 9 9 Synchronization LINK / PHY Yes Yes Yes Yes Yes Yes Repetition interval (ms) 1219 1219 1219 1219 1219 1219 Burst duration (ms) 135 135 135 135 135 135 CW per SH frame 63 70 79 90 105 126 TDM Symbols/ coded FEC 15120 13608 12096 10584 9072 7560 total/data/ifec/safety/psi TP per time slice 280/243/0/28/9 311/270/0/31/10 351/306/0/34/11 400/348/0/39/13 466/405/0/46/15 560/487/0/55/18 total/data/ifec/safety/psi kbps@TP per time 346/300//35/11 384/333//38/12 433/378//42/14 494/430//48/16 575/500//57/19 691/601//68/22 slice LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 283 316 362 408 480 578 Table A10.56 : TDM 16APSK variable code rate Uniform Late 10 seconds (O-QPSK) ETSI 405 ETSI TS 102 584 V1.1.2 (2010-XY) ID 1 2 3 4 5 new to IG rel 1 X X X X X Parameter/Case name T_8PSK_1/3_U_9.9s T_8PSK_1/3_U_5s T_8PSK_1/3_U_2s T_8PSK_1/3_U_1s T_8PSK_1/3_U_0.5s PHYSICAL CONFIGURATION OFDM related parameters 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK 5MHz, GI=1/4, QPSK (BW, GI, OFDM modulation order) TDM modulation parameters 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 5MHz, 8PSK, 0,15 (TDM bandwidth, TDM_modulation; roll-off) PHY FEC rate 1/3 1/3 1/3 1/3 1/3 Interleaver configuration (common_multiplier/nof_late_taps/ 40/0/16/5/2 60/0/12/4/2 20/0/18/1/2 13/0/9/1/1 5/0/5/1/1 nof_slices/slice_distance/non_late_increment) Min / Late / Max interleaver duration (ms) 0 / 0 / 9223 0 / 0 / 5549 0 / 0 / 2342 0 / 0 / 1022 0 / 0 / 516 MPEG TS total bit rate (Mbps) 3,851 3,851 3,851 3,851 3,851 TS LAYER CONFIGURATION  9 9 9 9 9 Synchronization LINK / PHY No No No No No Repetition interval (ms) 975 975 975 975 975 Burst duration (ms) 108 108 108 108 108 TDM Symbols/ coded FEC 12096 12096 12096 12096 12096 total/data/ifec/safety/psi kbps@TP per time 277/276/0/0/1 277/276/0/0/1 277/276/0/0/1 277/276/0/0/1 277/276/0/0/1 slice MPEG2 per service bit rate (kbps at TS level) 427/426///2 427/426///2 427/426///2 427/426///2 427/426///2 LINK LAYER CONFIGURATION LL-FEC rate min / recommended / max 1/1/1 1/1/1 1/1/1 1/1/1 1/1/1 Bitrate/service (kbps at IP level) 403 403 403 403 403 Table A10.57 : TDM 8PSK1/3 Uniform Long variable delay (O-16QAM) ETSI 406 ETSI TS 102 584 V1.1.2 (2010-XY) A.11 Reference Demodulator A further refinement has been performed by defining reference demodulator algorithms for investigating the impact of imperfect channel estimation. The demodulator algorithms that are described below are based on the data-aided approach and are considered representative of practical implementations (enhancements are possible). Carrier frequency and symbol clock extraction have not been included in the model as their impact are considered negligible. The models have been implemented in floating point. A.11.1 TDM Case A.11.1.1 TDM Reference Demodulator The proposed reference demodulator heavily exploits the TDM pilot symbols. The DVB-SH TDM slot structure is depicted in figure A11.5. For easing the acquisition and tracking a regular pilot spacing within each PL slot has been included in the DVB-SH standard. Each TDM slot of length LTOT is subdivided in two data sub-slots and pilot fields. Figure A11.5: Slot pilot insertion in DVB-SH TDM frame Carrier Phase Estimation: The DVB-S2 carrier phase estimator described in clause 3.6 of [i.16] has been adapted to the DVB-SH case. Following [i.16], the phase estimator consists of a ML feed-forward phase estimator operating on each pilot block belonging to the slot. The ML phase estimator algorithm adapted from [i.16] for the pilot group #n is given by: L * !  ! zn =  C [ k + (n − 1)( LSS + LPF ) ] r [ k + (n − 1)( LSS + LPF ) ], n = 1,.. N p PF k =1   θˆ ( n ) = arg { z } , n = 1,.. N PF n p where r (k) represents the k-th on-time baseband demodulator square-root raised-cosine chip matched filter output complex samples after frequency offset removal. The phase unwrapping allows to implement a simple linear phase interpolator between consecutive pilot estimations. ˆ ˆ The final unwrapped pilot estimates θ PFU ( n ) are computed from θ PF ( n ) as: θˆPFU ( n ) = θˆPFU ( n − 1) + SAW θˆPF ( n ) − θˆPFU ( n − 1)    where SAW [Φ ] ≡ [Φ ] +π −π is a saw tooth nonlinearity that reduces Φ to the interval (-, ). Linear interpolation between consecutive pilot blocks n and n+1 for sub-slice carrier phase removal then follows trough the following equation: l ˆ θˆPFUI ( l , n ) = θˆPFU ( n ) + θ PFU ( n + 1) − θˆPFU ( n )  LSS   ETSI 407 ETSI TS 102 584 V1.1.2 (2010-XY) where l = 1, 2, …, LSS, is the symbol index within the sub-slice n. Carrier amplitude estimation: By reusing the pilot-aided estimated signal phasor zn one can compute the signal amplitude simply as: ! ( n ) = z , n = 1,.. N A n p Linear amplitude interpolation between consecutive pilot blocks n and n+1 for sub-slice then follows trough the following equation: ! ( l , n ) = ! ( n ) + l  ! ( n + 1) − ! ( n )  A A A A LSS   where l = 1, 2, …, LSS, is the symbol index within the sub-slice n. Signal-to-Noise-Ratio Estimation: The signal to noise ratio estimation is required for the correct turbo decoder operation. For the SNR estimator the data-aided version of the SNORE [SNORE] algorithm based on pilot symbols have been adopted. The algorithmic details can be found in [i.16]. A.11.1.2 TDM Reference Demodulator Simulation Results To assess the performance of the DVB-SH TDM option a physical layer simulator inclusive of the above described channel estimation algorithms has been implemented. First a Ricean fading channel with K = 5 dB, mobile speed of v = 50 kmph and worst-case for single antenna reception SNR = -3,5 dB has been considered. Channel estimation results are plotted in figures A11.6 to A11.8. Some high error peaks are present corresponding to deep fading conditions. In these conditions, the channel estimation has no impact on BER as the Turbo decoder will not be able to decode even with perfect channel estimation. The BER curve confirms this analysis. Another set of channel estimation results have been obtained for the LMS intermediate tree shadowed channel with mobile speed of v = 70 kmph and a line-of-sight SNR = 10 dB (considered to be a representative demodulator operating point in this channel). The pilot-aided phase estimation show to perform well also in this difficult type of channel. Again large phase error peaks correspond to deep channel fading conditions. ETSI 408 ETSI TS 102 584 V1.1.2 (2010-XY) 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0 2 4 6 8 10 12 14 16 4 x 10 Figure A11.6: Simulated TDM reference demodulator channel estimation amplitude for Rice K = 5 dB, mobile v = 50 kmph channel, SNR =-3,5 dB. 2 1.5 1 0.5 0 -0.5 -1 -1.5 -2 -2.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 6 x 10 Figure A11.7: : Simulated TDM reference demodulator channel estimation phase error in radians for Rice K = 5 dB, mobile v = 50 kmph channel, SNR =-3,5 dB. ETSI 409 ETSI TS 102 584 V1.1.2 (2010-XY) 2 1.5 1 0.5 0 3750 3755 3760 3765 3770 3775 3780 3785 3790 3795 3800 0 -1 -2 -3 3750 3755 3760 3765 3770 3775 3780 3785 3790 3795 3800 Figure A11.8: : Simulated TDM reference demodulator channel estimation for Rice K = 5 dB, mobile v = 50 kmph channel, SNR =-3,5 dB: zoom of fading channel amplitude and estimated carrier phase error relation. 200 100 0 Phase [deg] / Power [dB] -100 -200 -300 -400 Faded signal phase Faded signal power x 10 [dB] Pilot-aided phase estimation -500 0 1 2 3 4 5 6 Symbol samples 5 x 10 Figure A11.9: Real and simulated TDM reference demodulator channel carrier phase estimation for v = 70 kmph, SNR = 10 dB and a LMS intermediate state of the intermediate tree shadowing environment. ETSI 410 ETSI TS 102 584 V1.1.2 (2010-XY) 60 40 20 Phase [deg] 0 -20 -40 -60 0 1 2 3 4 5 6 Symbol samples 5 x 10 Figure A11.10: TDM reference demodulator channel carrier phase estimation error for v = 70 kmph, SNR = 10 dB and a LMS intermediate state of the intermediate tree shadowing environment. A more comprehensive analysis of the channel estimation impact has been performed looking at the BER performance versus Es / N 0 in Ricean K=5 dB channels. Figure A11.11 shows the corresponding results for a short PL interleaving configuration. AWGN and Rice reference performance with ideal channel estimation (for short 200 ms interleaver and infinite length interleaver (uncorrelated fading)) are also reported for convenience. It can be concluded that the proposed reference demodulator channel estimation has negligible impact with Ricean channel K=5 dB. ETSI 411 ETSI TS 102 584 V1.1.2 (2010-XY) QPSK 1/3 AWGN 1,E+00 1,E-01 Rice K=5dB Uncorrelated Perfect Estimation 1,E-02 Rice K=5dB Correlated Short interleaver 1,E-03 Perfect Estimation BER 1,E-04 Rice K=5dB Correlated Short interleaver PHASE estimation 1,E-05 Rice K=5dB Correlated Short interleaver 1,E-06 PHASE estimation AMPLITUDE estimation 1,E-07 Rice K=5dB Correlated Short interleaver -2,00 -1,75 -1,50 -1,25 -1,00 -0,75 -0,50 -0,25 0,00 PHASE estimation AMPLITUDE Es/N0 (dB) estimation Figure A11.11: Simulated BER TDM reference demodulator BER performance for Rice K=5 dB, mobile v= 50 kmph channel, as a function of Es / N 0 Reference demodulator performance have also been assessed for the LMS 3-state channel case with the demodulator state machine on. The results at C/N = 10 dB are summarized in the following Table A11.15. It is apparent that the TDM reference demodulator has no practical impact on the ESR5 performance. The final verification performed on the TDM reference demodulator is related to the carrier phase noise impact. For this a synthetic phase noise nuisance following the mask of clause has been included in the simulation chain. We remark that the low frequency component of the DVB-SSP phase noise is less than the DVB-S2 one (about 30 dB at 100 Hz) but higher-frequency component (above 1E5 Hz) is larger. The mild slope of the DVB-SSP phase noise PSD makes the higher frequency component contribution dominating. This makes the phase noise time series much faster than the fading/shadowing ones. The simulated BER performance results are reported in figure A11.12. It can be concluded that for Rice channel with K = 5 dB and mobile speed of 50 kmph the carrier phase noise impact for QPSK r = 1/3 is less than 0,05 dB at BER = 1E-5. Table A11.15: Summary of ESR5 results at C/N = 10 dB with TDM state machine on and ideal or reference demodulator channel estimation Reference case FEC/Interleaver ESR5 satisfaction % ESR5 satisfaction % ideal channel estimation reference demodulator channel estimation SubUrban (SU) Short PL interleaver 45,6 48,9 V = 3 kmph Non uniform long PL interleaver 88,5 87,6 Uniform long PL interleaver 89,3 88,5 Short PL interleaver plus upper 87,6 Not available layer FEC Intermediate Tree Short PL interleaver 0,0 0,0 Shadowing (ITS) Non uniform long PL interleaver 25,2 25,6 V = 50 kmph Uniform long PL interleaver 80,4 78,2 Short PL interleaver plus upper 19,5 Not available layer FEC ETSI 412 ETSI TS 102 584 V1.1.2 (2010-XY) Table A11.16: S-band DVB-SH phase noise mask (Satellite contribution only) 10 Hz 100 Hz 1 000 Hz 10 kHz 100 kHz 1 MHz 10 MHz Phase noise -29 -59 -69 -74 -83 -95 -101 (dBc/Hz) AWGN QPSK 1/3 1,E+00 Rice K=5dB Uncorrelated 1,E-01 Perfect Estimation 1,E-02 Rice K=5dB Correlated Short interleaver Perfect Estimation 1,E-03 BER Rice K=5dB Correlated 1,E-04 Short interleaver PHASE NOISE 1,E-05 Rice K=5dB Correlated Short interleaver PHASE estimation AMPLITUDE estimation 1,E-06 SNR estmation Rice K=5dB Correlated 1,E-07 Short interleaver PHASE NOISE -2,00 -1,75 -1,50 -1,25 -1,00 -0,75 -0,50 -0,25 0,00 PHASE estimation AMPLITUDE estimation Es/N0 (dB) SNR estmation Figure A11.12: Simulated BER TDM reference demodulator BER performance for Rice K = 5 dB, mobile v = 50 kmph channel, as a function of Es / N 0 with and without phase noise. A.11.2 OFDM Case A.11.2.1 OFDM Reference Demodulator (see note) Channel transfer estimation is performed through adaptive Least Mean Square (LMS) interpolation both on time and frequency domain thanks to the knowledge of the channel at pilot positions. Firstly time-domain adaptive interpolation is performed to calculate the channel transfer function at scattered pilot subcarriers. Based on these estimates, frequency-domain adaptive interpolation is then performed at the non-pilot subcarriers. The following scheme is based on the assumption of a DVB-SH compliant OFDM transmission. At the pilot symbols location, the transfer function of the channel is calculated as: Y ( n, k ) H ( n, k ) = (A.1) X ( n, k ) where Y (n, k ) represents the received complex symbol at for the k-th subcarrier of the n-th OFDM symbol and X (n, k ) represents the transmitted pilot known at the demodulator corresponding to the same position in time and frequency. Time domain interpolation o) Filter coefficients update ETSI 413 ETSI TS 102 584 V1.1.2 (2010-XY) The interpolator coefficient update exploits the information carried by the continual pilot subcarriers, where the channel transfer function is known at each OFDM symbol. The result from (Eq.A.1) is then compared to the estimate of the channel in the same position (n,k) and performed through the interpolator filter. The filter coefficients are updated thanks to the estimation error, calculated as the distance between the estimated channel and the calculated one. The time-domain interpolator filter is designed to be 4(M1+M2)-OFDM symbol long, where 4M1 and 4M2 represent the number of exploited OFDM symbols before and after the instant to be estimated. The procedure is divided into three separate steps. Step 1: Channel transfer function estimation The channel estimation is performed filtering the pilots of the tone under analysis, using the set of coefficients calculated thanks to the previous (n-1)-th OFDM symbol. The estimation is repeated three times employing three different set of coefficients and pilot symbols belonging to continual subcarrier Ci (a graphical interpretation can be find in figure A11.13): M2 ! H L ( n, Ci ) =  m = − M1 +1 WL ( n − 1, m ) H ( n − L + 4m, Ci ) (A.2) where WL ( n − 1, m) represents the m-th coefficient of the L-th filter (L=1,2,3), updated during the previous (n-1)-th OFDM symbol.                   Figure A11.13: Graphical representation of the time domain channel estimation algorithm on continual pilot Ci Each filter performs the channel estimation for the same cell of the n-th OFDM symbol. Only one pilot every four is exploited and each set consists of a different pilot group. This redundancy is justified by the regular distribution of the scattered-pilot cells (see figure A11.14). The column Ci represents the (time) continual pilots while the column Sj represents the (time) scattered pilots. ETSI 414 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A11.14: Graphical representation of the DVB-SH OFDM pilot distribution in time (vertical axis) and frequency (horizontal axis) In general, the parameters M1 and M2 can be independently defined for each filter. Furthermore, when M2 is set to zero the filter is behaving as a predictor instead as an interpolator. This procedure is applied to each continual pilot tone (Ci) and repeated for each received OFDM symbol (n). Step 2: Estimation error calculation The estimation error for the filter updating is then derived from the estimates performed in Step1 (Eq.2) and the knowledge of the channel thanks to the pilot symbol transmission, according to the following equations for L=1,2,3: ! EL ( n, Ci ) = H ( n, Ci ) − H L ( n, Ci ) (A.3) The error calculation is performed on each continual pilot tone (Ci) and repeated for each received OFDM symbol (n). Step 3: Update of the time domain interpolation filter coefficients The interpolator filter coefficients are then updated using the errors calculated in Step2. The following summation are over all continual pilot tones (1..P) and repeated for every m ranging in [-M1+1:M2] for L=1,2,3: P WL ( n, m ) = WL ( n − 1, m ) + µ  EL ( n, Ci ) H ( n − L + 4m, Ci ) * (A.4) i =1 where µ represents the time interpolator filter LMS adaptation step. p) Scattered pilot interpolation The filter coefficients calculated exploiting the subcarriers containing continual pilot symbols are then employed to estimate the channel in non-pilot symbols position for the subcarriers containing scattered pilot symbols. The regular distribution of the scattered pilots ensures pilots to be transmitted every 4 OFDM symbols (then the factor 4 in Eq.A.2). The distance between a non-pilot carrier-symbol and the previously transmitted pilot is not higher then 3 OFDM symbols. According to which OFDM symbol is transmitting the pilot (n-1, n-2, n-3), the proper filter coefficient set W1, W2 or W3 must be used for the interpolation. More specifically the interpolated channel estimate for symbol n-L belonging to the scattered pilot subcarrier Si is computed as: M2 ! H ( n, Si ) =  m = − M1 +1 WL ( n, m ) H ( n − L + 4m, Si ) (A.5) ETSI 415 ETSI TS 102 584 V1.1.2 (2010-XY) ! The channel estimation H ( n, Si ) is based on the knowledge of the channel at H ( n − L + 4m, Si ) with L=1, 2, 3 where scattered pilot symbols are transmitted. This clarifies the necessity of three distinct sets of filter coefficients. This procedure is applied to each scattered-pilot subcarrier (Si) and repeated for each received OFDM symbol (n). Frequency domain interpolation The interpolation in the frequency domain is then performed to derive the channel estimation for the OFDM subcarriers not containing any pilot symbol. The channel transfer function estimation is based on the knowledge of the channel at the continual and scattered subcarriers derived during the previous time-domain interpolation phase. These stones are highlighted in figure A11.14. It clearly appears that two subcarriers are not carrying at all pilot symbols. The frequency-domain interpolator filter is designed to be 3(J1+J2) subcarrier long, where 3*J1 and 3*J2 represent the number of exploited OFDM subcarriers before and after the subcarrier to be estimated. q) Channel interpolation for pilot-less subcarriers Following an approach similar to the one applied for the time-domain interpolation, two distinct interpolation filters VI with I=1, 2 are defined in the frequency-domain: J2 ! H ( n, k ) =  m = − J1 +1 VI ( n − 1, j ) H ( n, k − I + 3 j ), for k = 3K + I (A.6) where K is an integer and VI (n-1,j) represents the j -th coefficient of the L-th filter (L=1,2), updated during the previous (n-1)-th OFDM symbol. Each filter performs the channel estimation for the k-th subcarrier. The indices relation k = 3K + I ensures that only ! the not-continual and not-scattered subcarriers are considered. The channel estimation H ( n, k ) is based on the knowledge of the channel cells H ( n, k − I + 3 j ) I=1, 2, calculated during the time-domain interpolation. r) Filter Coefficients Update Step 1: Estimation error calculation In the absence of pilot symbols, the error signals for the coefficient update are calculated based on a decision-direct approach. After the computation of the channel estimate H (n, k ) , the estimated transmitted symbol ! ( n, k ) can be X derived as: ! ( n, k ) = H  Y ( n, k )    X D   (A.7)  H ( n, k )    being H D { ⋅ } the function representing the complex symbol hard decision. For QSPK modulation the function 1 HD { ⋅ } is simply given by H D { zn } = {sign [ Re( zn )] + j sign [ Im( zn )]} . 2 Assuming the symbol hard decision reliable, it can be used as a pilot symbol so that the channel transfer function can be re-calculated according to: Y ( n, k ) H ( n, k ) = (A.8) ! ( n, k ) X The Estimation Error is defined as: ! D ( n, k ) = H ( n, k ) − H ( n, k ) (A.9) ETSI 416 ETSI TS 102 584 V1.1.2 (2010-XY) Step 2: Update of the Frequency Domain Interpolation Filter Coefficients The interpolator filter coefficients are then updated using the errors calculated in Step 1. The following summations are over all possible tones (k) to be updated (not-scattered and not-continual pilot subcarriers) i.e. for k = 3K+I, I=1, 2: VI ( n, j ) = VI ( n − 1, j ) + γ  D* ( n, k ) H ( n, k − L + 3 j ) (A.10) where γ represents the frequency interpolator filter LMS adaptation step. NOTE: The algorithms described here are inferred from the descriptions given in US patent US 2006/0269016 A1, Nov. 2006. A.11.2.2 OFDM Reference Demodulator Simulation Results A.11.2.2.1 Terrestrial Channel First the OFDM demodulator has been tested over the TU6 terrestrial reference channel, as described in TS 145.005 [29]. A fading Doppler spread range spanning from fd =50 Hz up to fd = 400 Hz (roughly equivalent to a mobile speed from v = 25 kmph up to v = 200 kmph at S-band) and an SNR=-2,0 dB have been used. Demodulator channel estimation simulation results are plotted in figures A11.15 and A11.16 where the OFDM subcarrier estimates are serialized in time (the frequency domain is read first). Figure A11.15 represents the signal amplitude estimation (orange curve) versus the real one (black curve) as a function of the received symbols. Figure A11.16 shows the correlation between large carrier phase estimation errors and deep channel fading conditions. These large phase errors have no practical impact on the end results. Turbo decoder will be anyway in error due to the very low SNR. Figures A11.18 to A11.20 confirm this analysis. Simulated OFDM reference demodulator channel estimation for TU6 with fading Doppler spread fd=50Hz, SNR=-2.0 dB. 2 2 1.51.5 Amplitude Amplitude 1 1 0.5 0.5 0 0 1.535 1.54 1.545 1.55 1.555 1.56 1.565 1.57 1.575 1.58 1.585 1.535 1.54 1.545 1.55 1.555 1.56 1.565 1.57 1.575 1.58 1.5855 x 10 1 OFDM symbol OF x 105 symbols Figure A11.15: Channel amplitude (black real, estimate orange) ETSI 417 ETSI TS 102 584 V1.1.2 (2010-XY) 2 (Estimated) Phase 0 -2 6.2 6.25 6.3 6.35 6.4 6.45 6.5 6.55 6.6 6.65 6.7 (perfect Estimation) 4 3 x 10 Amplitude 2 1 0 6.2 6.25 6.3 6.35 6.4 6.45 6.5 6.55 6.6 6.65 6.7 1 OFDM symbol symbols Figure A11.16: Fading channel amplitude (black) and estimated carrier phase error (orange) Another way to analyse the adaptive channel estimation algorithm performance is to look at its capability to estimate the fading bandwidth (Doppler spread). Figure A11.17 compares the simulated TU6 fading process bandwidth (blue curve) with the reference demodulator fading bandwidth estimate (red curve). It is remarked the good fading bandwidth estimation performed by the selected LMS adaptive algorithm. Example of filter adaptation at a continual-pilot-tone position. Power Spectral Density Estimate via Welch 10 Power/frequency (dB/rad/sample) 0 -10 -20 -30 -40 -50 0 0.05 0.1 0.15 0.2 0.25 0.3 Normalized Freque ncy (×π rad/sample) Figure A11.17: Channel response of LMS-filter (red plot) to a 50 Hz Doppler TU6 channel (blue plot) A more comprehensive evaluation of the OFDM reference demodulator channel estimation impact over terrestrial channels has been performed simulating the BER performance versus Es / N 0 in the TU6 channel. Figures A11.18 to A11.20 and figures A11.21 to A11.23 show the corresponding results for a short (±200 ms) PL interleaving configuration at different Doppler spread values. AWGN reference performances with ideal channel estimation are also reported for convenience. Figure A11.18 (A11.21) and figure A11.19 (A11.22) show results with perfect and real channel estimation respectively. Figure A11.20 (A11.23) show the dependency of demodulator losses to the fading Doppler frequency when the target BER is set to 10-5. The losses of the an ideal demodulator w.r.t. AWGN reference performance (blue curve) are decreasing with increasing Doppler frequency as the PL interleaver is increasingly able to ETSI 418 ETSI TS 102 584 V1.1.2 (2010-XY) decorrelate the fading. Instead losses are increasing with the Doppler spread in case of real channel estimation as the channel estimation errors are growing with the user (and channel) speed (dynamic). Finally the black curve in figure A11.16 shows the implementation losses of the reference demodulator compared to ideal channel estimation. In the QPSK-1/3 case, the channel estimation losses are below 1 dB for fd =50 Hz and goes up to about 4 dB for fd = 300 Hz. In the QPSK-1/2 case, the losses even lower at low Doppler frequency, below 0,5 dB for fd = 50 Hz, but remarkably higher, about 15 dB, for fd = 300. These results are considered representative of a typical OFDM demodulator. 1.E+00 AWGN - OFDM 1.E-01 TU6 - OFDM - Doppler = 50Hz 1.E-02 1.E-03 TU6 - OFDM - Doppler = 100Hz BER 1.E-04 TU6 - OFDM - Doppler = 200Hz 1.E-05 TU6 - OFDM - Doppler = 300Hz 1.E-06 TU6 - OFDM - Doppler = 400Hz 1.E-07 -2.00 -1.00 0.00 1.00 2.00 3.00 Es/N0 (dB) Figure A11.18: Simulated BER OFDM reference demodulator as a function of Es/N0, for QPSK-1/3 in TU6 channel with several Doppler frequencies: perfect channel estimation. 1.E+00 AWGN - OFDM 1.E-01 AWGN - OFDM - Channel Recovery 1.E-02 TU6 - OFDM - Doppler = 50Hz - Channel Recovery 1.E-03 BER TU6 - OFDM - Doppler = 100Hz - Channel Recovery 1.E-04 TU6 - OFDM - Doppler = 200Hz - Channel Recovery 1.E-05 TU6 - OFDM - Doppler = 300Hz - Channel Recovery 1.E-06 TU6 - OFDM - Doppler = 400Hz - Channel Recovery 1.E-07 -2.00 0.00 2.00 4.00 6.00 8.00 Es/N0 (dB) Figure A11.19: Simulated BER OFDM reference demodulator as a function of Es/N0, for QPSK-1/3 in TU6 channel with several Doppler frequencies: real channel estimation. ETSI 419 ETSI TS 102 584 V1.1.2 (2010-XY) 8.00 7.00 6.00 5.00 Perf ect Channel Estimation dB 4.00 Real Channel Estimation Implementation Loss 3.00 2.00 1.00 0.00 0 100 200 300 400 Doppler (Hz) Figure A11.20: Losses w.r.t. AWGN reference (perfect and real channel estimation) and Implementation losses of a real w.r.t. perfect channel estimation (at BER = 10-5). 1 .E +0 0 A WGN - OFDM 1.E-0 1 TU 6 - OFDM - D op pl er = 50 Hz 1.E-0 2 1.E-0 3 TU 6 - OFDM - D op pl er = 10 0H z BER 1.E-0 4 TU 6 - OFDM - D op pl er = 20 0H z 1.E-0 5 TU 6 - OFDM - D op pl er = 25 0H z 1.E-0 6 TU 6 - OFDM - D op pl er = 30 0H z 1.E-0 7 0 .0 0 1 .00 2.00 3.0 0 4 .00 5.00 6.0 0 7 .0 0 Es/N0 (dB) Figure A11.21: Simulated BER OFDM reference demodulator as a function of Es/N0, for QPSK rate 1/2 in TU6 channel with several Doppler frequencies: perfect channel estimation. ETSI 420 ETSI TS 102 584 V1.1.2 (2010-XY) 1 .E +0 0 AWGN - OFDM 1.E-0 1 TU6 - OFDM - Doppler = 50Hz 1.E-0 2 - Channel Recovery 1.E-0 3 TU6 - OFDM - Doppler = BER 100Hz - Channel Recovery 1.E-0 4 TU6 - OFDM - Doppler = 200Hz - Channel Recovery 1.E-0 5 TU6 - OFDM - Doppler = 250Hz - Channel Recovery 1.E-0 6 TU6 - OFDM - Doppler = 300Hz - Channel Recovery 1.E-0 7 -2 .00 3.0 0 8 .00 1 3.0 0 18 .00 Es/N0 (dB) Figure A11.22: Simulated BER OFDM reference demodulator as a function of Es/N0, for QPSK rate 1/2 in TU6 channel with several Doppler frequencies: real channel estimation. 20.00 18.00 16.00 14.00 12.00 Perfect Channel Estimation dB 10.00 Real Channel Estimation Implementation Loss 8.00 6.00 4.00 2.00 0.00 0 100 200 300 400 Doppler (Hz) Figure A11.23: Losses w.r.t. AWGN reference (Perfect and Real channel estimation) and Implementation losses of a real w.r.t. perfect channel estimation (at BER = 10-5). A.11.2.2.2 Satellite Channel Another set of channel estimation results have been obtained for a satellite flat Ricean fading channel with Rice factor K = 5 dB, mobile speed of v = 50 kmph and SNR = -2,0 dB (the same algorithm with the same parameters as above is used here). As described in clause A.10, the reference demodulator algorithm are the same for the terrestrial and satellite case. In case of satellite only operations, being the fading channel typically flat, there is no need for frequency interpolation of the channel estimates. No attempt has been done to optimize the channel estimation for the satellite-only type of operations. ETSI 421 ETSI TS 102 584 V1.1.2 (2010-XY) The reference demodulator channel estimation results are plotted in figures A11.24 to A11.25: BER performances versus Es/N0 are plotted in figure A11.26 for both perfect and real channel estimation. The impact of the proposed reference demodulator amount to 0,7 dB with Ricean channel and a mobile speed of 100 kmph compared to ideal channel estimation. 33 2.5 2.5 Amplitude 2 Amplitude 2 1.5 1.5 11 0.5 0.5 0 0 2.5 2.6 2.7 2.8 2.9 3 3.1 2.5 2.6 2.7 2.8 2.9 3 3.1 5 x 10 x 10 5 symbols 1 OFDM symbol Figure A11.24: Simulated OFDM reference demodulator channel amplitude estimation (orange line) versus real amplitude (black line) for Ricean, mobile speed 50 kmph, SNR=-2,0 dB. 4 2 Phase (rad) (Estimated) 0 -2 -4 0 100 200 300 400 500 600 2 (perfect Estimation) 1.5 Amplitude 1 0.5 0 0 100 200 300 400 500 600 Figure A11.25: Simulated OFDM reference demodulator channel estimation for Ricean, mobile speed 50 kmph, SNR=-2,0 dB: zoom of amplitude, zoom of fading channel amplitude and estimated carrier phase error relation and zoom of fading channel amplitude and estimated carrier phase error relation for a single FFT-tone. ETSI 422 ETSI TS 102 584 V1.1.2 (2010-XY) 1.E+00 1.E-01 AWGN - OFDM 1.E-02 RICE k = 5dB - 1.E-03 50Kmph - OFDM BER RICE k = 5dB - 50Kmph -OFDM - 1.E-04 Channel Recovery RICE k = 5dB - 100Kmph - OFDM 1.E-05 RICE k = 5dB - 100Kmph -OFDM - Channel Recovery 1.E-06 1.E-07 -2.00 -1.50 -1.00 -0.50 0.00 0.50 Es/N0 (dB) Figure A11.26: Simulated BER OFDM reference demodulator as a function of Es/N0, for Ricean channel with mobile speed 50 kmph and 100 kmph: perfect channel estimation versus real channel estimation. A.11.3 TDM Signalling Channel Performance Results In this clause we report some results obtained by simulation on the DVB-SH TDM signalling channel performance. At the terminal switch on it is important to receive the SH network key PL configuration parameters through correct signalling channel detection. The performance have been derived under the following conditions: • dynamic 3-state LMS-ITS, LMS-SU channel; • 400 000 signalling field simulated; • distance between two consecutive signalling fields (SH frame payload) is taken into account; • accumulation based on: - equal gain combining; - maximal ratio combining. The computed performances are: • signalling Block Word Error Rate; • probability of successful detection versus number of received signalling fields. ETSI 423 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A11.27 shows the simulated signalling field Word Error Rate as a function of the soft accumulated signalling fields for ITS LMS channel with a mobile speed of 50 kmph and Es / N 0 =-3,5 dB. Two way of performing soft combining have been considered the maximal ratio (MRC) and the equal gain (EGC) ones. As expected, the MRC provides superior performance compared to the ECG. Figure A11.28 shows the probability of correct TDM signalling field detection as a function of the number of received signalling fields for the 3-state LMS-ITS mobile channel v = 50 kmph, Es / N 0 =-3,5 dB. The results is of course dependent on how many signalling fields are softly combined, but in the worst case of no field combining after 30 signalling fields the signalling information is successfully detected with 99 % probability. With 5 fields soft combined 15 fields are sufficient to have the same detection probability. Figure Erreur ! Source du renvoi introuvable. reports the probability of correct TDM signalling field detection as a function of the number of received signalling fields for the 3-state LMS-SU mobile channel v = 3 kmph, Es / N 0 =-3,5 dB. The results is of course dependent on how many signalling fields are softly combined, but in the worst case of no field combining after 30 signalling fields the signalling information is successfully detected with 99 % probability. With 5 fields soft combined 10 fields are sufficient to have the same detection probability. Figure A11.27: TDM signalling Word Error Rate as a function of the number of soft accumulated signalling fields for the 3-state LMS-ITS mobile channel v = 50 kmph, Es / N 0 =3,5 dB ETSI 424 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A11.28: Probability of correct TDM signalling field detection as a function of the number of received signalling fields for the 3-state LMS-ITS mobile channel v = 50 kmph, Es / N 0 =-3,5 dB Figure A11.29: Probability of correct TDM signalling field detection as a function of the number of received signalling fields for the 3-state LMS-SU mobile channel v = 3 kmph, Es / N 0 =-3,5 dB A.12 Simulation results The simulation results reported here are first based on the system configurations described above, referred to as the "Reference Cases" and briefly recalled below. They allow comparison of different physical configurations under the same high level technical parameter settings (spectrum efficiency, satellite and terminal characteristics or ETSI 425 ETSI TS 102 584 V1.1.2 (2010-XY) interleaver length), irrespective of the receiver complexity trade-off, such as memory size. In addition, supplementary cases have also been performed, referred to as "Sensitivity Analysis", that give complementary information on respective capabilities of different physical configurations, when specific constraints are relaxed/changed. Because the constraints are not relaxed/changed in the same way, direct comparison of these additional results are not straightforward. Simulated environments The characteristics of the main terminal Categories are: Table A12.17 Handheld Handheld Portable Vehicular Category 3 category 2b category 2a category 1 Antenna polarization L or C L or C L or C C G/T (dB/K) -32,1 -29,1 -24,9 -21,0 Service delivery environments/use cases/terminal categories lead to a large combination of simulations targets. The simulations reported here are restricted to the following targets: 11) satellite coverage, Category 1 terminals, LMS-SU model (Suburban); 12) satellite coverage, Category 1 terminals, LMS-ITS model (Rural and forested); 13) satellite coverage, Category 2b terminals, LMS-SU (Suburban) model; 14) terrestrial coverage, any terminal category, TU6 propagation model (fast fading). Target QoS criteria for the simulations Two QoS parameters have been computed: 15) for a global approach to QoS the ESR5 fulfilment is used, according to definition in clause A3.8.5. 16) for easy reference and reuse of terrestrial network planning efforts conducted on other broadcasting services for terrestrial coverage the FER/MFER5 are also used. Simulation parameters The following parameters have been defined to identify uniquely each case: • waveform configuration, covering all main settings of the DVB-SH waveform: - physical layer parameters:  transmission Mode: OFDM or TDM for LMS environment, and OFDM only for TU6 environment;  physical layer Modulation: QPSK and 16QAM for OFDM, QPSK and 8PSK for TDM;  physical layer FEC code rate;  physical layer Interleaver Profile: Short (S), 200 ms long and Uniform (U) or Uniform Late (UL) (see clause A.10) with a reference total duration in the range of 10 s;  OFDM simulations have been performed using a Guard Interval of 1/4; - link Layer parameters (when applicable):  link Layer code rate;  link Layer Length of the Interleaving, defined as the B+S value (see clauses 6 and A.6 for details) with a reference total duration in the range of 10 s (clause A.3.1 for the delay breakdown); • channel propagation types as described in clause A.7; • state Machine Activation (clause A.5), simulating demodulator synchronization behaviour in LMS channel; ETSI 426 ETSI TS 102 584 V1.1.2 (2010-XY) • two reference terminal speeds: 3 kmph for Handheld, and 50 kmph for vehicular; • Carrier-to-Interference ratio (C/I) and Carrier-to-Noise ratio (C/N) according to the link budget calculations (section A.9); • two types of terminals are considered for LMS simulations, each associated with a specific satellite EIRP: - category 1 terminals, with circular polarized antenna, receiving signal from a 63 dBW EIRP satellite; - category 2b terminals, with linear polarized antenna, receiving from a 68 dBW EIRP satellite; • TU6 performance is simulated independently of the terminal category, as it characterizes the required C/N in terrestrial fast fading environment; • for LMS environments, Class 1 is always used together with MPE-IFEC. The complete list of Reference Cases is given in clause A10 (at Physical Layer level) for both terrestrial and satellite applications.. Simulation results presentation The following parts of this clause are organized as follows: • summary of the results for LMS is presented in clause A.12; • presentation of the simulated performance in LMS environment, including vehicular reception in LMS-Suburban, vehicular reception in LMS-ITS, and handheld reception in LMS Suburban is reported in clause A.12.2; • presentation of the simulated performance in TU6 environment is reported in clause A.12.3. A.12.1 Summary of the Results for LMS channel The results for the Reference Cases show the following main performance differences between the configurations: 17) Physical-Layer (class 2) vs Link-Layer (class 1+ MPE-IFEC) protection The reference configuration results show that protection provided exclusively at physical layer (class 2 receivers) performs better than an "equivalent" combination of protections at Physical Layer and Link Layer (class 1 receivers with MPE-IFEC), under the same constraints. More precisely: • when the constraint of handheld and portable terminals is removed, keeping only vehicular terminals (Category 1): - in LMS-SU, ESR5 fulfillment is above 99 % for Class 2, whereas for Class 1 it is from 88 % to 98 %; - in LMS-ITS, ESR5 fulfillment is from 86 % to 100 % for Class 2, whereas Class 1 can never achieve 90 %; - class 2 often exceeds the targeted performance, even for the lowest satellite EIRP considered, both in LMS-ITS and LMS-SU. Therefore, C/N reduction for Class 2 is addressed in the "Sensitivity Analysis" clause below; - class 1 with MPE-IFEC, most likely require interleavers longer than 10 s (see clause 10 of the guidelines). Therefore, ESR5 fulfillment improvement versus increased interleaver length are addressed in the "Sensitivity Analysis" clause below; • for what concerns handheld reception (Category 2b), in LMS-SU and at the highest satellite EIRP considered, the results indicate that ESR5 fulfillment is below target for all physical configurations. However, Class 2 is able to reach 88 % while Class 1+ MPE-IFEC reaches only 62 %. 18) Uniform Late (UL) and Uniform (U) long physical layer interleavers ETSI 427 ETSI TS 102 584 V1.1.2 (2010-XY) The Uniform Late interleaver profile is specifically designed to allow fast zapping in good reception condition. Also, it allows coexistence of Class 1 and Class 2 in the same network. However, these advantages entail some performance penalty. For the targeted 90 % of ESR5 fulfillment, the Uniform Late interleaver losses with respect to Uniform Long interleaver are: • up to 4 dB in the LMS-SU, for Category 1 terminal at 50 kmph; • less than 2 dB in the LMS-ITS, for Category 1 terminal at 50 kmph; • about 1,5 dB in LMS-SU, for Category 2b terminal at 3 kmph. 19) Usage of Higher Order Modulation with Lower coding rate When a constant overall spectral efficiency is considered, several configurations (Modulation / PHY-CodeRate / IFEC- CodeRate) can be envisaged. With reference to the Class 2, the results for the reference scenarios (cases 22/23-25/26 and 35/36-38/39) are inconclusive since these cases show 100 % fulfil of the ESR5. The results for Class 1 show a slight advantage for the Lower Order Modulation with a less robust FEC protection, in particular for the LMS-ITS (cases 14 vs. 18 and 21 vs. 24). For the analyzed configurations, the results show that the Higher Order Modulation with Lower Coding Rate obtains equivalent performances only when operating in high ESR5 fulfillment regions. A.12.2 Detailed results in LMS environments: Reference Cases and Sensitivity analysis A.12.2.1 Reference Cases results The results for the Reference Cases and the LMS environments are provided in the 3 tables below, corresponding respectively to the following 3 scenarios: • table A12.18, Scenario (a): Category 1 terminal in LMS-SU, speed of 50 kmph and 63 dBW EIRP Satellite; • table A12.19, Scenario (b): Category 1 terminal in LMS-ITS, speed of 50 kmph and 63 dBW EIRP Satellite; • table A12.20, Scenario (c): Category 2b terminal in LMS-SU, speed of 3 kmph and 68 dBW EIRP Satellite. The tables give the main parameters of the waveform, in particular the type of Physical Layer interleaver, the code rates at Physical Layer and, when used, the code rates at Link Layer level. The total code rate is the product of these two code rates. The total capacity is the equivalent capacity provided after all FEC decoding and computed at the MPEG2 TS interface. The total interleaving length is about 10 s for all cases. ETSI 428 ETSI TS 102 584 V1.1.2 (2010-XY) Table A12.18: ESR5 fulfilment in Scenario (a) LMS-SU environment, about 10 s of interleaving, 63 dBW EIRP Satellite, Category 1 terminal, speed =50 kmph, State Machine = "ON" Waveform configuration TOTAL PHY ESR5 ID transmission PHY Link TOTAL capacity Modulation Interleaver fuflfilment Mode code-rate code-rate code-rate (Mbps) Profile 27 OFDM 16QAM 1/4 2/3 1/6 S 2.24 88.20% 28 OFDM 16QAM 1/5 - 1/5 U 2.67 100.00% 29 OFDM 16QAM 1/5 - 1/5 UL 2.67 99.00% 30 OFDM 16QAM 2/7 7/12 1/6 S 2.19 93.30% 31 OFDM QPSK 1/2 2/3 1/3 S 2.24 89.70% 32 OFDM QPSK 1/3 - 1/3 U 2.22 100.00% 33 OFDM QPSK 1/3 - 1/3 UL 2.22 100.00% 34 TDM 8PSK 1/3 2/3 2/9 S 2.60 96.70% 35 TDM 8PSK 2/9 - 2/9 U 2.57 100.00% 36 TDM 8PSK 2/9 - 2/9 UL 2.57 100.00% 37 TDM QPSK 1/2 2/3 1/3 S 2.60 95.00% 38 TDM QPSK 1/3 - 1/3 U 2.57 100.00% 39 TDM QPSK 1/3 - 1/3 UL 2.57 100.00% Only two configurations (ID=27 and ID=31), corresponding to Class 1 and MPE-IFEC, have ESR5 fulfillment slightly below 90 %. When Class 2 is used, performance is above 99 %. Table A12.19: ESR5 fulfillment in Scenario (b) LMS-ITS environment, about 10 s of interleaving, 63 dBW EIRP Satellite, Category 1 terminals, speed= 50 kmph, State Machine ="ON" Waveform configuration TOTAL PHY ESR5 ID transmission PHY Link TOTAL capacity Modulation Interleaver fuflfilment Mode code-rate code-rate code-rate (Mbps) Profile 14 OFDM 16QAM 1/4 2/3 1/6 S 2.24 6.80% 15 OFDM 16QAM 1/5 - 1/5 U 2.67 95.50% 16 OFDM 16QAM 1/5 - 1/5 UL 2.67 86.00% 17 OFDM 16QAM 2/7 7/12 1/6 S 2.19 6.40% 18 OFDM QPSK 1/2 2/3 1/3 S 2.24 13.30% 19 OFDM QPSK 1/3 - 1/3 U 2.22 100.00% 20 OFDM QPSK 1/3 - 1/3 UL 2.22 99.00% 21 TDM 8PSK 1/3 2/3 2/9 S 2.60 52.80% 22 TDM 8PSK 2/9 - 2/9 U 2.57 100.00% 23 TDM 8PSK 2/9 - 2/9 UL 2.57 100.00% 24 TDM QPSK 1/2 2/3 1/3 S 2.60 64.70% 25 TDM QPSK 1/3 - 1/3 U 2.57 100.00% 26 TDM QPSK 1/3 - 1/3 UL 2.57 100.00% Most configurations with Class 2 have an ESR5 fulfillment above 90 %. Configurations using 16QAM OFDM and code rate 1/5 have slightly lower performance because of a higher spectrum efficiency. Results for Class 1 and IFEC show that a 10 s IFEC gives an insufficient protection in LMS-ITS. Results for Class 2 show that for most of the cases, performance is above 99 %. ETSI 429 ETSI TS 102 584 V1.1.2 (2010-XY) Table A12.20: ESR5 fulfillment for Scenario (c) LMS-SU environment, with about 10 s of interleaving, 68 dBW EIRP Satellite, Category 2b terminals, speed = 3 kmph, State Machine = "ON" Waveform configuration TOTAL PHY ESR5 ID transmission PHY Link TOTAL capacity Modulation Interleaver fuflfilment Mode code-rate code-rate code-rate (Mbps) Profile 72 OFDM 16QAM 1/5 - 1/5 U 2.67 51.00% 73 OFDM 16QAM 1/5 - 1/5 UL 2.67 42.00% 75 OFDM QPSK 1/3 - 1/3 U 2.22 84.00% 76 OFDM QPSK 1/3 - 1/3 UL 2.22 75.00% 77 TDM 8PSK 1/3 2/3 2/9 S 2.60 48.90% 78 TDM 8PSK 2/9 - 2/9 U 2.57 87.57% 79 TDM 8PSK 2/9 - 2/9 UL 2.57 84.43% 80 TDM QPSK 1/2 2/3 1/3 S 2.60 62.10% 81 TDM QPSK 1/3 - 1/3 U 2.57 87.98% 82 TDM QPSK 1/3 - 1/3 UL 2.57 86.89% In all configurations, ESR5 fulfillment is below 90 %. QPSK with PHY coding rate of 1/3 provide performance very close to this target (88 %). Possible improvements are either the use of longer interleavers or the increase of the transmission link margin, through the reduction of the capacity, or the increase of the satellite transmit power: • Interleaver impact is analyzed for IFEC cases in clause 1.2.2.2 (figure 9). • Link margin impact is analyzed for Long Physical Interleaver case in clause 1.2.2.2 (figure 10). To the reference cases reported in the 3 tables above: NOTE 1: When compared to OFDM QPSK, the Spectrum Efficiency for OFDM-16QAM code rate 1/5 (ID# 15-16, 28-29,72-73), is slightly higher (2,7 Mbps capacity instead of 2,2 Mbps). The rationale of this difference is due to the minimum allowed coding rate, equal to 1/5. This reflects both in an higher bit-rate and in a slight performance degradation. NOTE 2: The total channel capacity refers to the aggregate bit rate of the available services; target net bit-rate per service is roughly 280 kbps, then, depending on the system configuration, the number of services can be either 8 or 9. NOTE 3: In general it can be noticed that even given the same TOTAL EFFICIENCY, the system configurations can give slightly different net bit rates. The reason is related to a different Physical Layer configuration (modulation and/or coding rate) which could end up in a slightly different waveform configuration (e.g. CU-padding in the SH-Frame composition). NOTE 4: The exact C/N for each Scenario is reported in the link-budgets in clause 1.7.6. However, the results reported here already take into account the implementation losses, i.e for typical receivers (Sect. 10.4.3): 1,1 dB for OFDM QPSK, 0,5 dB for TDM QPSK, 1,5 dB for OFDM 16QAM and 1 dB for TDM 8PSK. A.12.2.2 Sensitivity Analysis This clause provides with performances in the LMS environments when different system configurations with respect to the Reference Cases are considered, relaxing or changing some constraints as appropriate. As noted, the performance of Class 1 receiver using MPE-IFEC with interleaving length up to 10 s do not always meet the 90 % ESR5 fulfillment target. Therefore, a sensitivity analysis to the length of the protection period and to the code rate is performed with the aim of finding the optimal IFEC parameters in each scenario, keeping the same reference C/N. The results are presented in a graphical format, reporting the minimum required (Link-Layer) interleaver length as a function of the supported aggregated net-bit rate. It is important to notice that, for each graphic, the curves represent the boundary conditions: all the system configurations which belong to the left-bottom part of the plane are viable solutions. Indeed, for a given interleaver length, all the bit-rates up to the point reported in the curve are guaranteed with 90 % of ESR5 fulfillment as minimum. On the contrary, Class 2 receivers in Category 1 terminals generally exceed the 90 % ESR5 fulfillment. Therefore, a parametric analysis of the ESR5 as a function of the C/N is performed for a set of cases when Class 2 is considered. The ETSI 430 ETSI TS 102 584 V1.1.2 (2010-XY) same C/I as reported in the reference cases is kept. Influence of the type of long Physical Layer interleaver (Uniform versus Uniform Late) is also addressed. The reader is reminded to be cautious when comparing in detail the OFDM versus TDM performances from the results presented, due to: a) the guard interval value chosen for OFDM; b) the different net bit rates; c) the different implementation losses; d) the different noise bandwidth assumption made. A.12.2.3 Vehicular reception in LMS environments A.12.2.3.1 Vehicular Reception in LMS-SU and SH-A Waveform (OFDM) The simulation performances reported in this clause have been obtained in LMS-SU environment and for SH-A waveform with a 63 dBW EIRP satellite (reference cases IDs 27 to 33 in table A12.18). These cases are considered for vehicular applications; therefore a typical speed of 50 kmph has been analyzed. Figure A12.30 shows the sensitivity analysis for Class 1 terminals. The supported bit rate for the reference case is indicated by the dashed line. Under these working conditions, it is possible to increase the supported bit rate while satisfying the 90 % ESR5 fulfillment target. This is achieved by increasing the MPE-IFEC code rate and interleaving depth. To be remarked that even assuming ideal channel estimation 16QAM combined with low FEC rate does not bring any advantage compared to QPSK. On the contrary performance and generally slightly worse for 16QAM. 60 50 Interleaver Length (s) 40 30 20 10 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Supported Bitrate (Mbps) QPSK 16QAM Figure A12.30: SH-A, Class 1 – QPSK 1/2 and 16QAM 1/4 – LMS-SU - 50 kmph - 63 dBW EIRP Satellite Different Link-Layer configuration fulfilling the ESR5(20) criterion at 90 % (corresponding reference cases IDs 27 to 31) Figure A12.31 shows the sensitivity analysis for Class 2 terminals. The C/N setting used for the reference case was 11,2 dB (shown by the vertical dashed line). The required C/N to guarantee the 90 % fulfillment of ESR5 is found to be below 5 dB, i.e. 7 dB below the reference configuration C/N setting. ETSI 431 ETSI TS 102 584 V1.1.2 (2010-XY) ESR5(20) fulfillment (%) 100 95 90 85 80 75 0 2 4 6 8 10 12 14 C/N (dB) Figure A12.31: SH-A, Class 2: Uniform Long Interleaver Profile – QPSK 1/3 – LMS-SU - 50 kmph Sensitivity Analysis to C/N value for an OFDM modulation (corresponding reference cases IDs 31 and 32) A.12.2.3.2 Vehicular Reception in a LMS-SU and SH-B Waveform (TDM) The simulation performances reported in this clause have been obtained in an LMS SU channel and for a SH-B waveform with a 63 dBW EIRP satellite (reference cases IDs 34 to 39 in table A12.18). These cases are considered for vehicular applications; therefore a typical speed of 50 kmph has been analyzed. Figure A12.32 shows the sensitivity analysis for Class 1 terminals. The same conclusions as for the corresponding case of SH-A can be drawn. ETSI 432 ETSI TS 102 584 V1.1.2 (2010-XY) 60 50 Interleaver Length (s) 40 30 20 10 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Supported Bitrate (Kbps) 8PSK QPSK Figure A12.32: SH-B, Class 1: – QPSK ½ and 8PSK1/3 - LMS-SU - 50 kmph - 63 dBW EIRP Satellite Different Link-Layer configuration fulfilling the ESR5(20) criterion at 90 % (corresponding reference cases IDs 34 to 37) Figure A12.33 shows the sensitivity analysis for Class 2. The C/N setting used for the reference case was 12,3 dB (shown by the vertical dashed line). The required C/N to guarantee the 90 % fulfillment of ESR5 is found to be below 4 dB, i.e. 8 dB below the reference configuration setting. 100 ESR5(20) fulfillment (%) 95 90 85 80 75 0 2 4 6 8 10 12 14 C/N (dB) Figure A12.33: SH-B, Class 2: with Uniform Long Interleaver Profile – QPSK 1/3 – LMS-SU - 50 kmph Sensitivity Analysis to C/N value for an OFDM modulation (corresponding reference cases IDs 37 and 38) ETSI 433 ETSI TS 102 584 V1.1.2 (2010-XY) A.12.2.3.3 Vehicular Reception in a LMS ITS and DVB SH-A Waveform (OFDM) The simulation performances reported in this clause have been obtained in LMS-ITS environment and for SH-A waveform with a 63 dBW EIRP satellite (reference cases IDs 14 to 20 in table A12.19). These cases are considered for vehicular applications; therefore a typical speed of 50 kmph has been analyzed. As can be seen in figure A12.34, in LMS-ITS, an SH-A Class 1 receiver cannot reach the target capacity at 63 dBW satellite EIRP for the reference case (represented as vertical dashed line). 60 50 Interleaver Length (s) 40 30 20 10 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Supported Bitrate (Mbps) QPSK 16QAM Figure A12.34: SH-A, Class 1: - 16QAM 1/4 and QPSK 1/2– LMS-ITS - 50 kmph - 63 dBW EIRP Satellite Different Link-Layer configuration fulfilling the ESR5(20) criterion at 90 % (corresponding reference cases IDs 14 to 18) Figure A12.35 shows the sensitivity analysis for Class 2. The C/N setting used for the reference case was 11,2 dB (shown by the vertical dashed line). The required C/N to guarantee the 90 % fulfillment of ESR5 is found to be 9 dB, i.e. 2 dB below the reference configuration setting. ETSI 434 ETSI TS 102 584 V1.1.2 (2010-XY) ESR5(20) fulfillment (%) 100 95 90 85 80 75 7 8 9 10 11 12 C/N (dB) Figure A12.35: SH-A, Class 2 with Uniform Long Interleaver Profile- QPSK 1/3 – LMS-ITS - 50 kmph Sensitivity Analysis to C/N value for a OFDM modulation (corresponding reference cases IDs 18 and 19) A.12.2.3.4 Vehicular Reception in a LMS ITS and DVB SH-B Waveform (TDM) The simulation performances reported in this clause have been obtained in an LMS ITS channel and for a SH-B waveform with a 63 dBW EIRP satellite (reference cases IDs 21 to 26 in table A12.19). These cases are considered characteristic for vehicular applications; in particular a typical speed of 50 kmph has been analyzed. From figure A12.36, it can be noticed that a SH-B Class 1 receiver is able to guarantee an aggregate traffic slightly lower than the target capacity for the reference case (vertical dashed line in the following figure) up to 2,5 Mbps with roughly 25 s of interleaver length at the reference C/N of 12,8 dB. The performances for the QPSK modulation (with a code rate of 1/2 at physical layer) are close to those for an 8PSK modulation (with a code rate of 1/3 at physical layer). Indeed, under this condition, the advantage for the QPSK for the reference 10 s of interleaver length is limited to about 10 % in terms of bit-rate or interleaver-length. ETSI 435 ETSI TS 102 584 V1.1.2 (2010-XY) 60 50 Interleaver Length (s) 40 30 20 10 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Supported Bitrate (Mbps) QPSK 8PSK Figure A12.36: TDM, Class 1 – QPSK ½ and 8PSK 1/3 – LMS-ITS - 50 kmph - 63 dBW EIRP Satellite Different Link-Layer configuration fulfilling the ESR5(20) criterion at 90 % (corresponding reference cases IDs 21 to 24) Figure A12.37 shows the same sensitivity analysis also for Class 2 under the same reference C/N simulation condition of 12,8 dB. Also in this case QPSK modulation performances are slightly better to those for an 8PSK modulation. Under this condition, both QPSK and 8PSK modulations guarantee high throughput for the reference 10 s of interleaver length or equivalently the possibility to reduce the interleaver length to get the target capacity up to 2,5 Mbps. 40 Interleaver Length (sec) 30 20 10 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 Supported Bit Rate (Mbps) QPSK 8PSK Figure A12.37: TDM, Class 2– QPSK – LMS-ITS - 50 kmph - 63 dBW EIRP Satellite Different Physical-Layer configuration fulfilling the ESR5(20) criterion at 90 % (corresponding reference cases IDs 21 to 24) ETSI 436 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A12.38 shows the sensitivity analysis for Class 2. The C/N setting used for the reference case was 12,3 dB (shown by the vertical dashed line). The required C/N to guarantee the 90 % fulfillment of ESR5 is found to be 8 dB, i.e. 4 dB below the reference configuration setting. 100 ESR5(20) fulfillment (%) 95 90 85 80 75 6 7 8 9 10 11 12 13 14 C/N (dB) Figure A12.38: SH-B, Class 2 with Uniform Long Interleaver Profile - QPSK – LMS-ITS - 50 kmph Sensitivity Analysis to C/N value for a TDM modulation (corresponding reference cases IDs 24 and 25) A.12.2.3.5 Category 2b reception in LMS-SU at 68 dBW satellite EIRP The simulation performances reported in this clause have been obtained in an LMS SU channel with a 68 dBW EIRP satellite (reference cases IDs 72 to 82 in table A12.20). These cases are considered for handheld applications; therefore a typical speed of 3 kmph has been analyzed. As can be seen in figure A12.39, in LMS-ITS, an SH-A Class 1 receiver cannot reach the target capacity of the reference case (represented as vertical dashed line) at 68 dBW satellite EIRP. QPSK performance are better than 8PSK. ETSI 437 ETSI TS 102 584 V1.1.2 (2010-XY) 60 50 Interleaver Length (s) 40 30 20 10 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Supported Bitrate (Mbps) QPSK 8PSK Figure A12.39: SH-B, Class 1: – QPSK ½ and 8PSK 1/3 – LMS-SU - 3 kmph - 68 dBW EIRP Satellite Different Link-Layer configuration fulfilling the ESR5(20) criterion at 90 % (corresponding reference cases IDs 77 and 78) Figures A12.40 and A12.41 give the sensitivity analysis for Class 2. The C/N has been increased with respect to the reference value of 5,1 dB and 6,2 dB for the reference OFDM and TDM cases respectively. For ID75 configuration the required C/N to guarantee the 90 % fulfillment of ESR5 criterion is 8 dB i.e. 3 dB above the reference C/N value.. For ID81 and ID82 configurations the required C/N to guarantee the 90 % fulfillment of ESR5 criterion is 8 dB i.e. 2 dB above the reference C/N value. 100 ESR5(20) fulfillment (%) 95 90 85 80 75 4 6 8 10 12 14 C/N (dB) Figure A12.40: SH-A, Class 2: Uniform Interleaver Profile – QPSK 1/3 – LMS-ITS - 3 kmph Sensitivity Analysis to C/N value for a TDM modulation (corresponding reference cases ID 75) ETSI 438 ETSI TS 102 584 V1.1.2 (2010-XY) ESR5(20) fulfillment (%) 100 95 90 85 80 75 4 6 8 10 12 14 C/N (dB) Figure A12.41: SH-B, Class 2: with Uniform Interleaver Profile - QPSK – LMS-ITS - 3 kmph Sensitivity Analysis to C/N value for a TDM modulation (corresponding reference cases IDs 80 and 81) A.12.2.4 Uniform Late (UL) and Uniform (U) long interleavers This clause provides with performances comparison between the interleaver profiles Uniform Long (U) and Uniform Late (UL) as defined in clause A.4 and used for the simulation campaign for the Class 2 receiver performance evaluation. Only SH-B waveform is considered, but the results are considered of general validity and applicable also to SH-A profile. Both interleaver profiles show excellent performances in the reference system configurations, if typical satellite EIRP is considered and vehicular terminals Category 1 are addressed (e.g. cases 19 and 20 or 25 and 26 in table A12.19). When the system is operating in such high ESR5 fulfillment regions, the performances of the two different profiles are almost equivalent and it is also possible to exploit the fast-zapping feature allowed by the UL profile. On the contrary, when the target 90 % of ESR5 fulfillment is considered, the U interleaver profile is characterized by a lower C/N working point compared to UL thus providing extra link margin for possible system optimization in terms of satellite EIRP, terminal RF performances and spectral efficiency is possible. In particular, this clause quantifies this UL interleaver advantage for the reference scenarios. Three different cases are analyzed in this clause: s) LMS-SU scenario at 50 kmph for vehicular reception (figure A12.42); t) LMS-ITS scenario at 50 kmph for vehicular reception (figure A12.43 and figure A12.45) u) LMS-SU scenario at 3 kmph for handheld reception (figure A12.44). With reference to a), the Uniform Long interleaver meets the 90 % ESR5 fulfillment at roughly 4 dB of C/N with a saving of 4 dB with respect the Uniform Late. This represent the worst case within the selected cases, indeed this difference reduces below 1 dB for both b) and c). Figure A12.45 shows the comparison results when different Physical-Layer configuration fulfilling the ESR5 criterion at 90 % are considered. When the interleaver length is set to the reference duration of 10 seconds (shown by the horizontal dashed line), the U interleaver allows supporting a bit rate higher than 4 Mbps with an increase of rougly 1 Mbps with respect the UL. Both configurations ensure a supported bit rate higher than the reference throughput (shown ETSI 439 ETSI TS 102 584 V1.1.2 (2010-XY) by the vertical dashed line), allowing for a interleaver duration reduction.When the same throughput is considered, the U interleaver requires a duration almost half of the UL one. 100% 95% ESR5(20) fulfillment (%) 90% 85% 80% 75% 0 2 4 6 8 10 12 14 C/N (dB) TDM - case38 (U) TDM - case39 (UL) Figure A12.42: SH-B , Class 2: - QPSK 1/3 – LMS-SU - 50 kmph Sensitivity Analysis to C/N value for a TDM modulation (corresponding reference case IDs 38 and 39) 100% 90% ESR5(20) fulfillment (%) 80% 70% 60% 50% 40% 30% 20% 0 2 4 6 8 10 12 14 C/N (dB) TDM - case25 (U) TDM - case26 (UL) Figure A12.43: SH-B, Class 2: - QPSK 1/3 – LMS-ITS - 50 kmph Sensitivity Analysis to C/N value for a TDM modulation (corresponding reference case IDs 25 and 26) ETSI 440 ETSI TS 102 584 V1.1.2 (2010-XY) 105% 100% ESR5(20) fulfillment (%) 95% 90% 85% 80% 75% 0 2 4 6 8 10 12 14 C/N (dB) TDM - case 81 (U) TDM - case 82 (UL) Figure A12.44: TDM, Class 2: - QPSK 1/3 – LMS-SU - 3 kmph Sensitivity Analysis to C/N value for a TDM modulation (corresponding reference case IDs 81 and 82) 40 Interleaver Length (sec) 30 20 10 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 Supported Bit Rate (Mbps) QPSK (UL) 8PSK (UL) QPSK (U) 8PSK (U) Figure A12.45: TDM, Class 2: - QPSK and 8PSK – LMS-ITS - 50 kmph Different Physical-Layer configuration fulfilling the ESR5(20) criterion at 90 % ETSI 441 ETSI TS 102 584 V1.1.2 (2010-XY) A.12.2.5 Demodulator state machine performance impact This clause provides with performances assessment when the demodulator state machine is not modelled (A.5). The results are to be considered as performances for an ideal receiver with zero re-acquisition time. In this clause only SH-B waveform is considered and for a specific case (case 25 in table A12.19), but the results are considered of general validity and applicable also to different environment and SH-A profile. The case analyzed case is the LMS-ITS scenario at 50 kmph for vehicular reception (figure A12.46). When the ideal demodulator is considered (without state machine), the target 90 % of ESR5 fulfilment is achieved at less than 7 dB of C/N with a saving of more than 1 dB with respect to the reference demodulator. This represent the best case where the behaviour of a real demodulator is not considered and corresponding losses not accounted for. 100 95 ESR5(20) fulfilment (%) 90 85 80 75 0 2 4 6 8 10 12 14 C/N (dB) TDM - case 25 TDM - case 25 (without SM) Figure A12.46: SH-B, Class 2: - QPSK 1/3 – LMS-ITS - 50 kmph Sensitivity Analysis to C/N value for a TDM modulation (corresponding reference case ID 25) A.12.2.6 C/I performance impact This clause provides with performances assessment when different values of co-channel useful signal over interference are considered (C/I as defined in A.9). It is to recall that satellite link C/I and C/N are provided separately as the interference is suppose to fade together with the signal and is simulated like an extra independent noise source modulated by the same fade as the useful signal. In this clause only the SH-B waveform is considered but the findings can be considered of general validity and applicable to different environment and SH-A profile with comparable coding and modulation formats. For the selected cases the ESR5performances are almost independent from the C/I value when typical satellite multibeam antenna values are considered. The very limited sensitivity to the C/I is noted, thus confirming that the performances of the selected configurations are not co-channel-interference-limited. Two different physical layer configuration are analyzed in this clause: ETSI 442 ETSI TS 102 584 V1.1.2 (2010-XY) v) LMS- ITS scenario at 50 kmph for vehicular reception, QPSK-1/3, 63 dBW EIRP Satellite (case 25 in table A12.19, figure A12.47). w) LMS-ITS scenario at 50 kmph for vehicular reception, 8PSK-2/5, 63 dBW EIRP Satellite (figure A12.48).. In the case a) for an ESR5 90% fulfillment the C/I reduction from 16 to 12 dB causes only a required C/N increase of 0.1 dB. This is because we are using a highly protected physical layer configuration (QPSK r=1/3), operating with quite high link margin for counteracting the LMS channel and because interference follows the LMS channel fluctuations. In the case b) for an ESR5 90% fulfillment the C/I reduction from 16 to 12 dB causes only a required C/N increase of 0,2 dB. The slight higher C/I sensitivity observed in this case is due to the use of a less protected physical layer configuration (8PSK r=2/5). In any case for a satellite mobile broadcasting system operating with spectral efficiencies less than 1 bps/Hz the antenna C/I is not a very critical parameter up to 10 dB isolation. 100% 95% ESR5(20) fulfilment (%) 90% 85% 80% 75% 0 2 4 6 8 10 12 14 TDM - case 25 - C/I = 10 dB C/N (dB) TDM - case 25 - C/I = 12 dB TDM - case 25 - C/I = 16 dB Figure A12.47: SH-B, Class 2: - QPSK 1/3 – LMS-ITS - 50 kmph Sensitivity Analysis to C/I value for a TDM modulation (corresponding reference case ID 25) ETSI 443 ETSI TS 102 584 V1.1.2 (2010-XY) 100% 95% ESR5(20) fulfilment (%) 90% 85% 80% 75% 5 7 9 11 13 15 17 19 TDM - 8PSK 2/5 - C/I = 10C/N (dB) TDM - 8PSK 2/5 - C/I = 12 dB dB TDM - 8PSK 2/5 - C/I = 16 dB Figure A12.48: SH-B, Class 2: - 8PSK 2/5 – LMS-ITS - 50 kmph Sensitivity Analysis to C/I value for a TDM modulation A.12.2.6 ESR5 fulfilment at different spectral efficiencies This clause provides with DVB-SH ESR5 performance assessment when different spectral efficiencies are considered. Only SH-B waveform results are reported but the findings are of general validity and also applicable with good approximation to different LMS environments and to the SH-A profile. The case analyzed corresponds to the LMS-ITS scenario for vehicular reception at 50 kmph with a 68 dBW EIRP satellite (figure A12.49). An Uniform Long interlaver with the reference duration of 10 seconds has been considered as well an ideal demodulator. QPSK and 8PSK modulation have been simulated for different physical layer FEC coding rates configurations corresponding to different supported bit rates..It is remarked that the performance of QPSK and 8PSK are essentially identical in terms of ESR5 fullfillemnt ratio as a function of the bit rate. 100% 8PSK 1/3 QPSK 1/4 QPSK 1/3 ESR5(20) fulfillment (%) QPSK 1/2 8PSK 2/9 80% 8PSK 2/5 60% QPSK 2/3 40% 0 1 2 3 4 5 6 Supported Bit Rate (Mb/sec) QPSK 8PSK Figure A12.49: SH-B, Class 2: - QPSK and 8PSKLMS-ITS - 50 kmph Sensitivity Analysis to supported bit rate for a TDM modulation ETSI 444 ETSI TS 102 584 V1.1.2 (2010-XY) A.12.3 Reception in TU6 environment TU6 propagation model is used as reference for characterization of the small scale fading effects encountered in terrestrial environment reception. Thus, it only applies to the DVB-SH OFDM waveform. Performance are defined in terms of required C/N, to be used for Terrestrial Network Planning (clause 11). Because of the hybrid structure of the DVB-SH system, two cases of terrestrial transmission are possible: • terrestrial retransmission of the satellite signal of an SH-A (SFN) system, where in particular long interleavers, at Physical Layer level, or at IFEC level, may be implemented; • dedicated terrestrial transmission, in case of retransmission of the satellite content for SH-A (non-SFN) or SH-B system, or in case of terrestrial-only local content broadcasting. This clause gives the simulation performance in a TU6 channel, for an OFDM waveform, with a terminal speed of 3 kmph and 50 kmph. They apply to a reference service of 280 kbps. Overall capacity is calculated on the basis of a 1/8 guard interval, as FER/ESR5 simulation results in TU6 are not sensitive to guard interval larger than 1/8. A.12.3.1 FER and ESR5 relationship The receiver threshold performance is a function of the quality criteria. Two criteria are commonly used in broadcasting systems, which are the FER and the ESR5 (see clause A.8). From the various simulations performed with the DVB-SH waveform in TU6, the relationship between ESR5 and FER reported in figure A12.50 is observed, whatever the code rate and the speed of the receiver. 100 90 80 ESR5 (%) Fulfilement 70 60 50 40 30 20 10 0 0 5 10 15 20 FER (%) Figure A12.50: ESR5 fulfillment performance versus FER performance for TU6 environment, for various speed, code rate and modulation Each plot on the figure refers to a different order of modulation (QPSK or 16QAM) and code rate (from ½ to 1/5). From this plot, 1-ESR5 = 99 % is equivalent to an FER = 1 %. In terms of C/N threshold difference between FER = 5 % and 1-ESR5 = 99 % (or 1 % FER), is typically less than 0,5 dB at 50 kmph, and is about 1 dB at 3 kmph. Exact values ETSI 445 ETSI TS 102 584 V1.1.2 (2010-XY) are provided in tables A12.21 and A12.22 as results of simulations. According to above results on FER and ESR relationship, FER 1% performance can be considered as equivalent to 1-ESR5(20) 99%. A.12.3.2 FER performance in TU6 Tables A12.21 and A12.22 give the FER=5 % and FER 1% performances of the DVB-SH OFDM waveform for all possible DVB-SHA physical layer coding rates and modulation (with short interleaver (200 ms QPSK and 100 ms for 16QAM with CM= 5 for both cases) and Uniform Late interleaving of 10s for both QPSK and 16QAM cases 1-4, Table A.9). The reference configuration of 5 MHz, FFT 2K and GI of ¼ and two speeds ( 3 km/h and 50 km/h) have been taken into account. Table A12.21: FER performance in TU6 (Short) Code rate PHY Capacity C/N(dB) @ C/N(dB) @ (Mbps) FER=5 % FER=1% or 1- ESR5(20) 99 TU6 OFDM QPSK - 50 kmph 1/5 1,333 -2,4 -2,2 2/9 1,481 -1,9 -1,5 1/4 1,679 -1,1 -0,8 2/7 1,876 -0,4 0 1/3 2,222 0,7 1 2/5 2,666 2,2 2,8 1/2 3,357 4 4,6 2/3 4,443 7,5 8,1 TU6 OFDM 16QAM - 50 kmph 1/5 2,666 2,8 3,2 2/9 2,962 3,5 4,1 1/4 3,357 4,4 5,1 2/7 3,752 5,1 5,8 1/3 4,443 6,5 7 2/5 5,332 8,5 9,5 1/2 6,714 10,8 11,5 2/3 8,887 15,1 16,5 TU6 OFDM QPSK - 3 kmph 1/5 1,333 -1 -0,1 2/9 1,481 -0,6 0,6 1/4 1,679 0,3 1,2 2/7 1,876 1,1 2,1 1/3 2,222 2 3 2/5 2,666 3,7 4,6 1/2 3,357 5,2 6,3 2/3 4,443 8,6 9,8 TU6 OFDM 16QAM - 3 kmph 1/5 2,666 4,5 5,7 2/9 2,962 5,3 6,7 1/4 3,357 6,2 7,5 2/7 3,752 6,9 8,2 1/3 4,443 8,1 9,4 2/5 5,332 10,1 11,3 1/2 6,714 12,2 13,7 2/3 8,887 15,8 18 ETSI 446 ETSI TS 102 584 V1.1.2 (2010-XY) Table A12.22 : FER performance in TU6 with Uniform Late interleaving Code rate PHY Capacity C/N (dB)@ C/N(dB @ (Mbps) FER=5 % FER=1%) or1- ESR5(20) 99 TU6 OFDM QPSK - 50 kmph 1/5 1.333 -2.7 -2.4 2/9 1.481 -2.2 -1.9 1/4 1.679 -1.3 -1 2/7 1.876 -0.5 -0.2 1/3 2.222 0.5 0.8 2/5 2.666 2.2 2.5 1/2 3.357 3.8 4.4 2/3 4.443 7.5 8 TU6 OFDM 16QAM - 50 kmph 1/5 2.666 2.3 2.7 2/9 2.962 3 3.4 1/4 3.357 4 4.5 2/7 3.752 4.5 5.2 1/3 4.443 6.1 6.6 2/5 5.332 8.2 8.6 1/2 6.714 10.4 11 2/3 8.887 14.5 15.5 TU6 OFDM QPSK - 3 kmph 1/5 1.333 -2.1 -1.3 2/9 1.481 -1.2 -0.7 1/4 1.679 -0.5 0.1 2/7 1.876 0.4 1 1/3 2.222 1.4 2.1 2/5 2.666 2.9 3.6 1/2 3.357 4.5 5.2 2/3 4.443 7.8 8.6 TU6 OFDM 16QAM - 3 kmph 1/5 2.666 3.2 4 2/9 2.962 3.9 4.7 1/4 3.357 5 5.7 2/7 3.752 5.8 6.4 1/3 4.443 6.6 7.6 2/5 5.332 8.9 9.6 1/2 6.714 10.8 11.8 2/3 8.887 14.5 15.5 From Tables A12.21 and A12.22 , it may be noted that: • with short interleaving (class 1 demodulator), C/N thresholds for ESR5 fulfilment at 99% at 3 kmph are typically 1.5/2.0 dB worse than those at 50 kmph when QPSK is considered; between 2.0/2.5 dB with 16QAM; • with Uniform Late interleaving of 10 s (class 2 demodulator), the improvement at 3 km/h is around 1 dB for QPSK, and 2 dB for 16QAM, while, as expected, at 50 km/h, the improvement is less significant and in average limited to 0,5 dB for both QPSK and 16QAM. A.12.3.3 ESR5 Performance in TU This clause reports the ESR5 performance in TU, arranged in terms of Physical interleaver length, and IFEC definition. The following configurations have been considered: a) Long Physical layer interleaving, without IFEC; b) Short Physical layer interleaving, with and without IFEC. Performance versus Long Physical layer interleaving length ETSI 447 ETSI TS 102 584 V1.1.2 (2010-XY) Short and Long Physical layer interleaving configurations are analyzed first. No IFEC is considered. The short interleaver is 100 ms long for 16QAM, and 200 ms long for QPSK. The UL interleaver is 10 s long, with half of the interleaver paths located within 200 ms (cases 1-4, Table A.9). Table A12.23 gives the ESR5 simulated performance based on FER simulation measurement in a TU6 channel, for an OFDM waveform, with a terminal speed of 3 kmph and 50 kmph. In both interleaver configurations, the same capacity is provided, as no IFEC is used. Guard Interval of 1/8 has been considered The results are reported for a ESR5 fulfilment of 99 %. The improvement on the C/N threshold between short physical interleaver and UL interleaver is up to 0,5 dB at 50 kmph with both modulations, 1 dB at 3 kmph with QPSK, and about 1.8 dB at 3 kmph with 16APSK. Table A12.23: C/N threshold in TU6 for ESR5 fulfilment of 99%, UL versus S interleaver (GI=1/8) Code Rate Interleaver Capacity C/N (dB) C/N (dB) PHY PHY (Mbps) @ 3 kmph @ 50 kmph TU6 OFDM 16QAM 1/3 S 4.9 9.4 7.0 1/3 UL 4.9 7.6 6.6 ¼ S 3.7 7.5 5.1 ¼ UL 3.7 5.7 4.5 1/5 S 3.0 5.7 3.2 1/5 UL 3.0 4.0 2.7 TU6 OFDM QPSK ½ S 3.7 6.3 4.6 ½ UL 3.7 5.2 4.4 1/3 S 2.5 3.0 1.0 1/3 UL 2.5 2.1 0.8 Performance with and without IFEC Performance of configurations with different IFEC setting are analyzed in this clause. In all cases Class 1 demodulator with short physical interleaver has been considered (cases 1-4, Table A.9). Tables A12.24 C/N threshold and ESR5 fulfilment in TU6 @ 3Kmph, with and without IFEC and Erreur ! Source du renvoi introuvable. give the ESR5 simulation performance in a TU6 channel, for an OFDM waveform, with a terminal speed of 3 kmph and 50 kmph. In most configurations, the same capacity after IFEC decoding is provided. Related effect on the required C/N of IFEC code rate together with IFEC interleaving length (B+S) is analyzed. The QoS target is 99 % of ESR5 fulfilment. At 3 kmph, which is somewhat representative of handheld reception in terrestrial environment, it is shown that: • In case of QPSK, same capacity is achieved with FEC code rate ½ plus IFEC rate 2/3 and FEC with code rate 1/3 and no IFEC. As expected a longer IFEC interleaver allows to improve the C/N threshold but the performance remains inferior to the only-phisical layer approach with code rate 1/3 and short interleaver (200ms); • In case of 16QAM it is not possible a direct comparison: slightly higher capacity is achieved in the case of only-phisical layer approach, with code rate 1/5 and no IFEC, at the expense of a larger required C/N. This is due to the fact that the selected 16QAM 1/5 configuration is deliberately more spectrum efficient. In a SH-A SFN configuration, IFEC settings defined for the satellite link will marginally penalize the reception in terrestrial environment for low speed users. However in case of SH-B or SH-A MFN it will be possible to get lower C/N threshold with low zapping time using physical layer FEC only and short interleaver (200 ms). In case of dedicated terrestrial transmission, the introduction of IFEC will offer fine tuning system capability. Typically, addition of IFEC will provide the possibility to control of the quality of service for each service based on high flexibility capability. ETSI 448 ETSI TS 102 584 V1.1.2 (2010-XY) Table A12.24 C/N threshold and ESR5 fulfilment in TU6 @ 3Kmph, with and without IFEC Code Rate Interleaver Capacity C/N (dB) Interleaving 1-ESR5 (%) PHY PHY Code Rate (Mbps) length (s) IFEC TU6 OFDM QPSK - 3 kmph 1/3 S - 2.5 3.0 0.200 99.00 1/2 S 0.67 2.5 5.5 10 99.50 1/2 S 0.67 2.5 5.0 16 99.20 1/2 S 0.67 2.5 4.5 30 99.00 TU6 OFDM 16QAM - 3 kmph 1/5 S - 3.0 5.7 0.100 99.00 1/4 S 0.67 2.5 5.0 10 to 16 > 99.20 1/4 S 0.60 2.2 5.0 10 99.30 1/4 S 0.42 1.5 4.0 20 to 30 > 99.10 IFEC Synthetic overall Performance Figure A12.51 gives a synthetic overview of the performance achieved at ESR5(20) @ 99 % criteria for all configurations, in TU6 at 3 kmph, and some IFEC configurations a) The different configurations are represented according to the legends; b) UL configurations and short configurations are represented by the different lines c) IFEC configurations are represented with larges squares DVB-SH includes 8 code rate settings at physical layer, from 1/5 to 2/3. All of them are presented in Figure A22 with QPSK modulation, and 3 with 16QAM modulation. The full set of code rates, together with the flexibility of IFEC, allows to provide a very fine system capacity optimization capability. It can be seen that class 1 demodulator with all protection at physical layer outperforms class 1 + IFEC demodulator for all the configurations reported. The C/N threshold gain is in the order of 1 dB for QPSK and 1.5 dB for 16QAM. The other advantage is that the FEC only configuration has a much lower zapping time than the one using the IFEC. Simulated Capacity at 3 km/h TU6 for ESR (5) @ 99 % 10 QPSK_Short_TU6_3km/h 9 QPSK_UL_TU6_3km/h 8 16QAM_Short_TU6_3km/ Capacity (Mb/s) 7 h 16QAM_UL_TU6_3km/h 6 QPSK_1/2_IFEC_26s_TU 5 6_3km/h 4 16QAM_1/4_IFEC_TU6_3 km/h 3 2 1 -3 -1 1 3 5 7 9 11 13 15 C/N (dB) Figure A12.51 : C/N performances for ESR (5) @ 99 % in TU6 3 km/h ETSI 449 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A12.52 gives a synthetic overview of the performance achieved at ESR5(20) @99 % criteria for all above configurations, in TU6 at 50 kmph. .Available dumps covered the IFEC performance for QPSK 1/2. Simulated Capacity at 50 km/h TU6 for ESR (5) @ 99 % 10 QPSK_Short_TU6_50km/h 9 QPSK_UL_TU6_50km/h 8 16QAM_Short_TU6_50km/h 7 Capacity (Mb/s) 16QAM_UL_TU6_50km/h 6 QPSK_1/2_IFEC_TU6_50 km/h 5 4 3 2 1 -3 -1 1 3 5 7 9 11 13 15 C/N (dB) Figure A12.52 C/N performances for ESR5(20) @ 99 % in TU6 50 km/h A.13 Experimental results A13.1 Laboratory results A.13.1.1 Terrestrial channels A13.1.1.1 Configurations For the terrestrial channel tests, the baseline configurations tested are described in the following table. A TU6 channel model (see clause A.7) has been used. For all configurations the impact of speed on required C/N for a QoS criterion corresponding to FER 5 %.has been analysed. Table A13.25: Baseline Terrestrial channels configuration BW Constellation Code FFT GI Short Class 2 IFEC length Capacity Capacity with (MHz) rate size Interleaving length and rate (Mb/s) IFEC 5 QPSK 1/5 2k 1/8 265 ms U 9s N/A 1,48 1,037 5 QPSK 2/9 2k 1/8 265 ms U 9s N/A 1,65 1,152 5 QPSK 1/4 2k 1/8 265 ms U 9s N/A 1,87 1,306 5 QPSK 2/7 2k 1/8 265 ms U 9s N/A 2,14 1,498 ETSI 450 ETSI TS 102 584 V1.1.2 (2010-XY) 5 QPSK 1/3 2k 1/8 265 ms U 9s Same as 2,47 LMS 1,728 5 QPSK 2/5 2k 1/8 265 ms U 9s N/A 2,96 2,074 5 QPSK 1/2 2k 1/8 265 ms U 9s N/A 3,73 2,611 5 QPSK 2/3 2k 1/8 265 ms U 9s N/A 4,94 3,456 5 16 QAM 1/5 2k 1/8 132 ms U 9s N/A 2,96 2,074 5 16 QAM 2/9 2k 1/8 132 ms U 9s N/A 3,29 2,304 5 16 QAM 1/4 2k 1/8 132 ms U 9s N/A 3,73 2,611 5 16 QAM 2/7 2k 1/8 132 ms U 9s N/A 4,28 2,995 5 16 QAM 1/3 2k 1/8 132 ms U 9s Same as 4,94 LMS 3,456 5 16 QAM 2/5 2k 1/8 132 ms U 9s N/A 5,92 4,147 5 16 QAM 1/2 2k 1/8 132 ms U 9s N/A 7,46 5,222 5 16 QAM 2/3 2k 1/8 132 ms U 9s N/A 9,87 6,912 Note : the different short interleaving lengths provided in the table correspond to a value compatible with approx. 4 Mbits on chip memory as minimum requirement in Chapter 10 for receivers Class 1. A part from these configurations, the impact of different parameters has been studied for some modulation-coding schemes. More precisely the following parameters have been considered: • Different bandwidth; • Different FFT sizes; • Different GI; • IFEC length; • Impact of Long Interleaving; • Uniform Long interleaving length. A.13.1.1.2 Synthetic results Table A13.26 summarizes the various C/N values required for achieving a FER=5 %. Table A13.26: Summary of results Mod Code Rate C/N C/N C/N Low Doppler Medium Doppler Doppler Max (dB) (dB) (dB) frequency frequency frequency Doppler @ Low @ (Hz) (Hz) (Hz) frequency speed Medium @ 3 dB (Hz) speed loss wrt @ > 30 dB medium loss wrt speed medium speed 1/5 0,5 0,0 3,0 10 334 639 669 2/9 0,8 0,3 3,3 10 325 620 650 QPSK 1/4 1,3 0,8 3,8 10 316 601 631 2/7 2,0 1,3 4,3 10 306 583 613 1/3 2,8 2,0 5,0 10 288 545 575 2/5 4,3 3,5 6,5 10 278 526 556 1/2 5,8 5,3 8,3 10 259 489 519 2/3 9,5 9,3 12,3 10 222 414 444 1/5 5,9 3,8 6,8 10 275 520 550 ETSI 451 ETSI TS 102 584 V1.1.2 (2010-XY) 2/9 6,3 4,4 7,4 10 269 508 538 16QAM 1/4 7,2 5,0 8,0 10 256 483 513 2/7 7,8 5,9 8,9 10 244 458 488 1/3 9,1 7,2 10,2 10 238 445 475 2/5 10,9 9,1 12,1 10 219 408 438 1/2 12,8 11,3 14,3 10 206 383 413 2/3 17,5 16,3 19,3 10 175 328 350 The following two figures represent the graphical representation of the previous table. DiBcom SHA2 Terrestrial Performances AWGN and TU6 @FER5% 20dB 5MHz 2k 1/8 QPSK 1/5 Class1 no iFEC 5MHz 2k 1/8 QPSK 2/9 Class1 no iFEC 5MHz 2k 1/8 QPSK 1/4 Class1 no iFEC 18dB 5MHz 2k 1/8 QPSK 2/7 Class1 no iFEC 5MHz 2k 1/8 QPSK 1/3 Class1 no iFEC 16dB 5MHz 2k 1/8 QPSK 2/5 Class1 no iFEC 5MHz 2k 1/8 QPSK 1/2 Class1 no iFEC 14dB 5MHz 2k 1/8 QPSK 2/3 Class1 no iFEC 12dB 10dB 8dB 6dB 4dB 4,6dB 2dB 1,9dB 0dB 0,5dB -0,7dB -2dB -1,5dB -2,8dB -2,1dB -4dB -3,6dB 1Hz 10Hz 100Hz 1000Hz Figure A13.53 : QPSK AWGN and terrestrial performances DiBcom SHA2 Terrestrial Performances AWGN and TU6 @FER5% 5MHz 2k 1/8 16QAM 1/5 Class1 no iFEC 20dB 5MHz 2k 1/8 16QAM 2/9 Class1 no iFEC 5MHz 2k 1/8 16QAM 1/4 Class1 no iFEC 18dB 5MHz 2k 1/8 16QAM 2/7 Class1 no iFEC 5MHz 2k 1/8 16QAM 1/3 Class1 no iFEC 16dB 5MHz 2k 1/8 16QAM 2/5 Class1 no iFEC 5MHz 2k 1/8 16QAM 1/2 Class1 no iFEC 5MHz 2k 1/8 16QAM 2/3 Class1 no iFEC 14dB 12dB 11,8dB 10dB 8dB 8,1dB 6dB 6,3dB 4dB 4,6dB 3,8dB 2,5dB 2dB 2,9dB 1,5dB 0dB -2dB -4dB 1Hz 10Hz 100Hz 1000Hz Figure A13.54: 16 QAM AWGN and terrestrial performances A13 1.1.3 Impact of different parameters In this clause the impact of the different parameters listed in clause A13.1.1.1 is presented in graphical format. Different bandwidth impact ETSI 452 ETSI TS 102 584 V1.1.2 (2010-XY) The chipset used for the laboratory tests supports 5, 6, 7 and 8 MHz bandwidth. Expected performance dependency on the OFDM signal bandwidth is the following: • No difference on the required AWGN C/N for FER=5%. • No significant difference for the required medium-speed C/N (time interleaving depth in Class1 is reduced but should be still sufficient for medium speed). • Small penalty on the 10 Hz required C/N in Class1 due to time interleaving duration reduction. • Linear increase of the maximum supported Doppler frequency with the signal bandwidth. Nevertheless, performances of SHA2 are not in line with these expectations. Due to throughput turbo-decoder limitations, 7 MHz and 8 MHz OFDM bandwidth configurations underperform compared to 5 and 6 MHz. Observed degradations are variable and depend on the actual bit rate: higher is the bit rate, higher is the measured degradation This finding is not related to the standard itself but rather on the specific ASIC implementation. DiBcom SHA2 Terrestrial Performances AWGN and TU6 @FER5% 20dB 18dB 16dB 14dB 12dB 10dB 9,9dB 8,9dB 8dB 8,3dB 8,1dB 6dB 5,5dB 4,9dB 4dB 4,6dB 4,9dB 5MHz 2k 1/8 16QAM 1/3 Class1 no iFEC 6MHz 2k 1/8 16QAM 1/3 Class1 no iFEC 2dB 7MHz 2k 1/8 16QAM 1/3 Class1 no iFEC 8MHz 2k 1/8 16QAM 1/3 Class1 no iFEC 0dB 5MHz 2k 1/8 16QAM 1/2 Class1 no iFEC 6MHz 2k 1/8 16QAM 1/2 Class1 no iFEC 7MHz 2k 1/8 16QAM 1/2 Class1 no iFEC -2dB 8MHz 2k 1/8 16QAM 1/2 Class1 no iFEC -4dB 1Hz 10Hz 100Hz 1000Hz Figure A13.55: Bandwidth influence Different FFT sizes Expected performance dependency on the OFDM FFT size variation is the following: • No difference on the required AWGN C/N for FER=5%. • No difference on the required TU6 C/N for FER=5%. • Linear increase of the maximum supported Doppler frequency with the OFDM FFT size. Only one modulation/coding case has been studied; indeed the maximum Doppler frequency for different cases can be derived applying the simple rule: Fdmax2K = 2 x Fdmax4K = 4 x Fdmax8K As it can be seen in Figure A13.56 , performance expectations are fulfilled by tested receiver. ETSI 453 ETSI TS 102 584 V1.1.2 (2010-XY) DiBcom SHA2 Terrestrial Performances AWGN and TU6 @FER5% 20dB 18dB 16dB 14dB 12dB 10dB 8dB 6dB 4dB 2dB 1,9dB 5MHz 2k 1/8 QPSK 1/2 Class1 no iFEC 1,9dB 1,8dB 5MHz 4k 1/8 QPSK 1/2 Class1 no iFEC 5MHz 8k 1/8 QPSK 1/2 Class1 no iFEC 0dB -2dB -4dB 1Hz 10Hz 100Hz 1000Hz Figure A13.56 : Impact of FFT size Impact of different Guard Interval Expected behaviour for guard interval size variation is the following: • No difference on the required AWGN C/N for FER=5%. • No difference on the required TU6 C/N for FER=5%. • Increase of Doppler-bandwidth when reducing the guard interval size. As it can be seen in Figure A13.57, performance expectations are fulfilled by tested receiver. AWGN and TU6 Characteristics for QPSK TC1/2 class1 Time Interleaver Doppler Frequency (Speed) vs C/N @ FER5 in TU6 for various GI 26dB 24dB QPSK TC 1/2 GI 1/8 22dB QPSK TC 1/2 GI 1/4 20dB QPSK TC 1/2 GI 1/16 18dB QPSK TC 1/2 GI 1/32 16dB C/N @ FER5 14dB 12dB 10dB 8dB 6dB 4dB 2dB 0dB -2dB -4dB 1 10 100 1000 Doppler Frequency (Hz) Figure A13.57: Guard interval size influence According to these results, the GI value has no impact on the C/N performances at any speed, except when very close to the maximum Doppler at the C/N + 3 dB point. This result legitimates the use of GI 1/8 instead of ¼ in all measurements. Impact of MPE-IFEC length Different B+S values have been tested in TU6 at 110Hz with 16QAM 1/3. ETSI 454 ETSI TS 102 584 V1.1.2 (2010-XY) C/N (dB) vs B+S for 16 QAM 1/3 iFEC 30% (C=142) in TU6 for SHA2 9,0 C/N vs B+S at medium speed C/N vs B+S at low speed (Doppler frequency = 10Hz) 8,5 Without MPE-iFEC at low speed (Fd=10Hz) Without MPE-iFEC at medium speed (Fd=250Hz) 8,0 C/N (dB) 7,5  7,0  6,5 6,0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 B + S (s) Figure A13.58 : Impact of B+S in MPE-IFEC The results summarised in Figure A13.58 confirm that: • The performances are improving when the B+S value increases . • The gain of performance is about 0,5dB for the low speed test mode (Fd is about 10Hz) and about 0,2dB for medium speed (Fd is about 250 Hz) for B+S varying from 3 s to 10 s. • At low speed, the minimum C/N is obtained for the highest B+S value (26 s). B+S at about 8 s presents good performance as well with a C/N required just 0,4 dB higher than for B+S=26 s. • At medium speed, the performances are not really improved when B+S varies (0,2 dB). • Concerning tests without MPE-iFEC, the gain brought by the MPE-iFEC at least is about 1,1dB at 10 Hz and about 0,3 dB at 250Hz. • The gain brought by B+S=26 s is about 0,3dB at low speed and about 0,1 dB at medium speed. This gain can be considered as insufficient compared to the length of the iFEC interleaver. A shorter B+S value (for example 8 s) is therefore considered sufficient for the terrestrial mode although when shadowing is present on top of TU6 fading a longer interleaver size may be beneficial. Impact of long interleaving in TU6 This clause reports the C/N performances in TU6 with a uniform long interleaver versus speed; sensitivity analysis to C/N is shown in Clause 7.3.1.2.2. It is observed that as expected the long interleaving impact is to uniform the required C/N for achieving FER=5 % versus the maximum Doppler-frequency. This is because the long interleaver breaks the fading correlation at low user speed (moderate to small fading bandwidth) reducing the corresponding loss observed in its absence. ETSI 455 ETSI TS 102 584 V1.1.2 (2010-XY) AWGN and TU6 Characteristics for class2 Time Interleaver Doppler Frequency (Speed) vs C/N @ FER5 in TU6 25dB WF01 QPSK 1/5 20dB WF05 QPSK 1/3 WF08 QPSK 2/3 WF11 16QAM 1/4 15dB WF13 16QAM 1/3 C/N @ FER5 10dB 5dB 0dB -5dB 1 10 100 1000 Doppler Frequency (Hz) Figure A13.59: Impact of long duration interleaving Impact of interleaving length in class 2 Using the same demodulator chipset, various measurements have been performed for 5 MHz signal bandwidth, 2k FFT and GI =1/8; both QPSK 1/3 and 16QAM 1/3 are considered. The channel model is TU6 at 10 Hz (5 km/h) The different configurations used during the tests are described in Table A13.27 and C/N @ FER 5% performances reported in Figure A13.60 Table A13.27: Configurations summary Duration Exact duration Configuration Receiver memory 9sec 8641ms 8 0 1 0 57 247 MBits 5sec 5003ms 6 0 1 0 44 143 MBits 2,5sec 2501ms 6 0 1 0 22 72 MBits 16QAM 1/3 1sec 985ms 4 0 1 0 13 28 MBits 0,5sec 493ms 2 0 1 0 13 14 MBits Class1 152ms 8 48 1 0 0 4,3 MBits 9sec 9551ms 4 0 1 0 63 137 MBits 5sec 5003ms 3 0 1 0 44 72 MBits 2,5sec 2501ms 3 0 1 0 22 28 MBits QPSK 1/3 1sec 985ms 2 0 1 0 13 14 MBits 0,5sec 493ms 1 0 1 0 13 7 MBits Class1 303ms 8 48 1 0 0 4,3 MBits ETSI 456 ETSI TS 102 584 V1.1.2 (2010-XY) Impact of Interleaving length for Class 2 TU 6 10Hz 10 dB C/N (dB) @ FER 5% 8 dB 6 dB 16QAM 1/3 4 dB QPSK 1/3 2 dB 0 dB 0s 2s 4s 6s 8s 10s Time (s) Figure A13.60 : Interleaving Length Impact Main outcomes are the following: • The overall long interleaver gain is 1.2 dB for QPSK and 2.5 dB for 16QAM. • The minimum C/N value of 1.8 dB for QPSK is already reached for a 2.5 s interleaver length. • For 16QAM at 2,5 and 5s, the required C/N is only 0.3 dB above the minimum value of 6.6 dB measured for an interleaver length of 9 s. • For both mod-codes a 2,5 s long interleaver seems an optimal value although when shadowing is present on top of TU6 fading a longer interleaver size may be beneficial. Note : the used configurations are not listed in Clause A.10. This is because the measurements reported in this clause had the main scope to show the trends when suing different interleaving depths. The results are in line with other simulations results (not presented here) using the Clause A.10 configurations. A13.1.1.4 Conclusions The different measurements presented above show that, with a single exception, the different simulations are consistent with the measured performances during laboratory tests with real chipsets. The introduction of a long interleaving at physical or link layer provides a real improvement at low mobile speed. Depending on the used configuration (Uniform Late and Long Uniform), the observed gain in TU6 channel is between 1 and 2 dB. At medium and higher speed, this gain is reduced at around 0,5 dB. In the presence of shadowing on top of TU6 fading the interleaver gain may be higher. The measurements show that the interleaving length could be reduced in TU6 to 1 or 2 s, but at the price of losing signal under bridges or small tunnels. The impact of fading bandwidth on the (user speed) has been analysed in detail showing the good system resilience to Doppler fading bandwidth. For S-band and typical OFDM parameters the maximum mobile speed supported by commercial demodulators exceeds the standard technical requirements. A.13.1.2 Satellite channels measurements results A13.1.2.1 Measurements with receiver 1 A.13.1.2.1.1 Configurations For the satellite channels, different configurations have been tested: • Class 1 with “standard” interleaving length, that is to say 132 ms for 16QAM and 265 ms for QPSK. • Class 1 with MPE-IFEC at 70 % code rate. • Class 2 with 9 s interleaving length. ETSI 457 ETSI TS 102 584 V1.1.2 (2010-XY) The different propagation channels are the following (see clause A.7 for details): • Sub-Urban (SU) at 50 km/h • Intermediate Tree Shadowing (ITS) at 50 km/h The different configurations are summarised in next table, with maximum burst duration is 200 ms. Table A13.28 : LMS measurements configurations BW Constellation Code FFT GI Class 1 Class 2 IFEC Configuration (MHz) rate size Interleaving Configuration 132/265 ms) (9s) 5 QPSK 1/5 2k 1/8 8/48/1/0/0 4/0/1/0/59 5 QPSK 2/9 2k 1/8 8/48/1/0/0 4/0/1/0/59 5 QPSK 1/4 2k 1/8 8/48/1/0/0 4/0/1/0/59 5 QPSK 2/7 2k 1/8 8/48/1/0/0 4/0/1/0/59 B=17, S=8, D=0, C=66, R=64 5 QPSK 1/3 2k 1/8 8/48/1/0/0 4/0/1/0/59 CR=2/3, 512 5 QPSK 2/5 2k 1/8 8/48/1/0/0 4/0/1/0/59 rows, , Max Burst Duration: 5 QPSK 1/2 2k 1/8 8/48/1/0/0 4/0/1/0/59 200ms 5 QPSK 2/3 2k 1/8 8/48/1/0/0 4/0/1/0/59 5 16 QAM 1/5 2k 1/8 8/48/1/0/0 8/0/1/0/59 5 16 QAM 2/9 2k 1/8 8/48/1/0/0 8/0/1/0/59 5 16 QAM 1/4 2k 1/8 8/48/1/0/0 8/0/1/0/59 5 16 QAM 2/7 2k 1/8 8/48/1/0/0 8/0/1/0/59 B=17, S=8, D=0, C=118, R=64 5 16 QAM 1/3 2k 1/8 8/48/1/0/0 8/0/1/0/59 CR=2/3, 768 5 16 QAM 2/5 2k 1/8 8/48/1/0/0 8/0/1/0/59 rows, Max Burst Duration :160ms 5 16 QAM 1/2 2k 1/8 8/48/1/0/0 8/0/1/0/59 5 16 QAM 2/3 2k 1/8 8/48/1/0/0 8/0/1/0/59 Note : The measurements are using configurations allowed by the 256 Mbits memory. The maximum obtained time interleaver length is 9,9 s. The different results were in line with simulations results and consistent with other receiver results when using equivalent channel emulator. This gain justifies the use of alternate configurations. A13.1.2.1.2 Examples of Results We focus in this clause on the results corresponding to the configurations defined in clause A.10. • QPSK 1/3 Long Uniform • QPSK ½ Short plus IFEC • 16QAM ¼ Long Uniform • 16QAM 2/5 Short plus IFEC • 16QAM 1/3 Long Uniform • 16QAM ½ Short plus IFEC. ETSI 458 ETSI TS 102 584 V1.1.2 (2010-XY) The next group of figures provides performances in SU and ITS 50Km/h for QPSK and an overall equivalent coding rate of 1/3: • Class 2 corresponds to QPSK 1/3 Long Uniform • Class 1 corresponds to QPSK 1/3 Short • Class 1 plus IFEC corresponds to QPSK ½ plus IFEC. Same Bit Rate Analysis : QPSK Overall CR=1/3 LMS SU50 Availability 100% 90% 80% F_ESR(5)/Mi_ESR(5) 70% 60% 50% 40% 30% Class1 only 20% Class1 + MPE-iFEC Class2 10% 0% 0 2 4 6 8 10 12 14 16 C/N (dB) Figure A13.61 : QPSK 1/3 equivalent LMS SU 50 results Same Bit Rate Analysis : QPSK Overall CR=1/3 LMS ITS50 Availability 100% 90% 80% F_ESR(5)/Mi_ESR(5) 70% 60% 50% 40% 30% Class1 only Class1 + MPE-iFEC 20% Class2 10% 0% 0 2 4 6 8 10 12 14 16 C/N (dB) Figure A13.62: QPSK 1/3 equivalent LMS ITS 50 results The next figures correspond to 16QAM and an overall equivalent coding rate of ¼: ETSI 459 ETSI TS 102 584 V1.1.2 (2010-XY) • Class 2 corresponds to 16QAM 1/4 Uniform • Class 1 corresponds to 16QAM ¼ Short • Class 1 plus IFEC corresponds to 16QAM 2/5 plus IFEC Same Bit Rate Analysis : 16QAM Overall CR=1/4 LMS SU50 Availability 100% 90% 80% F_ESR(5)/Mi_ESR(5) 70% 60% 50% 40% 30% 20% Class1 only Class1 + MPE-iFEC 10% Class2 0% 0 2 4 6 8 10 12 14 16 C/N (dB) Figure A13.63 : 16QAM 1/4 equivalent LMS SU 50 results Same Bit Rate Analysis : 16QAM Overall CR=1/4 LMS ITS50 Availability 100% 90% 80% F_ESR(5)/Mi_ESR(5) 70% 60% 50% 40% 30% 20% Class1 only Class1 + MPE-iFEC 10% Class2 0% 0 2 4 6 8 10 12 14 16 18 20 C/N (dB) Figure A13.64: 16QAM 1/4 equivalent LMS ITS 50 results The last figures correspond to 16QAM and an overall equivalent coding rate of 1/3 • Class 2 corresponds to 16QAM 1/4 Uniform • Class 1 corresponds to 16QAM 1/3 Short • Class 1 plus IFEC corresponds to 16QAM 1/2 plus IFEC. ETSI 460 ETSI TS 102 584 V1.1.2 (2010-XY) Same Bit Rate Analysis : 16QAM Overall CR=1/3 LMS SU50 Availability 100% 90% 80% F_ESR(5)/Mi_ESR(5) 70% 60% 50% 40% 30% 20% Class1 only Class1 + MPE-iFEC 10% Class2 0% 4 6 8 10 12 14 16 C/N (dB) Figure A13.65: 16QAM 1/3 equivalent LMS SU 50 results Same Bit Rate Analysis : 16QAM Overall CR=1/3 LMS ITS50 Availability 100% 90% 80% F_ESR(5)/Mi_ESR(5) 70% 60% 50% 40% 30% 20% Class1 only Class1 + MPE-iFEC 10% Class2 0% 0 2 4 6 8 10 12 14 16 18 20 22 C/N (dB) Figure A13.66: 16QAM 1/3 equivalent LMS ITS 50 results A.13.1.2.1.3 Conclusions When comparing the measurement results reported above to the simulation findings in clause A.12, we have the following results for SH-A (OFDM) Class 2 (long time interleaver): • QPSK 1/3 ITS 50 km/h : in clause A.12 the required simulated C/N is 8,5 dB for ESR5(20) 90 % fulfilment without implementation losses, versus 9,5 dB measured in laboratory tests. • QPSK 1/3 SU 50 km/h : in clause A.12 the required simulated C/N is 5 dB for ESR5(20) 90 % fulfilment without implementation losses, versus 6 dB in laboratory tests. • 16QAM 1/5 50 km/h: in the IG the required simulated C/N is 10.8 dB for ESR5(20) 90 % fulfilment without implementation losses, versus 11 dB in the laboratory tests. ETSI 461 ETSI TS 102 584 V1.1.2 (2010-XY) We can therefore conclude that there is a good matching between the clause A.12 simulations results and the laboratory measurements. Implementation losses over LMS channels are 1 dB or less. In the case of SH-A Class 1 configurations with MPE-IFEC, there are some discrepancies between simulations and laboratory tests results due to the different configurations adopted. From the previous curves above, we can derive the following conclusions: • Whatever the modulation and coding, in sub-urban environment, there is a need for long interleaving scheme: at physical or at link layer. • In sub-urban channel, Class 2 gain is about 1 dB compared to Class 1+ MPE-IFEC. • In ITS channel, Class 2 performs always better (2 dB or more) than Class 1 or Class 1 plus IFEC. • IT appears that Class 1 alone has comparable performances with Class 1 + IFEC, but in pure ITS channel: if the vehicle goes under a highway bridge, Class 1 will be useless, and it is thus generally preferable to have long interleaving i.e. Class 1 + IFEC. Important note: all the previous results have been obtained with a satellite channel emulator named CE1 implementing the 3-state Perez-Fontan LMS channel model described in clause A.7 with some approximation. Different results were obtained with a channel emulator named CE2 fully compliant with clause A.7 model. The measurements have been performed with QPSK 1/3 and 16QAM 1/3. The different channel emulators are referenced CE1 and CE2. SHA2 : QPSK 1/3 CLASS II LMS SU50 Availability SHA2 : QPSK 1/3 LMS ITS50 Availability 100% 90% 100% 90% 80% F_ESR(5)/Mi_ESR(5) F_ESR(5)/Mi_ESR(5) 80% 70% 70% 60% 60% 50% 50% 40% 40% CE1 - class II 30% 30% CE1 - class II CE2 - class II 20% 20% CE2 - class II 10% 10% 0% 0 2 4 6 8 10 12 14 16 0% 0 2 4 6 8 10 12 14 16 C/N (dB) C/N (dB) Figure A13.67: Comparison of QPSK 1/3 performances in SU and ITS with CE1 and CE2 channel emulators SHA2 : 16QAM 1/3 CLASS II LMS SU50 Availability SHA2 : 16QAM 1/3 LMS ITS50 Availability 100% 100% 90% 90% 80% 80% F_ESR(5)/Mi_ESR(5) F_ESR(5)/Mi_ESR(5) 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% CE1 - class II CE1 - class II 20% 20% CE2 - class II CE2 - class II 10% 10% 0% 0% 0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16 18 20 22 C/N (dB) C/N (dB) Figure A13.68: Comparison of 16QAM 1/3 performances in SU and ITS with CE1 and CE2 channel emulators Considering all cases, the curves show between 0,8 and 1,7 dB of performance difference in favour of CE2. We can conclude that the results reported in this clause are slightly pessimistic. ETSI 462 ETSI TS 102 584 V1.1.2 (2010-XY) A.13.1.2.2 Measurements with receiver 2 A.13.1.2.2.1 Configurations The second receiver has been used for both SH-B (TDM) and SH-A (OFDM) waveforms under LMS channel conditions. Only class 2 with long time interleaving has been considered. Table A13.29: Measurements configurations for Receiver 2 BW (MHz)/ Constellation Code rate FFT GI Class 2 Configuration (9s) Roll off OFDM 5 QPSK 1/3 2k 1/4 40/0/12/4/2 5 16 QAM 1/3 2k 1/4 40/0/12/8/4 TDM 5/ 0,15 QPSK 1/3 N/A N/A 40/0/12/4/2 5/ 0,15 16 APSK 1/3 N/A N/A 63/0/12/4/2 The measurements were made in ITS and SU channels at 50 km/h during 3000 s. The channel emulator used in this case (named CE3) is closely following the LMS model described in clause A.7. A.13.1.2.2.2 Results First, the results are given for TDM Figure A13.69; TDM QPSK 1/3 results in LMS-SU 50 km/h The ESR5(20) 90 % is obtained at C/N = 3,5 dB. ETSI 463 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.70: TDM QPSK 1/3 results in LMS-ITS 50 km/h The ESR5(20) 90 % is obtained at C/N = 7,7 dB. Figure A13.71: TDM 16 APSK 1/3 results in LMS-SU 50 km/h The ESR5(20) 90 % is obtained at C/N = 9,5 dB. Figure A13.72: TDM 16 APSK 1/3 results in LMS-ITS 50 km/h ETSI 464 ETSI TS 102 584 V1.1.2 (2010-XY) The ESR5(20) 90 % is obtained at C/N = 14,5 dB. For OFDM Figure A13.73: OFDM QPSK 1/3 results in LMS-SU 50 km/h The ESR5(20) 90 % is obtained at C/N = 4 dB. Figure A13.74; OFDM QPSK 1/3 results in LMS-ITS 50 km/h The ESR5(20) 90 % is obtained at C/N = 7,8 dB ETSI 465 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.75: OFDM 16 QAM 1/3 results in LMS-SU 50 km/h The ESR5(20) 90 % is obtained at C/N = 8,7 dB Figure A13.76: OFDM 16 QAM 1/3 results in LMS-ITS 50 km/h The ESR5(20) 90 % is obtained at C/N = 13 dB. A.13.1.2.2.3 Conclusions on receiver 2 The receiver 2 allows a comparison between TDM and OFDM, as the measurements have used the same channel emulator. Table A13.30: Summary of required C/N (dB) @ ESR5(20) 90 % with receiver 2 Receiver/Channel TDM OFDM Emulator Modulation/coding SU-50 ITS-50 SU-50 ITS-50 QPSK 1/3 3,5 7,8 4 7,9 16QAM or 16 APSK 9,5 14,5 8,7 13 1/3 ETSI 466 ETSI TS 102 584 V1.1.2 (2010-XY) If we compare TDM and OFDM, tested with the same channel emulator: • QPSK 1/3 in both TDM and OFDM waveforms have practically the same performances in SU and ITS • 16QAM 1/3and 16 APSK show some differences : OFDM performs better, especially in ITS (1,5 dB) but the interleaver configurations are not the same. A.13.1.2.3 Conclusions The following table gives a comparison in OFDM of the different receivers with different channel emulators. Table A13.31: Comparison between different receivers and channel emulators in LMS and OFDM waveform for ESR5(20) 90 % fulfilment Receiver/Channel Receiver 1/CE1 Receiver 1/ CE2 Receiver 2/CE3 Emulator Modulation/coding SU-50 ITS-50 SU-50 ITS-50 SU-50 ITS-50 QPSK 1/3 6 9,5 4,3 8,7 4 7,9 16QAM 1/3 10,5 14 9,7 13,2 8,7 13 Note : CE is for Channel Emulator. 1,2,3 refers to the type of Channel Emulator Observing the measurement results summarized in the above table we can derive the following conclusions: • With QPSK 1/3 the differences between Receiver 1/CE1 and Receiver 2/CE3 are of 2 dB in SU and 1,6 dB in ITS • These differences reduces to 0,3 dB for SU and 1,2 dB in ITS when using Receiver 1/CE2 (which tends to prove that CE2 and CE3 are based on the same LMS channel model. • With 16QAM 1/3, the differences between Receiver 1/CE1 and Receiver 2/CE3 are of 1,8 dB in SU and 1 dB in ITS. • These differences are reduced to 1 dB in SU, and 0,3 dB in ITS when using Receiver 1/CE2 (which tends to prove that CE2 and CE3 are based on the same LMS channel model . Considering that CE2 and CE3 are based on same LMS channel model, the two receivers perform approximately the same with OFDM in LMS channels (around 1 dB maximum difference). The difference is likely to be related to the different channel estimation algorithms implemented. Receiver 2 with CE3 show quite comparable performances in TDM and OFDM for LMS channels when QPSK modulation is used. For 16APSK/16QAM modulation the results are slightly better for OFDM compared to TDM. Concerning comparison with simulations results from clause A.12 with reference receiver channel estimation, we have the following table. Table A13.32: Results comparison with simulations results Receiver/Channel Receiver 1 CE2 Receiver 2 CE3 Simulations with channel Emulator estimation Modulation/coding SU-50 ITS-50 SU-50 ITS-50 SU-50 ITS-50 QPSK 1/3 OFDM 4,3 8,7 4 7,9 3,7 9,2 QPSK 1/3 TDM N/A N/A 3,5 7,8 3,7 8 Conclusions are as follows: • OFDM Receiver 1 with CE2 is in line with the simulations ETSI 467 ETSI TS 102 584 V1.1.2 (2010-XY) • OFDM, Receiver 2 is in line for SU channel, but better with ITS channel simulations. The difference is likely to be related to the faster reacquisition time of the Receiver 2 then what modelled in the simulations. • In TDM, receiver 2 is still slightly better than the simulations. The difference is likely to be related to the faster reacquisition time of the Receiver 2 then what modelled in the simulations. A.13.1.3 Hybrid channels A13.1.3.1 Configurations The Hybrid Channel tests aim to address hybrid situations when the on-air signal is a combination of terrestrial and satellite signals. This situation is particularly interesting when the satellite and the terrestrial signals are both using OFDM at the same frequency. This mode is called DVB-SHA SFN mode in the standard. Hybrid SFN channel is defined by the combination at the receiver of: • One satellite link under ITS or SU conditions; • One or several terrestrial links under TU6 conditions. In the case of Hybrid SFN, the two signals are at the same frequency, using the same modulation/coding/interleaving schemes, and carrying exactly the same data. In this section, only SHA-SFN configurations have been retained from simulations and laboratory tests: • OFDM over satellite and terrestrial at same frequency bands; • A single terrestrial repeater is emulated. The objectives of the measurements are: • Assess the improvement of system coverage beyond the edge of the terrestrial cell coverage where the received C/N is below the required threshold; • Assess the possible existence of interference when satellite and terrestrial signals are not within the GI window and the impact of these interferences; • Assess the influence of the GI on the interference reduction when the interferences are existing. Taking into account the above objectives it will not be necessary to test the following “weak” hybrid situations: c) A very high terrestrial C/N (for instance 20 dB): this terrestrial signal will be very “dominant” and no satellite component impact will be observed. d) A very low terrestrial C/N as also no terrestrial component impact will be observed being the satellite signal dominant. The different tests conditions are the following:  Waveform baseline configurations:  5 MHz/2k/GI 1/8 or 1/4  Long Interleaving : Class 2 and Class 1 plus IFEC  Satellite channels configurations:  ITS 50 km/h  SU 50 km/h  Terrestrial network configurations:  2 kW EIRP single repeater  Different repeater heights:30, 45 and 60 m  TU 6 channel  Use of Costa Hata model in SU and Rural conditions  The Receiver is a car receiver, single antenna with satellite oriented receiver (circular polarized antenna) as most of the satellite reception area will be on roads and highways, and is located at the point corresponding to the terrestrial C/N received value The operating mode for the simulations is the following: • Based on the terminal location and repeater characteristics, terrestrial C/N is settled at the wanted value, which corresponds to a certain differential delay • The satellite received C/N is settled at the required value to comply with the ESR5(20) fulfilment ratio of 90 %. • Then compute the ESR5 with the CGC repeater on. • Observe if there is a performance improvement or degradation. The basic principles of the laboratory tests are the following • Take the reference curve in LMS (ITS or SU) satellite reception with Class 2 and/or IFEC. ETSI 468 ETSI TS 102 584 V1.1.2 (2010-XY) • Set the terrestrial C/N to the wanted value, which corresponds to a certain differential delay. • Perform the hybrid configuration. tests for different satellite received C/N • Compare the ESR5 fulfilment curve with the LMS reference curve mentioned before. • Observe if there is a performance improvement or degradation. As said above, the tests will be performed with class 2 and class 1 plus IFEC receiver configurations. In class 2, only QPSK 1/3 and 16QAM 1/3 configurations have been tested. For Class 1 plus IFEC, the following configuration has been tested: 16QAM ½ at physical layer plus MPE-IFEC 70 % with GI = 1/8. In a hybrid channel, there are three main parameters to consider: • The received terrestrial C/N, • The received satellite C/N, • The difference of time of arrival of satellite and terrestrial signals, called here Delta T (T). In the test, the satellite signal C/N is varying for each couple: C/N terrestrial/Delta T corresponding to mobile position. The different couples are provided here below for each repeater configuration. To limit the number of cases, terrestrial C/N is comprised between -10 dB and + 10 dB (5 values) The following tables have been used for Class 2. Table A13.33: Configurations for 30 m Repeater: 30 m C/N (dB) Delta T (µs) To GI=1/8 To GI =1/4 Distance/Rep terrestrial (44.8µs) (89.6µs) (km) Sub-Urban Terr/ -10 57,07 127 % 64 % 9,4 LMS-SU 50 -5,1 40,82 91 % 46 % 6,8 0 28,35 63 % 32 % 4,9 4,3 19,79 44 % 22 % 3,7 10,4 12,3 27 % 14 % 2,5 Rural- -9,6 107 239 % 119 % 17,5 Terres/LMS-ITS 50 -5,2 80,26 179 % 90 % 13,1 0 57,07 127 % 64 % 9,4 4,9 40,82 91 % 46 % 6,8 10 6,99 16 % 8% 4,9 Table A13.34: Configurations for 45 m Repeater: 45 m C/N (dB) Delta T (µs) To GI=1/8 To GI =1/4 Distance/Re terrestrial (44.8µs) (89.6µs) p (km) Sub-Urban Terr/ -10,2 73,60 164% 82% 12 LMS-SU 50 -4,7 48,24 108% 54% 8,3 0,6 34,49 77% 39% 5,8 4,6 25,38 57% 29% 4,4 9,7 17,23 38% 19% 3,4 ETSI 469 ETSI TS 102 584 V1.1.2 (2010-XY) Rural- -9,8 140,57 314% 157% 17,5 Terres/LMS-ITS 50 -4,8 100,30 224% 112% 15,9 -0,2 73,60 164% 82% 12 5,3 48,24 108% 54% 8,3 9,4 25,38 57% 29% 4,4 Table A13.35: Configurations for 60 m Repeater : 60 m C/N (dB) Delta T (µs) To GI=1/8 To GI =1/4 Distance/Rep terrestrial (44.8µs) (89.6µs) (km) Sub-Urban Terr/ -7,6 73,60 164% 82% 12 LMS-SU 50 -2,2 48,24 108% 54% 8,3 3 34,49 77% 39% 5,8 6,8 25,38 57% 29% 4,4 10,8 17,23 38% 19% 3,4 Rural- - 6,9 140,57 314% 157% 17,5 Terres/LMS-ITS 50 -2 100,30 224% 112% 15,9 2,4 73,60 164% 82% 12 7,8 48,24 108% 54% 8,3 11,8 25,38 57% 29% 4,4 For Class 1 plus IFEC, different configurations have been tested. They are based on the DVB-SSSP Validation Task Force. In this configuration, the delta T is varying from 0% of the GT to 200 % by steps of 50 %, and the terrestrial C/N has the following values in dB: -10, - 6, -3, 0, 3, 6, 10. The satellite received C/N in line of sight is assumed to be 12 dB and is fixed. A13.1.3.2 Results A subset of the results contained in this clause have already been presented in the clause 7. The following graphs represent the required C/N in satellite channel when in presence of terrestrial channel to get 90 % ESR5(20). All possible repeater heights have been considered. The horizontal line represents the C/N required in LMS channel only. There are two cases: one with GI 1/8 and the second one with GI 1/4 . We consider first the cases with SHA-SFN Class 2 terminals. A13.1.3.2.1 Suburban channels, Two cases are provided as examples: a) QPSK 1/3 at 50 km/h; b) 16QAM 1/3 at 50 km/h. The satellite channel emulator is set in the SU configuration and the speed is 50 km/h. The terrestrial channel emulator is set at TU6 50 km/h, with the required C/N value. For a 30 m terrestrial repeater height, the following results have been obtained. ETSI 470 ETSI TS 102 584 V1.1.2 (2010-XY) SHA2 : QPSK 1/3 GI=1/8 Hybrid LMS SU50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% Satellite Only 40% Terrestrial C/N=-10dB 30% Terrestrial C/N=-5dB 20% Terrestrial C/N=0dB 10% 0% 3 5 7 9 11 C/N (dB) Figure A13.77: Performances in sub Urban Areas with QPSK rate 1/3 @ 30 m As the curve corresponding to C/N = 0 dB already shows a constant ESR5(20) of 100 %, the two other cases corresponding to C/N equal to 5 and 10 dB are not represented. SHA2 : 16QAM 1/3 GI=1/8 Hybrid LMS SU50 Availability 100% 90% 80% 70% 60% F_ESR(5) Satellite Only 50% 40% Terrestrial C/N=-10dB 30% Terrestrial C/N=-5dB 20% Terrestrial C/N=0dB 10% 0% 4 6 8 10 12 14 16 C/N (dB) Figure A13.78: Performances in sub Urban Areas with 16QAM rate 1/3 @ 30 m From this example derive the following conclusions can be derived: ETSI 471 ETSI TS 102 584 V1.1.2 (2010-XY) • Most of the hybrid reception cases provide improvement compared to the satellite only reception as the required C/N is decreasing • When the terrestrial signal is under the C/N threshold (for instance C/N = 0 dB in QPSK while 2 dB are required), the resulting quality of service with satellite on (and even with very low satellite C/N values) is above 99 % ESR5; • With 16QAM, even the case C/N = 0 dB provides improvement to the satellite channel. With 45 m terrestrial repeater height, the following results have been obtained. SHA2 : QPSK 1/3 GI=1/8 Hybrid LMS SU50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% Satellite Only 40% Terrestrial C/N=-10dB 30% Terrestrial C/N=-5dB 20% Terrestrial C/N=0dB 10% 0% 3 5 7 9 11 13 15 C/N (dB) Figure A13.79: Performances in sub Urban Areas with QPSK rate 1/3 @ 45 m ETSI 472 ETSI TS 102 584 V1.1.2 (2010-XY) SHA2 : 16QAM 1/3 GI=1/8 Hybrid LMS SU50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% Satellite Only Terrestrial C/N=-10dB 40% Terrestrial C/N=-5dB 30% Terrestrial C/N=0dB 20% Terrestrial C/N=5dB 10% Terrestrial C/N=10dB 0% 4 6 8 10 12 14 16 C/N (dB) Figure A13.80: Performances in sub Urban Areas with 16QAM rate 1/3 @ 45 m From this example we can derive the following conclusions • Most of the hybrid reception cases provide improvement compared to the satellite only reception as the required C/N is decreasing. • When the terrestrial signal is under the C/N threshold (for instance C/N = 0 dB in QPSK while 2 dB are required), the resulting quality of service with satellite on (and even with very low satellite C/N values) is above 99 % ESR5(20); • In QPSK, the case with terrestrial C/N=-10 dB, and with differential delay equal to 164 % of the GI, shows some slight degradation in the satellite reception (< 0.5 dB). • With GI of ¼, there is no more degradation with QPSK 1/3. ETSI 473 ETSI TS 102 584 V1.1.2 (2010-XY) SHA2 : QPSK 1/3 GI=1/4 Hybrid LMS SU50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% Satellite Only 40% 30% 20% Terrestrial C/N=-10dB 10% 0% 3 5 7 9 11 13 15 C/N (dB) Figure A13.81: Performances in sub Urban Areas with QPSK rate 1/3 @ 45 m and GI 1/4 Finally with 60 m height terrestrial repeater height, the following results have been obtained. SHA2 : QPSK 1/3 GI=1/8 Hybrid LMS SU50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% Satellite Only 40% Terrestrial C/N=-7dB 30% Terrestrial C/N=-2dB 20% Terrestrial C/N=2dB 10% 0% 3 5 7 9 11 13 15 C/N (dB) Figure A13.82: Performances in sub Urban Areas with QPSK rate 1/3 @ 60 m ETSI 474 ETSI TS 102 584 V1.1.2 (2010-XY) SHA2 : 16QAM 1/3 GI=1/8 Hybrid LMS SU50 Availability 100% 90% 80% 70% 60% Satellite Only F_ESR(5) 50% Terrestrial C/N=-7dB Terrestrial C/N=-2dB 40% Terrestrial C/N=2dB 30% Terrestrial C/N=7dB 20% Terrestrial C/N=11dB 10% 0% 4 6 8 10 12 14 16 C/N (dB) Figure A13.83: Performances in Sub Urban Areas with 16QAM rate 1/3 @ 60 m With 60 m height, there are some degradations for QPSK when differential delay is over 164 % of the GI (C/N = -7 dB) A13.1.3.2.2 Rural Channel, For the rural channel configuration two cases are provided: a) QPSK 1/3 at 50 km/h b) 16QAM 1/3 at 50 km/h The satellite channel emulator is set in the ITS configuration and the speed is 50 km/h. The terrestrial channel emulator is set at in the TU6 configuration at 50 km/h. The same three cases for the different repeater antenna height are provided. For a 30 m terrestrial repeater height, the following results have been obtained: • There is noticeable improvements when terrestrial C/N is over 0 dB; • With larger differential delays (more than 200 % of the GI), there are some degradations for both QPSK and 16QAM , below 0,5 dB. ETSI 475 ETSI TS 102 584 V1.1.2 (2010-XY) SHA2 : QPSK 1/3 GI=1/8 Hybrid LMS ITS50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% 40% Satellite Only 30% Terrestrial C/N=-10dB 20% Terrestrial C/N=-5dB 10% Terrestrial C/N=0dB 0% 4 6 8 10 12 14 16 C/N (dB) Figure A13.84: Performances in Rural Areas with QPSK rate 1/3 @ 30 m SHA2 : 16QAM 1/3 GI=1/8 Hybrid LMS ITS50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% 40% Satellite Only Terrestrial C/N=-10dB 30% Terrestrial C/N=-5dB 20% Terrestrial C/N=0dB 10% 0% 4 6 8 10 12 14 16 18 20 22 C/N (dB) Figure A13.85: Performances in Rural Areas with 16QAM rate 1/3 @ 30 m With 45 m terrestrial repeater height, the following results have been obtained: • In QPSK, there are some degradation when differential delay is over 200 % of the GI; • For the other cases there is still high improvement of the coverage; • In 16QAM, there is a slight degradation when differential delay are over 200 % of the GI, but below 0,5 dB. • The use of GI ¼ provides improvement with QPSK 1/3, reducing the interference ETSI 476 ETSI TS 102 584 V1.1.2 (2010-XY) SHA2 : QPSK 1/3 GI=1/8 Hybrid LMS ITS50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% Satellite Only 40% Terrestrial C/N=-10dB 30% Terrestrial C/N=-5dB 20% Terrestrial C/N=0dB 10% 0% 4 6 8 10 12 14 16 C/N (dB) Figure A13.86: Performances in Rural Areas with QPSK rate 1/3 @ 45 m SHA2 : 16QAM 1/3 GI=1/8 Hybrid LMS ITS50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% 40% Satellite Only Terrestrial C/N=-10dB 30% Terrestrial C/N=-5dB Terrestrial C/N=0dB Terrestrial C/N=5dB 20% Terrestrial C/N=10dB 10% 0% 4 6 8 10 12 14 16 18 20 22 C/N (dB) Figure A13.87: Performances in Rural Areas with 16QAM rate 1/3 @ 45 m ETSI 477 ETSI TS 102 584 V1.1.2 (2010-XY) SHA2 : QPSK 1/3 GI=1/4 Hybrid LMS ITS50 Availability 100% 90% 80% 70% Satellite Only F_ESR(5) 60% 50% Terrestrial C/N=-10dB 40% Terrestrial C/N=-5dB 30% 20% 10% 0% 4 6 8 10 12 14 16 C/N (dB) Figure A13.88: Performances in Rural Areas with QPSK rate 1/3 @ 45 m and GI 1/4 Finally with 60 m terrestrial repeater height, the following results have been obtained: • For both modulation schemes, when differential delay is above 200 % of the GI, there are noticeable degradations that can reach more than 2 dB • To mitigate these interference one can use GI ¼. SHA2 : QPSK 1/3 GI=1/8 Hybrid LMS ITS50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% 40% Satellite Only 30% Terrestrial C/N=-7dB 20% Terrestrial C/N=-2dB 10% Terrestrial C/N=2dB 0% 4 6 8 10 12 14 16 C/N (dB) Figure A13.89: Performances in Rural Areas with QPSK rate 1/3 @ 60 m ETSI 478 ETSI TS 102 584 V1.1.2 (2010-XY) SHA2 : 16QAM 1/3 GI=1/8 Hybrid LMS ITS50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% 40% Satellite Only Terrestrial C/N=-7dB 30% Terrestrial C/N=-2dB Terrestrial C/N=2dB 20% Terrestrial C/N=8dB Terrestrial C/N=12dB 10% 0% 4 6 8 10 12 14 16 18 20 22 C/N (dB) Figure A13.90: Performances in Rural Areas with 16QAM rate 1/3 @ 60 m SHA2 : QPSK 1/3 GI=1/4 Hybrid LMS ITS50 Availability 100% 90% 80% 70% 60% F_ESR(5) 50% 40% Satellite Only 30% Terrestrial C/N=-7dB 20% Terrestrial C/N=-2dB 10% 0% 4 6 8 10 12 14 16 C/N (dB) Figure A13.91: Performances in Rural Areas with QPSK rate 1/3 @ 60 m and GI ¼ A13.1.3.2.3 Case of Class 1 plus MPE-IFEC In the case of MPE-IFEC, the following configuration has been tested: 16QAM ½ at physical layer plus MPE-IFEC 70 % with GI 1/8. The results show some degradation for 60 m repeater height but improvement in all other cases. As said above, the case of ¼ with 16QAM 1/3 was not tested, but as degradation with 1/8 was lower for 16QAM than for QPSK, with GI ¼, there is no doubt that there will be no more degradation. With MPE-IFEC,the same results are ETSI 479 ETSI TS 102 584 V1.1.2 (2010-XY) provided with GI 1/8. For the case of MPE-IFEC, not all the cases have been measured, due to lack of time. The representative cases are summarized in the following table, which represents the ESR5(20) fulfilment ratio (last three lines). The satellite C/N is 12 dB which provides 90 % ESR5(20) in the sub urban 50 km/h configuration. Table A13.36: Summary of results for MPE-IFEC MPE-IFEC GI ratio C/N terrestrial 0% 50% 100% 150% 200% -6 dB 90% 90% 87% 68% 62% 0 dB 92% 92% 91% 51% 8% 6 dB 100% 100% 100% 99% 94% From the table above we can derive the following conclusions: • When the terrestrial signal is strong (C/N = 6 dB for instance) and whatever the differential delay, the improvement is always excellent: 100 % ESR5 when differential delay is up to 100 % of the GI. When differential delay exceeds 100 % of the GI, some improvement is also noticeable . • When terrestrial signal C/N is around 0 dB and with and differential delay below 100 % of the GI, the results are good. The performance is degraded when the differential delay increases, but we observe ESR5 fulfilment reduction to 8% corresponding to only 2 dB C/N degradation due to the sharp roll-off of the ESR5 curve. • When the terrestrial signal is weak, (C/N around – 6 dB), there is only a slight degradation when differential delay is equal to 100 % of the GI (around 0.1 dB), but some degradation for larger delays, quite low in terms of C/N, even if the ESR (5) value varies quite a lot. A13.1.3.3 Interpretation and Conclusions Summarizing some of different laboratory measurements, the following figure, illustrates the different hybrid network operational zones. Figure A13.92: Illustration of the different zones in rural areas with 16 QAM rate 1/3 ETSI 480 ETSI TS 102 584 V1.1.2 (2010-XY) It is possible to distinguish the different zones (starting from right in the diagram): • When received terrestrial signal C/N is above required threshold, we can consider that we are under terrestrial coverage, and the satellite signals provides some improvement, but is not essential. • The second zone corresponds to a zone where satellite signal is improved by the terrestrial one. Even if the satellite signal is quite weak (for instance if received by a handset, the combination of the two signals result in a good quality signal. • The third zone corresponds to an improvement of the satellite signal with a “residual” edge of terrestrial coverage signal. • The fourth zone corresponds to possible interference zone between the satellite and terrestrial signals. The border between the two previous zones corresponds to a differential delay of around 160 % of the GI. • The last zone corresponds to satellite only signal as terrestrial signal is falling below -10 dB, but it can have some interference impact. In the case of suburban area, the interference zone is reduced as shown in the following figure. 16QAM : Hybrid Channel Suburbain 50 km/h Performances Overview 12 dB   10 dB        C/N for 90% availability  8 dB 6 dB 30m high tower 45m high tower 4 dB 60m high tower Satellite Only 2 dB 0 dB -10 dB -5 dB 0 dB 5 dB 10 dB C/N (Ter)    Figure A13.93: Illustration of the different zones in Sub Urban areas with 16 QAM rate 1/3 Some conclusions and rules can be derived from the previous results: • The hybrid channel architecture gives an evident gain of quality of services, as it increases in most of the cases the ESR (5) ratio. • In the northern direction, the differential delay is below 100 % of the GI on along distance, even for GI 1/8, and there is no risk of interference, as the terrestrial signal becomes very low at the edges. • On the southern direction (in sight of the satellite), the gain is noticeable, even at more than 100 % of the GI. • When the delay is more than 150 % of the GI approximately , there can be some interference at te edges of the contour, but they can be mitigated by different means: using a large GI when possible, implement timing advance at repeaters level, or reducing radiated power towards southern direction; The repeater could use two sector antenna at the edge of the zone, and oriented towards northern directions; Hence a hybrid network planning tool can be designed according to the following principles: ETSI 481 ETSI TS 102 584 V1.1.2 (2010-XY) 1. Compute the geometrical hybrid SFN zones taking into account the possible timing advance: roughly the zones where differential delay is below 150 % (TBC) of the GI. In this region we will assume that satellite and terrestrial signal components will sum up constructively. 2. Compute terrestrial only coverage deriving the region within the SFN where considering the TU6 corresponding DVB-SH C/N threshold (C/N)Ter, C/N > (C/N)Ter . 3. Compute inside the geometrical SFN zone the equivalent C/N taking into consideration the satellite LOS by using the following simplified formula:   C  [dB]     [ dB]  C   10  N Sat  N Ter  C 10   [dB] = 10 ⋅ log 10 + 10 .  N eq     In this case the Q.o.S. could be determined by using maps like in clause 11.7.4. The calculation will give inputs for deriving possible reduction of number of sites or their transmitted power. 4. Outside the geometrical SFN zone we assume that the terrestrial signal component will now interfere with the satellite one. The resulting C/N is now computed by:    C  [dB]   N     N  [dB]  N  [dB]    C  C       Sat − 10      Sat − 10 −  Ter 10   C                [dB] = −10 ⋅ log 10 + 10 .  N eq       A.13.2 Field trials results A.13.2.1 Introduction Several organizations undertook a set of trial campaigns to validate the operation of DVB-SH. The aim of these trials was to confirm on the field and under realistic conditions the performances predicted via simulations and laboratory tests. The first two examples of trials (Pau and Torino) were terrestrial oriented (as deployed before the launch of any ad hoc satellite). The main objectives were the following: • Waveform performances evaluation; • SFN gain characterisation; • Receive antenna diversity characterisation; • Validation of network engineering rules (integration on UMTS sites for instance); • Validation of the Radio Planning strategy. The third example analyses the satellite coverage (using Eutelsat W2A S-band payload) with a SH-A configuration on a vehicular environment in a long route from Paris to Barcelona. The fourth campaign (Pisa) was focused on satellite and hybrid reception and it used the S-band signal from W2A satellite payload. The main objectives were: • Evaluation of Satellite coverage in different environments (urban, sub-urban and rural); • Assessment of hybrid gain: code combining; • Comparison of class 1 versus class 2 receivers. ETSI 482 ETSI TS 102 584 V1.1.2 (2010-XY) A.13.2.2 Pau trial A.13.2.2.1 Introduction For over one year, an intensive field test campaign of the DVB-SH technology was performed over an extensive terrestrial zone covering approximately 15 Km² of urban, suburban and rural environments jointly set-up by the French mobile network operator SFR and Alcatel-Lucent in the city of Pau (France), over a typical mobile cellular networking topology (sites disposition, engineering and heights), actually having a large sharing of site infrastructure between the mobile cellular UMTS service and the DVB-SH broadcast service (typically, in-building cabling and antennas, with limited modifications). Seven UMTS sites were selected and shared with DVB-SH transmitters. The Pau trial has been carried out in the French city of Pau in 2007 (Phase 1) and 2008 (Phase 2), sharing 7 UMTS sites of the French operator. In this document, we compare the performances on the field of the DVB-SH waveform in S- band (2170-2200 MHz) under terrestrial conditions with the laboratory tests done with a TU6 channel. The basic system architecture is represented here below. Helicopte Eutelsat W3A satellite (Ku r band) DVB-S2 DVB-SH Ku UMTS « Satellite » coverage DVB- SH B Terrestrial H coverage DTR1 DTR1 SAT Uplink Regenerative Regenerative repeater Figure A13.94: Pau trial high level setup configuration A.13.2.2.2 Experimental set up Several phases of measurements were conducted overall, benefiting in steps of successive improvements brought by field tools and by measurements protocols. Alcatel-Lucent and SFR were primarily interested in the end-to-end performances results, directly representative of actual service conditions delivered to end users while using SFN at network level and integrated reception antennas diversity at terminal level. SFN measurements benefited from the capabilities of the Alcatel-Lucent S-UMTS band emitter to broadcast several carriers (up to three in totally SFN conditions). The antenna diversity of order two was evaluated thanks to an innovative pre-industrial development by SAGEM Wireless (called the “T2 program”). Precise quantitative measurements were carried out using terrestrial-only coverage (in SFN configuration and with antenna diversity at the receiver). A simulation of a hybrid coverage (combining signals from terrestrial repeaters and a satellite “emulator” provided by helicopter stationary flight) was used for qualitative results not presented here. Several reception conditions were evaluated (Vehicular, In-car, Indoor and Outdoor pedestrian). The following picture gives an example of radio network planning including the sites location inside Pau. ETSI 483 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.95: Example of Radio Network Planning in Pau A.13.2.2.2.1 Base Station Configuration The 7 test sites, spread throughout the city, were reusing existing UMTS base station antennas and feeder. A diplexer, close to the UMTS antenna feeder, allowed injecting the DVB-SH signals (see figure A13.134). Sites equipped with a TMA (The Tower Mounted Amplifier) had to be upgraded as the one in place was filtering the DVB-SH signals out. Each site was equipped with a satellite dish receiving signal from the head-end located in SES-Astra premises. On each site, up to 3 carriers were broadcasted simultaneously. The first tests aimed at validating the impact of antenna sharing on the UMTS service. No impact of DVB-SH on the UMTS service was found by SFR using its key performance indicators (e.g. call setup and completion rate). Figure A13.96: Diplexer multiplexing UMTS and DVB-SH signals A.13.2.2.2.2 Helicopter Configuration A helicopter flying in stationary position with a constant elevation angle emulated the satellite. The helicopter embedded a transparent repeater (i.e. a simple frequency transponder with amplifier) equipped with a circular ETSI 484 ETSI TS 102 584 V1.1.2 (2010-XY) polarization antenna as on an actual satellite (see Figure A13.97). The radiated power was adjusted in order to deliver on the ground a power equivalent to the one of the satellite. Figure A13.97 The helicopter transmit antenna system A.13.2.2.2.3 Receiver Configurations Three types of receiving situations were evaluated using various antennas setup (see figures below): • Vehicular reception using 2 test receiver and 2 outdoor car-mounted antennas; • In-car reception (see figure 4) using 2 test receivers and 2 handheld antenna systems; • Indoor and outdoor pedestrian reception using up to 3 test receivers and 3 handheld antenna systems installed on a cart. For outdoor pedestrian the cart was pushed by a person along the test road, each SH receiver was connected to an independent antenna system. The indoor pedestrian conditions were emulated with the cart pulled by a rope and a motor, the SH receivers were connected via splitters (1 in diversity, 2 without diversity). Figure A13.98: Different kinds of roof top antennas ETSI 485 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.99: In-car receivers . Figure A13.100: Indoor receivers and processing unit installed on a cart A.13.2.2.3 Waveform performance results We consider the two cases in outdoor and indoor. A.13.2.2.3.1 Waveform configurations In most of the cases, unless specified, the characteristics of the waveform are the following, using a frequency slot in: inside 2170-2185 MHz. Table A13.37: Pau waveform configurations BW Modulation/ FFT GI Capacity Short Short Uniform Uniform MPE-IFEC (MHZ) coding (Mbps) Interleaving Interleaving Interleaving duration (s) duration (s) structure 5 QPSK 1/3 2k 1/8 2,468 8 0,265 5 QPSK 1/3 2k 1/8 2,468 4/0/1/0/59 10 5 QPSK 1/3 2k 1/8 2,468 8 0,265 B=17 S=9 B+S = 26 5 16QAM 1/3 2k 1/8 4,494 8 0,132 ETSI 486 ETSI TS 102 584 V1.1.2 (2010-XY) A.13.2.2.3.2 Outdoor pedestrian The outdoor vehicle route and the corresponding network planning results are shown here below: éè Figure A13.101: Reference Circuits in Pau For the measurements, the results for 16QAM 1/3 and QPSK 1/3 are presented in the following figure. 5000 points were measured, filtered and declared “good”. Figure A13.102: Results of outdoor measurements: FER versus SNR In conclusion, we have the following C/N values • For QPSK 1/3 : C/N = 4 dB @ Low Speed • For 16QAM 1/3 : C/N =9,5 dB @ Low Speed The above results are slightly above to the TU6 channel laboratory test results reported in clause A13. It is recalled that at low speed and using a TU6 channel emulator, the following laboratory measurements were observed: ETSI 487 ETSI TS 102 584 V1.1.2 (2010-XY) • QPSK 1/3 : 2,8 -3,5 dB • 16QAM 1/3: 8,5-9,1 dB. A.13.2.2.3.3 Indoor pedestrian Two places were selected for indoor measurements: • Leclerc commercial center; • La Pepinière for its SFN gain. Figure A13.103: Location of indoor sites In this case only 16QAM 1/3 results are available and summarized in the following figure: Figure A13.104: Results of indoor measurements: FER versus C/N with 16QAM rate 1/3 The obtained C/N value for 16QAM 1/3 @ FER 5% is 8,5 dB , which is exactly the value obtained with the same equipment using TU6 channel emulator at 5 km/h (19 Hz). ETSI 488 ETSI TS 102 584 V1.1.2 (2010-XY) A.13.2.2.3.4 Waveform performances synthesis The different waveform performances at low speed are summarized in the table here below and compared with the results obtained with the same receiver under laboratory conditions. Table A13.38: Summary of waveform performances pedestrian Modulation/coding Terminal Terminal C/N @ FER 5 C/N @FER 5% Type Class Indoor Low speed Outdoor Low speed QPSK 1/3 field trials Class 1 Not available 4 Lab test Class 1 3,2 3,2 16QAM 1/3 field trials Class 1 8,5 9,5 Lab test Class 1 8,5 8,5 The differences between laboratory and field trial results is less than 1 dB. A.13.2.2.4 SFN measurements SFN gain is generally defined as the difference between the MFN case (i.e. each repeater uses a given frequency; the receiver selects along the time the best repeater) and the SFN case (i.e. all repeaters use the same frequency and the receiver combines the signal received from the different repeaters). Clearly, this is a purely “local” (point by point) definition. Local signals fluctuate widely within very short spatial ranges, especially in indoor conditions. When the various SFN-combining signals are very power unbalanced, the effect on “real end-user” service conditions may not be measurable under the usual “macroscopic” service availability criteria such as FER. Therefore, SFN measurements must be done in locations where SFN conditions are optimally met. This requires a specific radio network planning and network set-up optimization that, at the same time, respects propagation conditions in real networks, namely: • Cell edge with low power signal, so that SFN creates an extended coverage (e.g. one individual signal does not ensure service while cooperative SFN signals do). Sources must be roughly of the same power level to guarantee non-biased conditions. • The receiver antenna configuration must be representative of a real scenario; this was achieved using the RF part of a complete SAGEM T2 mobile terminal. The purpose of the test was to compare MFN and SFN in a real quasi-static environment using two repeaters. This was evaluated in two ways: 1) Field strength measurements (in dB):  Signal level measured in SFN (i.e. receiving simultaneously both repeaters on the same frequency),  Best signal level measured in MFN (i.e. receiving the repeater with the best signal. Both repeaters using different frequencies). 2) Transmission quality (in %):  Frame Error Rate (FER) in SFN,  Best FER measured in MFN. A.13.2.2.4.1 Methodology Two methods were used to assess the SFN gain: Method 1: simultaneous runs This method benefits from the capability of the repeater used in the trials to broadcast simultaneously up to three DVB-SH 5 MHz signals: • Repeater 1 is configured to broadcast two signals, one at frequency F1, the other one at frequency F3 , ETSI 489 ETSI TS 102 584 V1.1.2 (2010-XY) • Repeater 2 is configured to broadcast two signals, one at frequency F2, the other one at frequency F3. Three receivers are set up to receive one of the three signals. Since frequencies F1, F2 and F3 are contiguous, we assumed similar propagation channel. This methodology allows simultaneous measurement of all signals, through a single RF chain, thus eliminating any propagation hazard, and it requires three different receiver tools, duly calibrated for simultaneous reception of the three carriers, and equipped with ad hoc filters. . Figure A13.105: Repeater configuration for simultaneous run tests Tests were done on a single run on the test path; the three receivers measured parameters continuously over the test path. Method 2: Successive runs In this method, a single frequency F2 was used, from one repeater, then from the other one, then from both repeaters in SFN. A single receiver was set-up and received frequency F2. Independently of the method used, the following conditions were also respected: • For outdoor, about 100 measurement sites were selected along the roads and streets located in the test area. • Each measurement site was a few meters long and separated by about 50 meters from the next measurement site. • One minute measurements were performed in each site with the cart moving around. For indoor tests were done in a room with continuous measurements at a speed of about 1m/mn. A.13.2.2.4.2 The results Field strength measurements results SFN gain evaluation is done comparing the distribution of received signal levels. For each repeater, a CDF of the measured field strength was derived. Figure A13.106 represents the percentage of time the received level is above abscise. 77% of the time, the received signal level was higher than -100 dBm for “Centre”, -99 dBm for “Gabard” and -96.5 dBm in SFN. Measured SFN gain versus the single “Gabard” signal is 2.5 dB. ETSI 490 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.106: Example of Rx level compared distribution The next figure represents the percentage of time C/N is below abscise, 70% of the time (complementary to 30 %), the C/N is higher than 2.0 dB for “Centre”, 3.5 dB for “Gabard” and 6.5 dB in SFN case, giving 3.0 dB of SFN gain. Figure A13.107: C/N compared distribution Transmission Quality Measurement Results The three curves of FER versus time are plotted over the measurement path. The simultaneous measurements method is used. Figure A13.108: FER measurements A.13.2.2.5 Indoor diversity measurements Space diversity works on the principle that when considering two points separated by half a wavelength, the probability that these two points correspond simultaneously to a minimum is very low. Using two receive antennas allows taking advantage of the space diversity. In S-band, antennas need to be separated by only a few centimetres which makes antenna diversity a feasible and attractive solution. Several methods exist to combine the signal from both antennas. The method used in the trial is the Maximum Ratio Combining (MRC) which weights the signal amplitude according to the square-root of their SNR, before coherently adding them. A.13.2.2.5.1 Methodology As for SFN measurements, it is preferable to do measurements in the area of the network where diversity clearly creates the conditions of a measurable improved reception. To characterize the diversity gain, 3 receivers were used, all connected to the same single handheld antenna system provided by the SAGEM Mobiles T2 terminal. Figure A13.109 details how the single antenna signals are fed to the several receivers: “Receiver 1” received only signal from “Antenna 1”, “Receiver 2” received only signal from “Antenna 2”, “Receiver 3” received signal from both antennas and combined these signals using the MRC method.. Receiver 3 signals inputs could, in addition, be attenuated selectively, using wideband attenuators. ETSI 491 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.109: Receivers configuration These receivers, installed on a cart, were moved around the room and several thousand measurements were performed. At each run, the signal sent to “Receiver 3” was progressively attenuated until the quality of the video (FER) was equivalent to the one of the best receiver without diversity. A.13.2.2.5.2 Diversity Measurements The comparison is based on the global distribution of FER5 values (A FER below 5% allows to display video without image freeze). This comparison is made between the receiver with diversity (“Receiver 3”) and the best receiver without diversity (“Receiver 1” or “Receiver 2”). One run was performed per attenuation, with 3600 to 6000 measurements per run. Figure A13.110: Percentage of time with a FER below 5% The results show that the signal sent to the “Receiver 3” needs to be attenuated by 5.5 dB to get a performance similar to the best receiver without diversity (“Receiver 1” in the example of figure A13.110). Therefore the diversity gain from this experiment can be estimated to be 5,5 dB. The diversity gain obtained in outdoor urban has been measured to be in the range of 4 to 6 dB. Diversity gain appears to be highly dependant on the urban environment. A.13.2.3 Torino trial A.13.2.3.1 Introduction The objectives of this trial was to demonstrate the quality of the DVB-SH technology on the field.. The focus of this campaign was to assess the antenna diversity and the SFN gains. The capacity of DVB-SH waveform to counteract channel impairments thanks to efficient coding and interleaving schemes was also explored. ETSI 492 ETSI TS 102 584 V1.1.2 (2010-XY) Eutelsat W3A K Ku u DVB- S2 Terrestrial coverage Alcatel-Lucent DTRs Hosted in RAI & H3G Figure A13.111: Overview of the Torino field trial set up The terrestrial network is composed of a combination of high power / high height sites (up to 480 m for Eremo) and lower power (30 w) regular heights collocated with 3G sites (provided by H3G operator). A.13.2.3.2 Experimental set up The experiment is based on the use of a combination of five Low Power sites collocated with 3G sites, and two high power sites (Eremo and Cernaia) located as shown in the next figure. Figure A13.112: Torino sites location The following picture describes the block diagram of a low power site collocated with 3G site. ETSI 493 ETSI TS 102 584 V1.1.2 (2010-XY) • The antennas are shared, thanks a diplexer, as already described in the Pau trial • The satellite dish is used for receiving the different services that will be demodulated, multiplexed and DVB- SH modulated through the tri sectors of the site. • The TMA (Tower Mounted Amplifiers) are changed to increase bandwidth. Figure A13.113: Typical Low Power site configuration Figure A13.114: Low Power site: Cavour site The next site is a high power high altitude site: Eremo ETSI 494 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.115: Eremo site and Eremo configuration The different receiver configurations are depicted hereafter: Figure A13.116: Configuration of car receiver with roof top antennas ETSI 495 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.117: In car reception configuration A.13.2.3.3 Performances of DVB-SH waveform and diversity gain analysis in reception condition C A.13.2.3.3.1 Waveform configurations In most of the cases, unless specified, the characteristics of the waveform are the following, using a frequency slot inside the 2170-2185 MHz band. Table A13.39 : Torino waveform configurations BW Modulation/ FFT GI Capacity Short Short Uniform Uniform MPE-IFEC (MHZ) coding (Mbps) Interleaving Interleaving Interleaving duration (s) duration (s) structure 5 QPSK 1/3 2k 1/8 2,468 8 0,265 5 QPSK 1/3 2k 1/8 2,468 2/0/1/0/59 5 5 QPSK 1/3 2k 1/8 2,468 4/0/1/0/59 10 5 QPSK 1/3 2k 1/8 2,468 8 0,265 B=3 S=2 B+S = 5 5 QPSK 1/3 2k 1/8 2,468 8 0,265 B=17 S=9 B+S = 26 ETSI 496 ETSI TS 102 584 V1.1.2 (2010-XY) BW Modulation/ FFT GI Capacity Short Short Uniform Uniform MPE-IFEC (MHZ) coding (Mbps) Interleaving Interleaving Interleaving duration (s) duration (s) structure 5 16QAM 1/3 2k 1/8 4,494 8 0,132 5 16QAM 1/3 2k 1/8 4,494 8/0/1/0/59 10 5 16QAM 1/3 2k 1/8 4,494 8 0,132 B=3 S=2 B+S = 5 5 16QAM 1/3 2k 1/8 4,494 8 0,132 B=17 S=9 B+S = 26 A.13.2.3.3.2 Diversity gain evaluation with the SNR @ FER 5 method In order to evaluate the diversity gain in outdoor environment, and possibly compare directly none-diversity values with the theory and laboratory tests, additional measurements have been performed using only one single rooftop antenna.This test was done with SH-A receivers from DIBCOM (SHA2). Figure A13.118: FER vs SNR vs modulations for SHA2 (average) without diversity The summary of the results from the graphs above is in the table below. For the comparison with lab test, a SHA2 receiver has been used with a TU6 channel. Since the tests in lab have been made only for 5 and 90 km/h the C/N value extrapolated is an average of those 3 values. Table A13.40: Summary field tests vs lab tests vs receivers for QPSK 1/3 and 16QAM 1/3 without diversity for SHA2 Receivers Modulation Field SNR Lab SNR @ FER=5% @ FER=5% (50 km/h) SHA2 average QPSK 1/3 3 dB 2 dB SHA2 average 16QAM 1/3 8 dB 7.2 dB Table A13.41 Modulation C/N difference without diversity in in reception condition C for SHA2 ETSI 497 ETSI TS 102 584 V1.1.2 (2010-XY) Receivers Modulation 1 Modulation 2 Field difference Lab difference (50km/h) SHA2 QPSK 1/3 short 16QAM 1/3 short +5dB +3,6dB The results were conservative whatever the antenna used (antenna 1 or antenna 2). The trade-off is about the same as when diversity 2 is used. In the following, the results with diversity are presented: Figure A13.119: FER vs SNR vs modulations for SHA2 (average) with diversity The diversity gain can be calculated from the previous results: Table A13.42: Diversity gain vs modulations in reception condition C for SHA2 Receivers Modulation C/N with diversity Diversity gain (SNR @ FER=5%) SHA2 average QPSK 1/3 0 dB 3 dB SHA2 average 16QAM 1/3 4,5 dB 3.5 dB With this method, the diversity gain is 3 dB for QPSK 1/3 and 3.5dB for 16QAM 1/3. We can notice also that the performances in lab trials using a TU6 channel emulator are in line (less than 1 dB difference) with on field performances, confirming the TU 6 channel model validity. A.13.2.3.3.3 Conclusion on diversity gain in outdoor environment (reception condition C) In outdoor environment, reception diversity of second order brings a significant advantage (more than 3dB of gain in the tested configurations) on service quality and availability and above all on coverage extension. A.13.2.3.4 Performances of DVB-SH waveform and diversity gain analysis in reception condition D A.13.2.3.4.1 Diversity gain evaluation with the SNR @ FER 5 method In order to evaluate the diversity gain in in-car environment, and compare directly none-diversity values with the theory and laboratory tests, additional measurements have been performed using only one single antenna.. Both antennas have been tested in order to ensure the results are conservative and that the diversity gain is equal whatever the reference antenna. As before, that the test was done with SHA2 receivers. ETSI 498 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.120: FER vs SNR vs modulations for SHA2 (average) without diversity The summary of the results is presented in Table A13.43. Regarding the comparison with lab test; the results have been taken from measurements for TU6 channels.. As for the outdoor configuration, since the tests in lab have been made only for 5 and 90km/h, , the C/N value extrapolated from those 3 values. Table A13.43: Summary field tests vs lab tests for QPSK 1/3 and 16QAM 1/3 without diversity for SHA2 Receivers Modulation Field SNR Lab SNR @ FER=5% @ FER=5% (50km/h) SHA2 average QPSK 1/3 3.5 dB 3 dB SHA2 average 16QAM 1/3 8.5 dB 6.6 dB The following figure and table show the results with antenna diversity: Figure A13.121: Results with diversity Table A13.44 shows the diversity gain calculated from the trial results: Table A13.44: Diversity gain vs modulations for in-car environment for SHA2 Receivers Modulation C/N with diversity Diversity gain (SNR @ FER=5%) SHA2 average QPSK 1/3 0,5 2.9d B SHA2 average 16QAM 1/3 4 4.5 dB ETSI 499 ETSI TS 102 584 V1.1.2 (2010-XY) A.13.2.3.4.2 Conclusion on Diversity gain in reception condition D In reception condition D, reception diversity of second order brings a significant advantage (more than 3dB of gain in the tested configurations). A.13.2.3.5 Impact of the interleaving scheme The purpose of this set of trials is to assess the impact of the long interleaving scheme. In the following, a Class 2 interleaver is compared with and class 1 plus iFEC configuration. A.13.2.3.5.1 MPE-IFEC interleaving gain in Reception Conditions C and D The following figures compare DVB-SH C/N performance for Class I without iFEC (named in the legend Class 1) and with i-FEC of 5s duration (named iFEC). Results are presented for the two selected modulations. FER vs SNR for QPSK 1/3 Class I vs Class I + i-FEC FER vs SNR for 16QAM 1/3 Class I vs Class I + i-FEC 80% 100% Slight recovery in highly 60% degraded areas 80% 60% FER5 FER5 40% Class I Class I 40% i-FEC5 i-FEC5 20%  20% 0% 0% -5 -4 -3 -2 -1 0 1 2 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 SNR (dB) SNR (dB) Figure A13.122: Impact of IFEC 5s Figure A13.122 shows that the 5 s. iFEC lightly improves performances in the high PER area while the advantage at FER 5% goes from marginal (in QPSK) to 0.2 dB (in 16QAM). There is a sensible improvement when the iFEC length is extended to 26 seconds as seen in Figure A13.123: FER vs SNR for QPSK 1/3 Class I vs Class I + i-FEC FER vs SNR for 16QAM 1/3 Class I vs Class I + i-FEC 80% 100% 60%  80% 60% FER5 FER5 40% Class I Class I 40% i-FEC26 i-FEC26 20% 20% 0% 0%  -5 -4 -3 -2 -1 0 1 2 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 SNR (dB) SNR (dB) Figure A13.123: : Impact of IFEC 26 s An iFEC of 26 seconds provides between 0.5 and 0.75dB gain under pure terrestrial coverage. This is in line with TU6 lab measurements reported in clause A13.1. A.13.2.3.5.2 Class 2 interleaving gain in Reception Conditions C and D This section tackles the comparison between Class 1 (short interleaver) and Class 2 (long interleaver). For Class 2, two interleaver durations are analysed i.e. 5 and 10s. Class 1 results do not include iFEC. ETSI 500 ETSI TS 102 584 V1.1.2 (2010-XY) FER vs SNR for QPSK 1/3 Class I vs Class II 80%  60%  FER5 40% Class I Class2 5s 20%  0% -5 -4 -3 -2 -1 0 1 2 SNR (dB) Figure A13.124: DVB-SH C/N performance for Class I and Class2.5s An interleaver of 2.5 seconds allows recovering up to 20% of the FER in highly degraded areas with 1 dB of improvement at FER 5% (see Figure A13.124). Figure A13.125 shows the trials results for a 10 s interleaving. FER vs SNR for QPSK 1/3 Class I vs Class II FER vs SNR for 16QAM 1/3 Class I vs Class II 100% 80%  80% 60% FER5 60% FER5 Class I 40% Class I Class2 10s 40% Class2 10s 20% 20%  0% 0% -5 -4 -3 -2 SNR (dB) -1 0  1 2 -5 -4 -3 -2 -1 0 1 SNR (dB) 2 3 4 5 6 Figure A13.125: DVB-SH C/N performance for Class I and Class2 10s Increasing the interleaving length improves slightly the performance both in highly degraded areas and at PER5% Gains range from 0.5 dB (16QAM) to 1.25 dB (QPSK) under pure terrestrial coverage (slightly better than TU6 laboratory measurements). A.13.2.3.5.3 Conclusion on the use of long interleaving Figure A13.126 compares the performance of IFEC and Class 2 configurations. FER vs SNR for QPSK 1/3 for Class II & i-FEC FER vs SNR for 16QAM 1/3 for Class II & i-FEC 80% 100% 80% 60% FER5 FER5 Class I 60% Class I i-FEC5 40% i-FEC5 i-FEC26 40% i-FEC26 Class2 5s Class2 10s Class2 10s 20% 20% 0% 0% -5 -4 -3 -2 -1 0 1 2 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 SNR (dB) SNR (dB) Figure A13.126: DVB-SH C/N performance for i-FEC and Class2 interleavers The following conclusions can be drawn: • 5 s iFEC brings light improvements even in pure terrestrial coverage mainly in degraded areas; • 26 s iFEC brings more important improvements and a significant gain of C/N in terrestrial coverage; ETSI 501 ETSI TS 102 584 V1.1.2 (2010-XY) • Class 2 is very efficient in highly degraded areas (20 to 30% recovery rate) and with an important gain of C/N for FER5 criteria; • 5 s and 10 s Class2 show close performances. A.13.2.6 Synthesis of results Measurements without antenna diversity Table A13.45: Torino results without diversity at medium speed Modulation/coding Terminal Terminal C/N (dB) C/N (dB) Type Class @ FER=5 % @FER=5 % vehicular In-car medium speed medium speed QPSK 1/3 SHA2 Field Class 1 3 3,5 SHA2 Lab Class 1 2 2 16QAM 1/3 SHA2 Field Class 1 8 8,5 SHA2 Lab Class 1 7,2 7,2 Measurements of diversity gain outdoor and in car During the Torino Field Trial diversity measurements have been made with vehicular terminal and with terminal in-car at medium average speed. Outdoor: Table A13.46: Diversity gain in C conditions Modulation No diversity C/N (dB) Diversity C/N (dB) Diversity gain (dB) QPSK 1/3 Class 1 3 0 3 16QAM 1/3 Class 1 8 4,5 3,5 In car diversity gain Table A13.47: Diversity gain in D conditions Modulation No diversity C/N (dB) Diversity C/N (dB) Diversity gain (dB) QPSK 1/3 Class 1 3,5 0,5 3 16QAM 1/3 Class 1 8,5 4 4,5 Improvement with Long Interleaver Table A13.48: Improvement with Long Interleaving Modulation/coding Terminal Terminal Class Improvement in C/N (dB) @ Type FER = 5 % vehicular medium speed QPSK 1/3 SHA2 Field Class 1 +IFEC/ 5s 0,1 (Diversity 2) SHA2 Lab Class 1 +IFEC/ 5s 0,2 SHA2 Field Class 1 +IFEC/ 25 0,75 (Diversity 2) s ETSI 502 ETSI TS 102 584 V1.1.2 (2010-XY) SHA2 Lab Class 1 +IFEC/ Bellow 0,5 25s SHA2 Field Class 2 /9s 1,25 (Diversity 2) SHA2 Lab Class 2/ 9s No impact 16QAM 1/3 SHA2 Field Class 1 +IFEC/ 5s 0,2 (Diversity v 2) SHA2 Lab Class 1 +IFEC/ 5s 0,2 SHA2 Field Class 1 +IFEC/ 25 0,5 (Diversity 2) s SHA2 Lab Class 1 +IFEC/ 0,5 25s SHA2 Field Class 2 /9s 0,5 (Diversity 2) SHA2 Lab Class 2/ 9s No impact A.13.2.4 Measurements results of a DVB-SHA satellite signal from Paris to Barcelona During February 2010, the Q.o.S. of the DVB-SHA signal transmitted in S-Band by the W2A satellite was measured in a car diversity receiver via two antennas mounted on the roof. A.13.2.4.1 Receiver platform description The modulation parameters were as follows: • DVB-SHA Class1 • Frequency carrier : 2197.5 MHz • Right Hand Circular Polarization (RHCP) • C/N threshold measured in Laboratory : o AWGN: C/N = -1dB o TU6 : C/N = 1.3 dB The detail of the antennas mounted on the car roof is shown on the following picture: Table A13.127: Waveform configuration BW Modulation/ FFT GI Capacity Short Short Uniform Uniform MPE-IFEC (MHZ) coding (Mbps) Interleaving Interleaving Interleaving duration (s) duration (s) structure 5 QPSK 1/2 2k 1/4 2,222 8 0,265 N/A N/A N/A The configurations corresponds to Table A10.8 ID n° 7 of clause A.10. ETSI 503 ETSI TS 102 584 V1.1.2 (2010-XY)  Figure A13.128: Vehicle equipment overview The vehicle roof rails were removed to avoid masking the satellite signal. The chipset used for the test integrates two tuner-demodulator working in diversity mode (MRC). The received signal was sampled every second and a monitoring software delivered all the needed information to monitor and post-process the trial information(RF level, C/N, Frame_Lock, GPS parameters,…) A.13.2.4.2 Main measurements results A.13.2.4.2.1 Erroneous seconds on over 1000km of highways Considering the Frame_Lock criteria, the erroneous second rate (~MFER) is depicted on Figure A13.129 map for route segments of about one hour of duration:           Figure A13.129: MFER results all over the travel The global MFER calculated along the whole route is of only 4.2 % (without MPE_iFEC); ESR5(20) fulfilment is about 85%, in line with the results in Clause A12 (Figure A12.57). ETSI 504 ETSI TS 102 584 V1.1.2 (2010-XY) A.13.2.4.2.2 ”Frame_lock” loss events distribution analysis The probability density function of the loss of frame lock (masks) duration (Figure A13.130 ) shows that on the Highway: • up to 62.6 % of the frame lock loss events have a duration less or equal to one second • up to 24.7% of the frame lock loss events have a duration between 1s and 2s Probability density function of Masks duration Paris to Barcelona   400 350 Trip duration : 32312 s (~ 9 h) Number of seconds with no errors : 30951 (95.8%) 300 Number of seconds with errors : 589 ( 4.2%) 250   200       150     100     50  0  1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 Masks durations (s)  Figure A13.130: Probability density function of masks duration During the trip from Paris to Barcelona it was observed, that most of these short masks were due to short bridges crossing over the highway and to road signs masking the satellite. The associated cumulative distribution is depicted on the following figure.  ETSI 505 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.131: Distribution of the W2A satellite masks duration A.13.2.4.2.3 Erroneous seconds versus interleaving depth The tests described above were made without Class 2 interleaver or MPE-iFEC protection (only the 300 ms interleaver was active). However, from the traces recorded in the field, the performance of a Class 2 interleaver can be emulated. A13.133 shows the remaining erroneous seconds versus the interleaver depth, calculated along the whole route. % erroneous seconds vs interleaver depth 4.5% 4.0% 3.5% % of erroneous seconds 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% 0 1 2 3 4 5 6 7 8 9 10 Interl. corr. depth (s) Figure A13.132: Percentage of erroneous seconds versus interleaving depth In conclusion: • If the interleaver is able to compensate for interruption up to 1 second the erroneous second ratio is reduced to 3.1 % instead of 4.2 % • If the interleaver is able to compensate for interruption mask of up to 2 seconds the erroneous second ratio is divided by nearly two: 2.2 % instead of 4.2 % A.13.2.4.2.4 Percentage of interruptions versus interleaving depths In addition to the decreasing of erroneous seconds ratio the interleaver decreases also dramatically the number of interruptions as seen in Figure A13.133  % of masks vs interleaver depth (Total test duration =32312 s)      100% 90%  80%  70% % of masks 60%  50% 40%   30% 20%   10%  0% 0 1 2 3 4 5 6 7 8 9 10 Interl. corr. depth (s)  Figure A13.133: Percentage of interruptions versus interleaving depth ETSI 506 ETSI TS 102 584 V1.1.2 (2010-XY) Only 13% of the initial number of interruptions remains after correction by an interleaver of 2 s depth. A.13.2.4.3 Conclusions • The vehicular outdoor quality of coverage of the W2A satellite tested between Paris and Amsterdam and Paris to Barcelona confirms the good performance of DVB-SH in satellite only coverage in a highway environment with ICO and Devas satellites trials. • Between Paris and Barcelona the measured MFER (in DVB-SHA and without any long interleaving process) was of around 4% along 1000 km and during 9 hours, corresponding to ESR5(20) fulfilment ratio of 85%.. • An interleaving process correction simulation based on “frameLock” burst modelisation over the Paris- Barcelona highway shows that up to 87% of the burst have a duration less than 2 seconds • For Mobile TV reception from satellite, in a car travelling on highway, the results indicate a small interleaving with a capacity correction of about 2 seconds is sufficient for correcting most of the interruptions (road bridges, signalling panels,…). • Along the Paris-Barcelona route the resulting MFER, after a 2s Interleaving/de-interleaving process, is approximately divided by two compared to the original one, corresponding to ESR5(20) fulfilment ratio of 95%. A.13.2.5 Pisa trial These trials were carried out by a consortium of companies and institutions: Eutelsat, Fraunhofer IIS, DLR, UBS, SIDSA, MBI and Calearo co-funded under ESA ARTES34 program. The objective of these trials was to assess the performance of DVB-SH in the satellite and hybrid coverage provided in a variety of environments. Although the trials covered an extended set of configurations, this section will focus of DVB-SHB results. A.13.2.5.1 Experimental Setup The position of the two CGC repeaters (named MBI-1 and MBI-2) with respect to the test area is indicated in Figure A13.134. The two repeaters, both located at a rough height of 15 meters, aim to cover the city centre of Pisa, equipped with a 50 W amplifier over a unique sector, providing an EIRP of ~47dBm+18 dB (antenna gain) – 3 dB (losses) = 62 dBm. Both repeaters were directly pointing the center of Pisa. The CGC intends to complement the satellite coverage provided by the W2A satellite. The overall duration of the route does indeed depend on the traffic conditions. Nevertheless, the estimated time for the completion of each session is in the order of ~38 minutes. ETSI 507 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.134. Position of the CGC repeaters over the test area (above). The route used for the measurements is tracked (red-urban, blue-suburban, green-tree shadowing, below). Pisa, Italy. These pictures have been obtained via GoogleMaps (maps.google.com). The measured received signal power (terrestrial only) along the route is depicted in Figure A13.135, while the predicted C/N is displayed in Figure A13.136. Figure A13.135. Measured received power on Route 1 (terrestrial signal only). ETSI 508 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.136. Predicted C/N for the terrestrial signal. Figure A13.137. Route 1. ETSI 509 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.138. Route 1, point A (left) and B (right). Figure A13.139. Route 1, point C (left) and D (right). Figure A13.140. Route 1, point E. A.13.2.5.2 Description of the trials conditions Trials were carried out on the above-described route in different configurations of the network: in satellite-only coverage, in terrestrial-only coverage, and with code combining. Whenever possible, for the tests, two receivers provided by two distinct manufacturers have been simultaneously used. The measurement setup has been organized in a way that the two receivers had access to the same antenna signal. The receivers were operated with a G/T of -21.5 dB/ ºK. The LOS C/N measured on the satellite component is around 9 dB. The output of the two receives was input to dumping servers, collecting measures of quality of service at different layers, e.g.: ETSI 510 ETSI TS 102 584 V1.1.2 (2010-XY) - C/N values - Physical layer / Transport Stream level error flags - MPE-level (burst) error flags A common dumping architecture was foreseen for all the receivers used during the Pisa trials. As showed in Figure A13.141 each receiver has to provide the output data to a centralized server, that will handle them in parallel with the data extracted from a GPS receiver, from a gyroscope system and from a fish-eye camera. Figure A13.141. Dumping architecture The first receiver was equipped with the following features: • S-Band Tuner • Antenna diversity with up to 4 antennas • DVB-SH OFDM and TDM demodulator • Maximum-Ratio-Combining(MRC) • Code-Combining(CC) between OFDM and TDM signals • Class 1 and Class 2 decoding capabilities The log-likelihood ratio data streams coming from identical demodulators are combined using the MRC method. The combined LLR are de-interleaved and then handed over to the turbo decoder. MPE-iFEC decoding (for Class 1 configurations) is performed on a separate platform with a software (ANSI-C) implementation. The second receiver is a modular, multi-standard (DVB-T/H/SH) FPGA-based real-time prototyping. The real-time prototype is divided in two main parts: • The main board. It is built around a Virtex-5 LX330 FPGA. This board contains the digital part of the demodulator. It also contains the external memories needed for Class-2 deinterleaver, along with the ARM7 main processor and the interfaces with the host. • The RF board. This board contains a multi-band (UHF, S-Band) NXP tuner. For the these field trials, the demodulator contains two RF boards and two main (digital) boards where the whole receiver is mapped. The ARM7 processor subsystem adds flexibility to the receiver and controls the reception process. The receiver supports the entire DVB-SH standard: • Class-1 and Class-2 deinterleaver profiles. • SH-A, either MFN or SFN configurations. In SFN configuration, MRC antenna diversity is supported. ETSI 511 ETSI TS 102 584 V1.1.2 (2010-XY) • SH-B, with Code-Combining between TDM and OFDM signals. • MPE-IFEC processing is also performed by the demodulator. • SH-B with code-combining and MRC antenna diversity of 2 could be supported by interconnecting several receivers for a total of 4 receiving antennas. A.13.2.5.3 Selected configurations The two configurations (denoted in the following of the document as SHB#2 and SHB#8) selected for the trials and are summarized in the following tables: Table A13.49: Configurations used for the on-field trials                                                                                                                                                                                                          Clearly the configuration SH B#2 deals with a Class-1 type, with satellite component configured with QPSK modulation and coding rate ½ (ref. to Table A32, id 7 in clause A.10, with a slightly longer interleaver, 244 ms vs. 181 ms), while the CGC part is configured with 16-QAM and code rate 1/3 (ref. to Table A6, id 5 clause A.10, with a slightly longer interleaver, 122 ms vs. 105 ms ). The scheme is complemented by an MPE-iFEC with B=17 and S=9, with nominal coding rate 2/3 (although after post-processing, it resulted to be actually closer to ½ due to padding). ETSI 512 ETSI TS 102 584 V1.1.2 (2010-XY) The configuration SH B#8 is a Class-2 type, with an interleaver of 5.6 seconds. The satellite component is protected by a rate 1/3 code, with 8-PSK modulation (close to ref. to Table A53, id 2 in clause A.10), while the CGC is still based on a 16-QAM signalling and on coding rate 1/3 (ref. to Table A25, id 2 in clause A.10, with a different interleaver profile). Table A13.50: Summary of the configurations used for the trials, with reference to clause A.10. ID Waveform configuration Additional Spectral SAT C/N C/I Comments Parameters Efficiency SHB2 SAT:T-QPSK_1/2_S Table A32 id 7 0.67 (with iFEC) AWGN C/N req=1.1 dB TER:16QAM_1/3_S Table A6 id 5 0.89 (with iFEC) AWGN C/N req=3.7 dB SHB8 SAT:T-8PSK_1/3_U Table A53 id 2 1 AWGN C/N req=1.6 dB TER:16QAM_1/3_UL Table A25 id 2 1.25 AWGN C/N req=3.7 dB A.13.2.5.4 Trial results The trials results are summarized in table Table A13.26 The overall test duration is indicated, and it is an approximated value given by the average dump lengths for the two receivers. Note that the slight differences in length of the dumps is due to different starting/stopping instants, and are usually less than 1 minute, which in any case may result in a bias of the measured service availabilities in the order of +/- 1%. We recall that the two configurations tested on the field: • A class-1 configuration (SHB2), with satellite component configured with modulation QPSK and coding rate ½, while the CGC part is configured with 16-QAM and code rate 1/3. The scheme is complemented by an MPE-iFEC with B=17 and S=9, with nominal coding rate 2/3. The MPE-iFEC duration is 25 seconds. The actual measured code rate for MPE-iFEC is actually lower, due to padding in the ADT tables, and it is close to ½. Hence, the overall efficiency of the class-1 configuration for the TDM part is approximately 2*(1/2)*(1/2)=0.5 b/s/Hz. • A class-2 configuration (SHB8), with an interleaver of 5.4 seconds. The satellite component is protected by a rate 1/3 code, with 8-PSK modulation, while the CGC is still based on a 16-QAM signalling and on coding rate 1/3. Hence, the overall efficiency of the class-2 configuration for the TDM part is approximately 3*(1/3)=1 b/s/Hz. ETSI 513 ETSI TS 102 584 V1.1.2 (2010-XY) Table A13.51: Summary of the field trial results. Receiver Test Network status ESR5(20) Service Availability Configuration Class duration [s] Satellite CGC RX 1 RX 2 SHB#2 1 1900 ON ON 94% - SHB#2 1 2050 ON ON 95% - SHB#2 1 2100 ON ON 92% 93% SHB#2 1 1700 ON - 55% 54% SHB#2 1 2050 ON - 48% 52% SHB#2 1 1750 OFF ON 72% 70% SHB#8 2 2100 ON ON 89% - SHB#8 2 2120 ON ON 91% - SHB#8 2 2050 ON ON - 90% SHB#8 2 2150 ON ON 90% - SHB#8 2 1850 ON - 39% 40% SHB#8 2 2600 ON OFF* 50% - SHB#8 2 2200 OFF ON 68% 65% * - CGC turned off to avoid interference of the sat signal Tests have been carried out with 3 different network topologies: satellite-only (with CGC switched on) trials, terrestrial- only trials, and code-combining trials. A summary of test results is terms of service availability (ESR5(20) criterion) is provided in table Table A13.26. The Class-1 configuration is clearly advantaged by its lower spectral efficiency with respect to the Class-2 configuration. However, in both cases, the code combining feature has ensured a seamless handover between satellite and terrestrial signals that is visible in the good service availability percentages achieved in hybrid reception. For the Class-1 configuration, service availability close to 95% is achieved in hybrid reception, while a terrestrial-only reception would provide roughly 73%. Similarly, for the Class-2 configuration, a service availability of 90% can be achieved thanks to code combining, while in the terrestrial-only case it would not exceed 70%. In the table provided next, the performance of the two receivers in different sessions is depicted as well, confirming the stability of the obtained results More insights on the performance of the system can be obtained by evaluating the ESR5(20) criterion over location. The results for the class-1 case are provided in Figure 7.50A13.142, while those for the class-2 configuration are presented in Figure 7.51A13.143. Figure A13.142. Service availability vs. location, Class 1 configuration. Terrestrial only (left), Satellite only (mid), code combining (right). Red spots identify areas in which the ESR5(20) criterion is not fulfilled. Receiver 1 (top) and 2 (bottom). In both cases, the satellite component seems to suffer mostly in the parts of the route which are developed along the East-West axis. Interestingly, in areas where the neither the satellite component nor the CGC alone can provide ETSI 514 ETSI TS 102 584 V1.1.2 (2010-XY) coverage, their combination is successful. This is especially evident in Figure 7.51A13.143, on the western and in the northern parts of the route. Figure A13.143. Service availability vs. location, Class 2 configuration. Terrestrial only (left), Satellite only (mid), code combining (right). Red spots identify areas in which the ESR5(20) criterion is not fulfilled. Receiver 1 (top) and 2 (bottom). The role of code combining is especially visible in the class 2 configuration case. A larger detail on that is presented in Figure A13.144: a part of the route (urban environment) where both the terrestrial only and the satellite only coverage are not sufficient to fulfil the ESR5(20) criterion while the code combining allows recovering the errors. ETSI 515 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.144. Role of code combining. Class 2 configuration. Terrestrial only (top), Satellite only (mid), Code combining (bottom). An environment-based analysis is presented hereafter. The route has been split in 5 parts (according to Figure ). The initial/final part of the route has been divided in two sub-parts (to take into account the different travelling direction), see Figure A13.145, parts A1/A2. ETSI 516 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.145: Route segmentation by environment. For the class-1 configuration, the analysis is provided in the next figures. ETSI 517 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.146: Window status and ESR5(20) calculated for the different environments: A1 (top left), B (top right), C (centre left), D (centre right), E (bottom left), A2 (bottom right). Code combining. ETSI 518 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.147: Window status and ESR5(20) calculated for the different environments: A1 (top left), B (top right), C (centre left), D (centre right), E (bottom left), A2 (bottom right). Terrestrial only. ETSI 519 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.148: Window status and ESR5(20) for the complete session (top) and for the different environments: A1 (top left), B (top right), C (centre left), D (centre right), E (bottom left), A2 (bottom right). Satellite only. ETSI 520 ETSI TS 102 584 V1.1.2 (2010-XY) The results for the class-2 configuration are depicted next. Figure A13.149: Window status and ESR5(20) for the complete session (top) and for the different environments: A1 (top left), B (top right), C (centre left), D (centre right), E (bottom left), A2 (bottom right). Code combining. ETSI 521 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.150: Window status and ESR5(20) for the complete session (top) and for the different environments: A1 (top left), B (top right), C (centre left), D (centre right), E (bottom left), A2 (bottom right). Terrestrial only. ETSI 522 ETSI TS 102 584 V1.1.2 (2010-XY) Figure A13.151: Window status and ESR5(20) for the complete session (top) and for the different environments: A1 (top left), B (top right), C (centre left), D (centre right), E (bottom left), A2 (bottom right). Satellite only. ETSI 523 ETSI TS 102 584 V1.1.2 (2010-XY) A.13.2.5.4 Hybrid SFN trials Trials in Pisa also covered hybrid SFN configurations where signals from satellite and terrestrial were combined in SHA SFN mode. A.13.2.5.5 Conclusion on Pisa trials These trials have proved that the code combining mechanism allows improving the coverage of DVB-SH systems in areas where neither the satellite nor the terrestrial signal could provide a good Q.o.S.. The capacity of the satellite to improve the CGC coverage in urban and sub-urban areas has also been confirmed. By comparing two different waveforms with different spectral efficiencies, it has been shown that the long interleaver could be used to match the performance of a more robust (i.e. less spectral efficient) modulation and coding without trading-off the system capacity. This comes at the expense of the need of a longer memory and additional complexity in the receiver. A.13.2.6 Conclusion on trial results This section has provided a compilation of the different trial campaigns carried out to assess the performances of DVB- SH in real scenarios. All trials results confirm the performances expected and they do not deviate in more than few tens of dB from the results obtained in the laboratory tests. The performance of the main features of DVB-SH has been assessed, namely: SFN gain, Code Combining gain, long Physical layer interleaver and MPE IFEC protection. Satellite coverage has been evaluated in a highway scenario and in urban, sub-urban and rural environments. DVB-SH performance in satellite-only coverage is very good for the highway scenario and it can be further boosted thanks to the long interleaver protection. Satellite does also improve the coverage of the CGC in rural and sub-urban areas. Even though long protection mechanisms (both long interleaver and MPEiFEC) where introduced to improve the performance on satellite channels, results show that they are also beneficial for terrestrial-only channels. Antenna diversity has been proved to be an interesting technique to improve the coverage with a gain between 3 to 5,5 dB in the configurations analysed. ETSI 524 ETSI TS 102 584 V1.1.2 (2010-XY) Clause B (informative): Interoperability with cellular telephony networks B.1 Introduction Mobile terminals can contain several radios and therefore co-existence of the DVB-SH receiver working in L-band (1 452 MHz to 1492 MHz) and S-band (2 170 MHz to 2 200 MHz) with other radios, either cellular or connectivity, is required. The cellular radios can be GSM/EDGE (including GSM850, GSM900, DCS1800 and PCS1900) or UMTS while the connectivity radios can be wireless LAN or Bluetooth. Simultaneous operation of the DVB-SH receiver in combination with one of the cellular or connectivity radios in a small sized terminal is very challenging. B.1.1 General coexistence issues The co-existence issues described in this clause consider worst case conditions: two main co-existence issues for the DVB-SH receiver can be distinguished: • interference of the uplink signal of a simultaneous (cellular or connectivity) transmission with the DVB-SH receiver: - wanted transmitted signal: desensitization; - unwanted transmitted signal: power amplifier noise and spurious responses; • interference of the downlink signal of a simultaneous (cellular or connectivity) reception with the DVB-SH receiver: - wanted transmitted signal from a base station: adjacent channel selectivity (ACS); - unwanted transmitted signal from a base station: downlink Adjacent Channel Leakage Ratio (ACLR). Next to this, undisturbed operation of the cellular or connectivity radio in presence of the DVB-SH receiver must also be maintained. Because DVB-SH does not have a uplink signal, the possible impairments caused by DVB-SH receiver are limited to: • out of band unwanted signals in cellular or connectivity downlink (RX) band; • effects to the cellular antenna pattern. These problems are pure implementation issues and can be solved by proper terminal design. B.1.2 Terminal Architectures The architecture (relevant parts) of a typical terminal category 3 containing a DVB-SH receiver and a GSM, UMTS or WLAN radio is presented in figure B.1. ETSI 525 ETSI TS 102 584 V1.1.2 (2010-XY) Figure B.1: Terminal architectures with cellular radios Most probably the DVB-SH receiver and the cellular radio will have physically separate antennas. The antenna isolation between the antennas is frequency dependent and will be detailed in the next clause. An important difference between UMTS and GSM radios is the duplex filter. UMTS will use duplex filters, but the majority of modern GSM radios uses a Tx/Rx switch. This has a major implication on the co-existence. The cellular radio uplink unwanted signal interference to the DVB-SH receiver will not be a problem in UMTS terminal if a duplexer is used. However the problem will be severe in a GSM terminal with Tx/Rx switch. It is worth noting that in case of terminal category 2 the cellular radio transmitters are not considered. However, the WLAN radio transmitter might be present causing also signal interferences. B.2 Frequency bands and power levels Table B.1 gives an overview of the most relevant interferers for DVB-SH including frequency bands and power. For uplink signals, the transmit power and the assumed antenna coupling values are given for S-band receiver. Together they give the received power at the DVB-SH antenna. Table B.1: Overview of interferers for DVB-SH receiver Name Frequency band Transmit Antenna Power at DVB-SH power coupling antenna GSM850 Tx 824 MHz to 849 MHz 33 dBm -18 dB 15 dBm GSM900 Tx 880 MHz to 915 MHz 33 dBm -18 dB 15 dBm DCS1800 TX 1 710 MHz to 30 dBm -15 dB 15 dBm 1 785 MHz PCS1900 TX 1 850 MHz to 30 dBm -15 dB 15 dBm 1 910 MHz FDD 3G band 1 Tx 1 920 MHz to 24 dBm -9 dB 15 dBm 1 980 MHz DD 3G band 1 Rx 2 110 MHz to n.a. n.a. 2 170 MHz ISM2400 (BT-WLAN) 2 400 MHz to 20 dBm -9 dB 11 dBm 2 484 MHz The most problematic cases are the wanted uplink and downlink signals from the FDD 3G band 1 (UMTS) due to its close proximity to the DVB-SH signal band in S-band. ETSI 526 ETSI TS 102 584 V1.1.2 (2010-XY) B.3 Interference due to uplink signal B.3.1 Desensitization The transmitted cellular or connectivity signal has a very high power compared to the received DVB-SH signals. FDD 3G band 1 is the closest one with high power and will be considered as the worst-case situation. FDD 3G band 1 transmitted power is +24 dBm. Part of this power is coupled to the DVB-SH antenna. The coupling loss is between the FDD 3G band 1 antenna and the DVB-SH antenna is assumed to be 9 dB. The received power at the DVB-SH antenna is therefore +15 dBm. Without any filtering the cellular TX signal present in the DVB-SH receiver input would be also +15 dBm. This very high interference signal level would cause severe blocking effects and reciprocal mixing. The practical solution for co-existence is to insert a rejection filter in front of the DVB-SH receiver. The filter has to attenuate the FDD 3G band 1 Tx-signal to the allowed out of band unwanted signal level. The maximum level for hand portable terminal is -25 dBm. Stop band attenuation of the filter thus becomes: A filter = PTX − AA − Pmax = +24 dBm − 9 dB − ( −25 dBm ) = 40 dB (B.1) where: Afilter = stop band attenuation of the reject filter; PTX = Tx output power; AA = coupling between the antennas (-9 dB); Pmax = maximum allowed power at the DVB-SH receiver input. Typically the insertion loss of the filter is in the order of 1,5 dB as described in clause 10. The sensitivity degradation is affected by the attenuation of the RF filter, by the receiver desensitization and phase noise performances and by antenna coupling between the two antennas at the blocker frequency. B.3.2 Spurious and transmitted PA noise Besides the wanted part of the FDD 3G band 1 uplink signal, the signal also contains unwanted parts like spurious responses and the out of band PA noise for terminal architectures without duplexer. The GSM specification (TS 100 910 [30]) defines that within 100 kHz measurement bandwidth the power shall not be greater than -79 dBm within GSM900 Rx frequency band (-71 dBm in case of DCS1800). This noise floor corresponds to the noise contribution of the transmitter power amplifier. That value can be extended for a noise floor value within the DVB-SH bands as a maximum value. Assuming a 18 dB coupling loss between GSM900 and DVB-SH S-band antennas, the interference power entering DVB-SH receiver would be -80 dBm for 5 MHz bandwidth. With a 4,5 dB receiver noise figure, the total receiver input noise floor is –102,7 dBm. From this it is obvious that the transmitter output noise reduces the DVB-SH receiver sensitivity considerably. In order to degrade DVB-SH receiver sensitivity "only" by 3 dB the transmitter output noise would need to be lower or equal to the receiver thermal noise floor (-102,7 dBm). Fortunately, noise in S band from GSM quadriband PA is lower than ETSI specification (20 dB typically). Nevertheless, if DVB-SH high sensitivity performance is targeted, an additional filtering will be required on TX power amplifier outputs. Interoperability requirements To guarantee interoperability between the radio systems the noise power at the DVB-SH receiver input (@ 5 MHz band) must fulfil the mask shown in figure B.2. ETSI 527 ETSI TS 102 584 V1.1.2 (2010-XY) Figure B.2: Tx PA-noise Mask in S-band DVB-SH Receiver Input The noise level is affected by the noise power of the power amplifier, by the attenuation of the possible low pass filter at the output of the transmitter PA as shown in figure B.3 and by antenna coupling between the two antennas at the DVB-SH reception frequency band. Figure B.3: GSM Tx Block Diagram B.4 Cellular Radio Downlink Signal Interference to DVB-SH Receiver The location of the DVB-SH received signal in S-band is next to the UMTS FDD downlink frequency range. This represents an additional interferer. Consequently, this is the most critical case for downlink signal interference. B.4.1 Adjacent channel Adjacent channel denomination will characterize hereafter 5 MHz adjacent channel (UMTS downlink channel next to DVB-SH channel) and per extension 10 MHz adjacent channel (one unoccupied UMTS downlink channel between UMTS downlink interferer and DVB-SH channel). Adjacent interference is summarized through Adjacent Channel Interference Ratio (ACIR). This is defined by: 1 1 1 = + ACIR ACS ACLR ACLR: Adjacent Channel Leakage Ratio which determines the power ratio transmitted by the UMTS downlink Base Station in its adjacent channel. ACS: Adjacent Channel Selectivity which defines the ability of the DVB-SH receiver to demodulate the wanted signal in presence of interferer in its adjacent channel. The UMTS specification TS 125 104 [15] gives a minimum of 45 dB ACLR for 5 MHz adjacent channel and a minimum of 50 dB ACLR for 10 MHz adjacent channel. ETSI 528 ETSI TS 102 584 V1.1.2 (2010-XY) B.4.2 ACS requirements A minimum receiver ACS must be achieved in order to avoid high DVB-SH receiver ACS contribution in ACIR value considering implementation margins in UMTS downlink ACLR. B.4.2.1 Example of ACS value measurement Different measurements have been made on DVB-SH receivers. The following pictures summarises the measurements methods., which is also explained in detail in ETSI/MSS/ cc01d04r1.        Figure B.4: Adjacent channel test bench In this picture, the signal generator plays the role of the interferer, and generates a signal in excess power of M dB above the useful signal          Figure B.5: Tx Adjacent channel spectrum This picture shows the position of the main signal between two potential interferers at N-1 and N+1 frequency slots, where N is the main. The margin, called M represents the excess power of the interferer. The useful signal input power is set at -70 dBm, which corresponds to at least 25 dB above the required power to get the FER 5 % in AWGN. ETSI 529 ETSI TS 102 584 V1.1.2 (2010-XY) Then the interferer power is increased until the useful signal is at the limit of required QoS. At that time we can write : ACS = M +C/N0.. The different measurements show that ACS is above 50 dB for N±1 and above 60 dB for N±2. In the Future Harmonised Standard ETSI EN 302 574-2 for User equipment in the MSS bands, the requirement for the ACS is the following: • 48 dB for N±1 • 55 dB for N±2. B.4.2.2 Recommendations The ACS performances requirement shall be at minimum 48 dB for the N±1 channels and 55 dB for N±2 channels. It should be possible to reach 50 dB and 60 dB as in the studied chipset. B.4.3 Co channel Interference The co-channel test consists of adding an interfering DVB-SH spectrum on the considered spectrum. This situation can be encountered in MFN networks on cell borders in area where signals from both cells can be received with equivalent level. The same test bench as adjacent channel tests is used except that the frequency of the interfering signal is the same as the useful signal. The interfering signal uses the same DVB-SH modulation parameters with PRBS mode. The method of the measurement consists of: • Set the power of the useful signal to a given fixed value. • Set the power of the interferer to find the FER5% criterion on the receiver. Two scenarios are provided: one with QPSK 1/3 and one with 16QAM 1/3. The results are presented in Table B.2. Table B.2: Co-channels test results Max interferer Co-channel limit AWGN C/N C/I @ FER5 compared to AWGN Scenario 1 0,4dB -0,7dB 1,1dB Scenario 2 6,1dB 4,6dB 1,5dB So C / I co − channel [dB] = 1.5 + [C / N AWGN ]dB ETSI 530 ETSI TS 102 584 V1.1.2 (2010-XY) Clause C (informative): Spectrum efficiency and system throughput C.1 Spectrum efficiency analysis The following spectral efficiency calculations shall be considered as illustrative and by no means representative of a specific real system. In this analysis, the frequency plan is assumed to be homogenous over the whole coverage region. It is considered thatwhen the common content retransmission takes place in another frequency sub-band (MFN configuration) we assumed that the terrestrial physical and upper layer parameters can be different from the satellite ones. This allows inclusion of local content on top of the common content in the terrestrial retransmission. The following configurations can be identified. Table C.1 Waveform SFN/MFN Config 1 SH-A MFN Config 2 SH-A SFN Config 3 SH-B MFN Examples for a 3-color reuse pattern of a 3x5 MHz satellite system are depicted below. The drawings show the frequency plan in a particular beam, with "Slot A" designating the frequency assigned to the satellite beam in question (this is the "satellite-protected" frequency, in which no adjacent beam interference occurs). Not the drawings are not to scale and that the frequency slots can be of different width in the different cases.             Figure C.1: Configuration 2 (SH-A, MFN, No split band) Figure C.2: Configuration 3 (SHA, SFN, No Split band) ETSI 531 ETSI TS 102 584 V1.1.2 (2010-XY) Figure C.3: Configuration 5 (SH-B, MFN, No split band) C.1.1 Spectral Efficiency Calculations We define the waveform spectral efficiency as the net bit rate of each waveform divided by the channelbandwidth. The following equations give the spectral efficiencies and the net bit rate as a function of modulation and coding parameters: for TDM: the net bit rate supported by the TDM carrier Cap _ TP[TDM ] is computed in clause 7. The TDM spectral efficiency is then: TDM η(TDM) = Cap _ TP[TDM ] / BW , TDM being BW the TDM channelization bandwidth; for OFDM: the net bit rate supported by the OFDM carrier Cap _ TP[OFDM ] is computed in clause 7. The OFDM spectral efficiency is then: OFDM η(OFDM) = Cap _ TP[OFDM ] / BW , OFDM being BW the OFDM channelization bandwidth. Table C.2 contains numerical examples for the bit rate and spectral efficiencies calculations for TDM and OFDM waveforms having assumed: OFDM TDM channelization bandwidth = BW = BW = 5 MHz; TDM roll-off factor αTDM = 0,15; OFDM guard interval GI = 1/8. Table C.2 TDM, QPSK, FEC_rate 1/3 Net bit rate in 5 MHz= 2,633 Mbps η(TDM) = 0,53 b/s/Hz TDM, 16APSK, FEC_rate ¼ Net bit rate in 5 MHz= 3,950 Mbps η(TDM) = 0,79 b/s/Hz OFDM, QPSK, FEC_rate 1/3 Net bit rate in 5 MHz= 2,468 Mbps η(OFDM) = 0,49 b/s/Hz OFDM, 16QAM, FEC_rate ¼ Net bit rate in 5 MHz= 3,730 Mbps η(OFDM) = 0,75 b/s/Hz We are interested in two spectral efficiencies at beam level: • the Total spectral efficiency per beam, η(TOTAL), defined as the net bit rate used for all content (Common+Local) available in the CGC coverage of a beam divided by the satellite-protected bandwidth allocated to that beam; ETSI 532 ETSI TS 102 584 V1.1.2 (2010-XY) • the Common content spectral efficiency per beam, η(COM), defined as the net bit rate of the Common content available in the beam divided by the same bandwidth as above. For the clarity of the formulas below, we use the following notations to further distinguish between different use cases of the OFDM waveform: • η(OFDM): Efficiency of the OFDM waveform when used in CGC only; • η(OFDM,Sat): Efficiency of the OFDM waveform when used in satellite, without SFN; • η(OFDM,Hyb): Efficiency of the OFDM waveform when used jointly in satellite and CGC (SFN). Let fR be the number of colours in which we divide the overall bandwidth available in our system (i.e., the total number of sub-bands in a beam). Table C.3 gives the formulas for computing these efficiencies, once the waveform spectral efficiencies and the satellite frequency reuse have been selected. Table C.3 Configuration Total spectral efficiency per Common content spectral beam, η(TOTAL) efficiency per beam, η(COM) 1) SH-A, MFN (fR -1) x η(OFDM) η(OFDM,Sat) 2) SH-A, SFN (fR -1) x η(OFDM) + η(OFDM,Hyb) η(OFDM,Hyb) 3) SH-B, MFN (fR -1) x η(OFDM) η(TDM) Comments: In deriving the above formulas for MFN, it has been implicitly assumed that the terrestrial efficiency is always higher than or equal to the satellite efficiency so that local content can be included on top of the common content in the terrestrial retransmission. In configuration 2), if the Hybrid frequency sub-band uses the same modulation and coding as the other terrestrial sub-bands then η(OFDM) = η(OFDM, Hyb) and the Total spectral efficiency per beam is fR x η(OFDM). The terrestrial frequency reuse in a beam is exactly the satellite frequency reuse between beams. If Config 2 is to be compared with Config 3, their Total spectral efficiency per beam differs by the η(OFDM,Hyb), in favour of Config 2. As concerns the Common content spectral efficiency per beam, η(TDM) is usually higher than η(OFDM,Hyb), in favour of Config 3. If the satellite system is composed of Nbeams, the system spectral efficiency can be defined as the total bitrates summed over all beams divided by the total frequency allocated to that satellite system. The following relationship can be easily shown: System spectral efficiency = (Nbeams/fR) x (Total spectral efficiency per beam) C.1.2 Numerical examples In the following example of efficiency comparison between the different 5 possible configurations in a 6-beam, 3-color reuse system. For sake of simplicity the waveform used in all cases has a 5 MHz bandwidth. For this reason the split configurations 1) and 4) require a 10 MHz beam sub-slot bandwidth instead of 5 MHz as for cases 2), 3) and 5). Table C.4 TDM, QPSK, FEC 1/3 η(TDM) = 0,53 b/s/Hz Satellite in Config 3 OFDM, QPSK, GI=1/8, FEC 1/3 η(OFDM) = 0,49 b/s/Hz Sat/Hybrid in Config 1, 2 OFDM, QPSK, GI=1/8, FEC ½ η(OFDM) = 0,75 b/s/Hz CGC only ETSI 533 ETSI TS 102 584 V1.1.2 (2010-XY) Table C.5 Configuration η(TDM) η(OFDM) η(OFDM) η(COM) η(TOTAL) with System Hybrid or Sat fR =3 (6 beams) 2) SH-A, MFN NA 0,75 0,49 0,49 2 x 0,75 = 1,5 6/3 x 1,5 = 3 3) SH-A, SFN NA 0,75 0,49 0,49 2 x 0,75 + 0,49 = 1,99 6/3 x1,99 = 3,98 5) SH-B, MFN 0,53 0,75 NA 0,53 2 x 0,75 = 1,5 6/3 x1,5 = 3 The efficiencies above can be converted into net bitrates as follows, assuming a 5 MHz waveform channelization. Table C.6 Configuration Total system Common bit Total bit rate in System bandwidth/Beam rate in a beam a beam (6 beams) sub-slot bandwidth 2) SH-A, MFN 15/5 MHz 2,47 Mbps 7,5 Mbps 45 Mbps 3) SH-A, SFNt 15/5 MHz 2,47 Mbps 10 Mbps 60 Mbps 5) SH-B, MFN 15/5 MHz 2,63 Mbps 7,5 Mbps 45 Mbps C.2.3 Conclusions on System Spectrum Efficiency It is difficult to provide general recommendations on the best system frequency planning configuration as decision depends not only on spectral efficiency. In fact when deciding the optimal frequency planning one should also consider other system parameters such satellite antenna beam patters and gap fillers location which may impact the location and extent of exclusion zones (see clause 11). Based on the examples illustrated before from a pure spectral efficiency perspective the following conclusions can be derived: • when the content to be broadcasted is different for different regions (e.g. linguistic regions) multibeam satellite configuration increase the common content spectral efficiency proportionally to the ratio between the number of beams and the number of colours used ( Nbeams / f R ) ; • SH-A/SFN configuration provides the highest overall spectral efficiency thanks to SFN operation between satellite and terrestrial gap fillers; • SH-B achieves a slightly higher spectral efficiency than SH-A for common content delivery but has a lower spectral efficiency for local content distribution than SH-A. Power efficiency issues (e.g. HPA nonlinearity effects) are not part of this comparison. . As stated above the above discussion is only looking at the spectral efficiency aspects. More extensive discussion about SH-A and SH-B waveform pro and contra can be found in clause 7. C.2 Throughput calculations DVB-SH capacity at MPEG TS interface level may be calculated from the parameters defined in the waveform definition [1]. Specifically, it is found by first calculating the duration of the SH frame, and the number of MPEG TS packets in an SH Frame, and then using these results to determine the capacity. To support seamless hand-over between OFDM and TDM, the SH frame length of the TDM part has been aligned to the SH frame length in OFDM. Therefore, the TDM parameters can not be calculated independently but imply a selection of OFDM parameters which is represented in the following equations. C.2.1 Reference DVB-SH parameters The throughput calculation is based on some reference DVB-SH parameters which are recalled in this section. ETSI 534 ETSI TS 102 584 V1.1.2 (2010-XY) C.2.1.1 OFDM Frame The time related parameters of the OFDM frame are defined in multiple of the elementary period T. From EN 302 583 [1], clause 5.7.4.1, 7 T= (C.1) 8 ⋅ BW _ OFDM with BW_OFDM being the OFDM signal bandwidth, in MHz; BW_OFDM=1,6 MHz (for a "1,7 MHz" channel), 5 MHz, 6 MHz, 7 MHz or 8 MHz. NOTE 1: T is in S in the above equation; alternatively, T is in seconds if BW_OFDM is in Hz. NOTE 2: the elementary period of the 1,7 MHz Channel, defined in EN 302 583 [1], clause 5.7.4, is equal to 7/12,8 µs; this corresponds to a BW_OFDM of 1,6 MHz. The duration Ts of an OFDM symbol, defined in EN 302 583 [1], clause 5.7.4.1, is: TS = 1024 ⋅ Mod _ FFT ⋅ (1 + GI ) ⋅ T (C.2) with Mod_FFT being the FFT length in k; Mod_FFT = 1, 2, 4 or 8 and GI being the Guard Interval ratio; GI=1/4,1/8,1/16 or 1/32. The number Nb_Symb_OFDM_Frame of OFDM symbol per OFDM frame, defined in EN 302 583 [1], clause 5.7.4.1, is: Nb_Symb_OFDM_Frame = 68 (C.3) The number of data subcarriers Nb_Data_Carrier per OFDM symbol, defined in EN 302 583 [1], clause 5.7.2, is: Nb_Data_Carrier = 756 Mod_FFT (C.4) with Mod_FFT being the FFT length in k; Mod_FFT = 1, 2, 4 or 8. The number of bits per modulated data subcarrier in OFDM BpS_OFDM is: BpS_OFDM=2 or 4 (C.5) with BpS_OFDM= 2 for QPSK modulation, and 4 for 16QAM modulation. The SH Frame is, as defined in EN 302 583 [1], clause 5.5.2.3, composed of 816 Capacity Units (CU) of 2 016 bit each, thus: Nb_CU_SHF_OFDM= 816 (C.6) Nb_Bit_CU=2 016 (C.7) C.2.1.2 TDM Frame The TDM signal symbol rate defined in EN 302 583 [1], clause 5.6.3 is calculated with the following formula:   1 + GI   BW _ TDM Sr _ TDM = int 32   (C.8)   1 + α   32(1 + GI ) with α being the TDM Square-Root Raised-Cosine (SRRC) filter roll off factor; α =0,15; 0,25; or 0,35, GI being the OFDM guard interval, BW_TDM being the TDM signal bandwidth with BW_TDM=BW_OFDM.NOTE: the DVB-SH waveform specification does not preclude the use of unequal bandwidth between TDM and OFDM modulations. However this usage is not allowed and left FFS. NOTE: The rationale of this equation is that the TDM symbol rate may be simply derived from the OFDM elementary period T, in order to optimize the receiver design. Using this definition as well as (1) and (2): ETSI 535 ETSI TS 102 584 V1.1.2 (2010-XY)   1 + GI    7  Sr _ TDM ⋅ T = int 32       1 + α    8*32(1 + GI )  result is a rational number, whatever the GI and α values. Thus Sr_TDM may be easily generated in any case from the 1/T clock reference. As defined in EN 302 583 [1], clause 5.6.4, the TDM SH Frame is composed of Physical Layer Slots (PL SLOTS) of Nb_Symb_TDM_SLOT symbols: Nb_Symb_TDM_SLOT=2 176 (C.9) Each PL SLOT carries 2, 3, or 4 CU, equal to the order of modulation BpS_TDM (see EN 302 583 [1], clause 5.6.4.1): Nb_CU_PLSLOT=BpS_TDM =2, 3 or 4 (C.10) C.2.2 Calculation of the SH Frame duration The SH frame length is defined as a function of the length of OFDM frames. Therefore, the frame length durations of both an SH-frame in OFDM and TDM modes are identical. The number of OFDM Frames per SH frame is defined in the table 5.10 of the waveform definition [1]. It may be calculated from the equations (C.3) to (C.7) from above. The related equation is the following: Nb _ CU _ SHF _ OFDM ⋅ Nb _ Bit _ CU Nb _ OFF _ SHF = BpS _ OFDM ⋅ Nb _ Data _ Carrier ⋅ Nb _ Symb _ OFDM _ Frame where the numerator is the number of data bits per SH Frame, and the denominator is the number of bits per OFDM Frame. Using (3)-(7), the above equation reduces to: 32 Nb _ OFF _ SHF = (C.11) Mod _ FFT ⋅ BpS _ OFDM So, the equation of the SH Frame duration is the following: TSHF = Nb _ OFF _ SHF ⋅ Nb _ Symb _ OFDM _ Frame ⋅ TS (C.12) Using (2), (3) and (8), it reduces to: 1 + GI TSHF = 2176 * 896 ⋅ , (C.13) BpS _ OFDM ⋅ BW _ OFDM with TSHF in S. C.2.3 Calculation of the number of MPEG TS packets per SH Frame Each turbo code word contains exactly 8 MPEG TS Packets. Thus, the number of MPEG TS packets per SH Frame is: Nb _ TP _ SHF = Nb _ TP _ CW ⋅ Nb _ CW _ SHF , (C.14) with Nb_CW_SHF being the number of FEC code words per SH frame, and Nb_TP_CW being equal to 8. Following FEC coding and rate adaptation, the FEC codeword length in bits is: Nb _ Bit _ CW _ 0 Nb _ Bit _ CW = , cr ETSI 536 ETSI TS 102 584 V1.1.2 (2010-XY) with Nb_Bit_CW_0 = 12 096, that is the codeword length assuming a code rate of 1, and cr being the nominal FEC code rate, as defined in EN 302 583 [1], clause 5.3.1. The number of code word per SH Frame is related to the number of CU per SH Frame:  (Nb _ CU _ SHF − 3)  Nb _ CW _ SHF = int   Nb _ Bit _ CW / Nb _ Bit _ CU     with Nb_CU_SHF being the number of CU per SH Frame, int(x) being a function which returns the largest integer less than the value x.  ( Nb _ CU _ SHF − 3) ⋅ Nb _ Bit _ CU ⋅ cr  Nb _ CW _ SHF = int   ,  (C.15)  Nb _ Bit _ CW _ 0  Nb _ Bit _ CW _ 0 NOTE: simplifies to 6 (a codeword occupies 6/cr CUs). Nb _ Bit _ CU For OFDM, as defined in (6), the number of CU per SH frame is: Nb _ CU _ SHF _ OFDM = 816 (C.16) For TDM, the number of CU per SH frame is given in EN 302 583 [1], clause 5.6.1. It is obtained from the following calculation: Nb _ CU _ PLSLOT Nb _ CU _ SHF _ TDM = Sr _ TDM ⋅ TSHF ⋅ (C.17) Nb _ Symb _ TDM _ SLOT Using (8), (9), (10) and (13), it reduces to the following expression:  1 + GI  BpS _ TDM Nb _ CU _ SHF _ TDM = 28 ⋅ int  32 ⋅ (C.18)  1 + α  BpS _ OFDM So, the number of MPEG TS packets per SH Frame is: for OFDM, using (14), (15) and (16): Nb _ TP _ SHF = 8 ⋅ int (136 ⋅ cr ) (C.19) for TDM, using (14), (15) and (18):  cr   1 + GI  BpS _ TDM  Nb _ TP _ SHF = 8 ⋅ int  . 28 ⋅ int  32 6  ⋅ − 3   (C.20)    1 + α  BpS _ OFDM  C.2.4 Calculation of the capacity An MPEG TS packet is 188 bytes long, that is Nb_Bit_TP bits. Nb_Bit_TP=1 504 (C.21) So the bit rate capacity at the MPEG-TS interface is: Nb _ TP _ SHF Cap _ TP = Nb _ Bit _ TP ⋅ [bps], (C.22) TSHF for OFDM, using (13), (19) and (22): ETSI 537 ETSI TS 102 584 V1.1.2 (2010-XY) 47 ⋅ 8 int(136.cr ) Cap _ TP _ OFDM = ⋅ ⋅ BpS _ OFDM ⋅ BW _ OFDM ; 2176 ⋅ 28 1 + GI for TDM, using (13), (20) and (22):  cr   1 + GI  BpS _ TDM  Cap _ TP _ TDM = 12032 ⋅ int  . 28 ⋅ int  32 6  ⋅ − 3  .     1 + α  BpS _ OFDM  ETSI 538 ETSI TS 102 584 V1.1.2 (2010-XY) History Document history V1.0 June 1 2010 First version approved by the DVB-TM ETSI