Advances in Communications Satellite Systems: Proceedings of the 37th International Communications Satellite Systems Conference (Icssc-2019) 9781839531453, 9781839531460

Citation preview

IET TELECOMMUNICATIONS SERIES 95

Advances in Communications Satellite Systems 2

Other volumes in this series: Volume 9 Volume 12 Volume 13 Volume 19 Volume 20 Volume 26 Volume 28 Volume 29 Volume 31 Volume 32 Volume 33 Volume 34 Volume 35 Volume 36 Volume 37 Volume 38 Volume 40 Volume 41 Volume 43 Volume 44 Volume 45 Volume 46 Volume 47 Volume 48 Volume 49 Volume 50 Volume 51 Volume 52 Volume 53 Volume 54 Volume 58 Volume 59 Volume 60 Volume 65 Volume 67 Volume 68 Volume 69 Volume 70 Volume 71 Volume 72 Volume 73 Volume 74

Phase Noise in Signal Sources W.P. Robins Spread Spectrum in Communications R. Skaug and J.F. Hjelmstad Advanced Signal Processing D.J. Creasey (Editor) Telecommunications Traffic, Tariffs and Costs R.E. Farr An Introduction to Satellite Communications D.I. Dalgleish Common-Channel Signalling R.J. Manterfield Very Small Aperture Terminals (VSATs) J.L. Everett (Editor) ATM: The broadband telecommunications solution L.G. Cuthbert and J.C. Sapanel Data Communications and Networks, 3rd Edition R.L. Brewster (Editor) Analogue Optical Fibre Communications B. Wilson, Z. Ghassemlooy and I.Z. Darwazeh (Editors) Modern Personal Radio Systems R.C.V. Macario (Editor) Digital Broadcasting P. Dambacher Principles of Performance Engineering for Telecommunication and Information Systems M. Ghanbari, C.J. Hughes, M.C. Sinclair and J.P. Eade Telecommunication Networks, 2nd Edition J.E. Flood (Editor) Optical Communication Receiver Design S.B. Alexander Satellite Communication Systems, 3rd Edition B.G. Evans (Editor) Spread Spectrum in Mobile Communication O. Berg, T. Berg, J.F. Hjelmstad, S. Haavik and R. Skaug World Telecommunications Economics J.J. Wheatley Telecommunications Signalling R.J. Manterfield Digital Signal Filtering, Analysis and Restoration J. Jan Radio Spectrum Management, 2nd Edition D.J. Withers Intelligent Networks: Principles and applications J.R. Anderson Local Access Network Technologies P. France Telecommunications Quality of Service Management A.P. Oodan (Editor) Standard Codecs: Image compression to advanced video coding M. Ghanbari Telecommunications Regulation J. Buckley Security for Mobility C. Mitchell (Editor) Understanding Telecommunications Networks A. Valdar Video Compression Systems: From first principles to concatenated codecs A. Bock Standard Codecs: Image compression to advanced video coding, 3rd Edition M. Ghanbari Digital Communications: Principles and Systems Ifiok Otung Dynamic Ad Hoc Networks H. Rashvand and H. Chao (Editors) Understanding Telecommunications Business A Valdar and I Morfett Advances in Body-Centric Wireless Communication: Applications and state-of-the- art Q.H. Abbasi, M.U. Rehman, K. Qaraqe and A. Alomainy (Editors) Managing the Internet of Things: Architectures, theories and applications J. Huang and K. Hua (Editors) Advanced Relay Technologies in Next Generation Wireless Communications I. Krikidis and G. Zheng 5G Wireless Technologies A. Alexiou (Editor) Cloud and Fog Computing in 5G Mobile Networks E. Markakis, G. Mastorakis, C.X. Mavromoustakis and E. Pallis (Editors) Understanding Telecommunications Networks, 2nd Edition A. Valdar Introduction to Digital Wireless Communications Hong-Chuan Yang Network as a Service for Next Generation Internet Q. Duan and S. Wang (Editors) Access, Fronthaul and Backhaul Networks for 5G & Beyond M.A. Imran, S.A.R. Zaidi and M.Z. Shakir (Editors)

Volume 76 Volume 77 Volume 78 Volume 79 Volume 80 Volume 81 Volume 83 Volume 84 Volume 86 Volume 89 Volume 91 Volume 92 Volume 93 Volume 95

Trusted Communications with Physical Layer Security for 5G and Beyond T.Q. Duong, X. Zhou and H.V. Poor (Editors) Network Design, Modelling and Performance Evaluation Q. Vien Principles and Applications of Free Space Optical Communications A.K. Majumdar, Z. Ghassemlooy, A.A.B. Raj (Editors) Satellite Communications in the 5G Era S.K. Sharma, S. Chatzinotas and D. Arapoglou Transceiver and System Design for Digital Communications, 5th Edition Scott R. Bullock Applications of Machine Learning in Wireless Communications R. He and Z. Ding (Editors) Microstrip and Printed Antenna Design, 3rd Edition R. Bancroft Low Electromagnetic Emission Wireless Network Technologies: 5G and beyond M.A. Imran, F. He´liot and Y.A. Sambo (Editors) Advances in Communications Satellite Systems Proceedings of the 36th International Communications Satellite Systems Conference (ICSSC-2018) I. Otung, T. Butash and P. Garland (Editors) Information and Communication Technologies for Humanitarian Services M.N. Islam (Editor) Green Communications for Energy-Efficient Wireless Systems and Networks Himal Asanga Suraweera, Jing Yang, Alessio Zappone and John S. Thompson (Editors) Flexible and Cognitive Radio Access Technologies for 5G and Beyond H. Arslan and E. Bas¸ar (Editors) Antennas and Propagation for 5G and Beyond Q. Abbasi, S.F. Jilani, A. Alomainy and M.A. Imran (Editors) ISDN Applications in Education and Training R. Mason and P.D. Bacsich

This page intentionally left blank

Advances in Communications Satellite Systems 2 Proceedings of the 37th International Communications Satellite Systems Conference (ICSSC-2019) Edited by Ifiok Otung, Thomas Butash and Tetsushi Ikegami

The Institution of Engineering and Technology

Published by The Institution of Engineering and Technology, London, United Kingdom The Institution of Engineering and Technology is registered as a Charity in England & Wales (no. 211014) and Scotland (no. SC038698). † The Institution of Engineering and Technology 2021 First published 2020 This publication is copyright under the Berne Convention and the Universal Copyright Convention. All rights reserved. Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act 1988, this publication may be reproduced, stored or transmitted, in any form or by any means, only with the prior permission in writing of the publishers, or in the case of reprographic reproduction in accordance with the terms of licences issued by the Copyright Licensing Agency. Enquiries concerning reproduction outside those terms should be sent to the publisher at the undermentioned address: The Institution of Engineering and Technology Michael Faraday House Six Hills Way, Stevenage Herts, SG1 2AY, United Kingdom www.theiet.org While the authors and publisher believe that the information and guidance given in this work are correct, all parties must rely upon their own skill and judgement when making use of them. Neither the authors nor publisher assumes any liability to anyone for any loss or damage caused by any error or omission in the work, whether such an error or omission is the result of negligence or any other cause. Any and all such liability is disclaimed. The moral rights of the authors to be identified as authors of this work have been asserted by them in accordance with the Copyright, Designs and Patents Act 1988.

British Library Cataloguing in Publication Data A catalogue record for this product is available from the British Library

ISBN 978-1-83953-145-3 (hardback) ISBN 978-1-83953-146-0 (PDF)

Typeset in India by MPS Limited Printed in the UK by CPI Group (UK) Ltd, Croydon

Contents

About the editors Acknowledgments

Communications satellite systems: retrospect and prospect Ifiok Otung, Thomas Butash and Tetsushi Ikegami Section 1: Broadband satellite communication architectures and applications 1 The results of WINDS experiments of NICT Tajkashi Takahashi, Naoko Yoshimura, Kaszuyoshi Kawasaki, Tomoshige Kan, Byeongpyo Jeong and Morio Toyoshima 1.1 1.2 1.3 1.4 1.5

Introduction WINDS WINDS experiments NICT fundamental experiments in the regular operational phase NICT fundamental experiments in postoperational phase 1.5.1 Function verification experiment of fully automatic transportable earth station 1.5.2 Function verification experiment of small-sized vehicle transportable earth station 1.5.3 Function verification experiment of small earth station for vessel 1.5.4 Satellite communication experiments for disaster countermeasures 1.5.5 Basic experiments for future satellite technology 1.6 Conclusion References 2 Report of 3.2 Gbps transmission experiment result using WINDS satellite Kenji Suzuki, Masatomo Yahata, Tetsuya Watanabe, Kenichi Hoshi, Tamio Okui, Yoshiki Arakawa, Toshio Asai, Tomoshige Kan, Takashi Takahashi and Morio Toyoshima 2.1

Introduction

xxxiii xxxv

1

9 11

11 12 13 14 14 14 16 18 19 19 22 22

25

25

viii

Advances in Communications Satellite Systems 2 2.2 2.3

16APSK/QAM-OFDM 3.2 Gbps transmitter and receiver WINDS satellite communication experiment: Uncompressed 4K UHDTV 2.4 Conclusion Acknowledgments References 3

Disaster countermeasure experiments using WINDS and the response of 2016 Kumamoto earthquakes Byeongpyo Jeong, Hajime Susukita, Tomoshige Kan, Kazuyoshi Kawasaki and Takashi Takahashi 3.1 3.2 3.3

Introduction Development of mobile vehicle earth station for WINDS Development of application based on mobile vehicle earth station for WINDS and demonstration experiment of disaster countermeasures 3.3.1 Development of detection system of the road bump based on the mobile vehicle earth station for WINDS 3.3.2 Mobile communication experiment between fire vehicles and WINDS mobile vehicle earth station 3.3.3 Speech quality evaluation of voice communication system using WINDS 3.4 Open experiment of disaster countermeasure 3.4.1 Open experiment to build an emergency communication network by the cooperation of WINDS and a small unmanned aircraft on “Sanuki medical rally” and to collect and convey information 3.4.2 Open experiment of the medical activity training at a large-scale earthquake in the government’s comprehensive disaster preparation drill in 2016 3.4.3 Open experiment on the 2016 comprehensive disaster preparation drill of Ehime Prefecture 3.5 Response of 2016 Kumamoto earthquakes: building and operating emergency network based on WINDS 3.5.1 Purpose of dispatch of NICT 3.5.2 Emergency network building using WINDS, and usage conditions of the network 3.6 Conclusion References 4

26 28 33 33 33

35

35 36

37 37 40 41 44

44

45 47 47 48 49 51 51

Satellite integrated communication system for marine robots operations Hiroshi Yoshida

53

4.1

53

Introduction

Contents 4.2

Past trials 4.2.1 Satellite remote control of an ROV 4.2.2 Data uploading and mission downloading of an AUV via the satellite 4.2.3 Wideband real-time data transmission 4.2.4 ASV controls with a satellite 4.3 Artificial satellites and marine robot 4.3.1 Operation basis of marine robot 4.3.2 Challenge to a new paradigm where robots play the leading role in ocean business 4.4 Conclusion References 5 Evolution of Ka-band on-the-move terminals for land and maritime broadband communications Marshall Lewis, John Ness, John Logan, Tomoshige Kan, Takashi Takahashi, Naoko Yoshimura and Morio Toyoshima 5.1 5.2

Introduction Methodology 5.2.1 Monopulse tracking 5.2.2 Linear polarization 5.2.3 Heat and power consumption reduction 5.3 Results 5.3.1 Monopulse tracking 5.3.2 Linear polarization 5.3.3 Power consumption and heat transfer 5.3.4 Terminal operation results 5.4 Conclusion Acknowledgments References

ix 54 55 57 57 59 59 59 61 62 62

65

65 66 66 66 70 70 70 71 72 73 76 76 76

Section 2: Integrated applications and architectures for vessels and IoT

77

6 Mega-constellations as enabler of autonomous shipping Anastasia Yastrebova, Marko Ho¨yhtya¨ and Mikko Majanen

79

6.1 6.2 6.3

6.4

Introduction System model and requirements 6.2.1 System requirements Use cases for autonomous ship-based LEO satellite connectivity 6.3.1 Continuous satellite connectivity use case: collision avoidance with remote control and AR support 6.3.2 Discontinuous satellite connectivity use case: fully autonomous monitoring operation Design methodology

79 81 82 82 83 84 85

x

7

Advances in Communications Satellite Systems 2 6.4.1 Minimal number of satellites for the continuous coverage 6.4.2 Optimal number of satellites 6.5 Results 6.5.1 Coverage 6.5.2 Satellite visibility 6.5.3 Latency 6.5.4 Link duration 6.6 Conclusion Acknowledgments References

85 87 88 89 91 91 92 93 93 93

On decoding strategies for satellite uplink asynchronous random access channels Farbod Kayhan and Nader Alagha

95

7.1 7.2

8

9

Introduction System model 7.2.1 Coding scheme selection 7.3 Decoding strategies in the presence of interference 7.3.1 System model and simulation results 7.4 Physical layer abstraction 7.5 Conclusions and future studies Acknowledgments References

95 97 98 98 98 100 102 102 102

Software-defined radio implementation of UHF RFID tags in 5G Internet of things application Gursajan Singh

105

8.1 Introduction 8.2 Methodology 8.3 Results 8.4 Conclusion 8.5 Future work Acknowledgments References

105 106 108 111 112 113 113

Energy-efficient user terminals for Internet of things applications over satellite Uyen L. Dang, Henrik Zessin, Christian Rohde and Florian Leschka

115

9.1 9.2

Introduction Energy-efficient user terminal 9.2.1 Energy-efficient transmission

115 117 117

Contents 9.2.2 Power supply 9.2.3 Intelligent energy and system management 9.3 Conclusion Acknowledgments References Section 3: DTN and HTS technologies 10 Rising above the cloud: toward high-rate delay-tolerant networking in low earth orbit Alan Hylton, Daniel Raible, Gilbert Clark, Rachel Dudukovich, Brian Tomko and Laura Burke 10.1 Introduction 10.1.1 Network model of optical communications 10.1.2 Experiment network 10.2 Delay-tolerant networking 10.2.1 Interoperability of software components 10.3 Scenario: generating a connectivity model 10.4 High-rate delay-tolerant networking 10.4.1 Introduction 10.4.2 Components 10.4.3 Interconnect 10.4.4 System flow 10.5 Test discussion 10.5.1 Networking test results 10.6 Conclusion Acknowledgments References 11 Integration of high-performance V-band GAN MMIC HPA for the QVlift project Alejandro Rodrı´guez, Jaime Cagigas, Noelia Santos, Joe¨l Moron, Fabio Vitobello, Giorgia Parca, Giuseppe Valente and Giuseppe Codispoti 11.1 Introduction 11.2 Solid state GaN power amplifier design 11.3 HPA measurement test setup and results 11.3.1 Output power and gain compression of drivers 11.3.2 Insertion loss measurements 11.3.3 HPA on-wafer measurements 11.3.4 HPA preliminary power measurements 11.4 Conclusions Acknowledgments References

xi 122 125 126 127 127 129 131

132 133 133 134 136 137 140 140 140 141 142 143 144 145 145 145

149

149 150 150 152 152 156 156 158 159 159

xii

Advances in Communications Satellite Systems 2

12 Performance study of frequency flexibility in high-throughput satellites and its contribution to operations strategy Yuma Abe, Mitsugu Okawa, Amane Miura, Kazunori Okada, Maki Akioka and Morio Toyoshima 12.1 Introduction 12.2 Summary of previous results 12.2.1 System configuration 12.2.2 Frequency assignment method 12.2.3 Previous evaluation results 12.3 Comprehensive evaluation plan 12.3.1 Performance index 12.3.2 Communication traffic 12.3.3 Link assignment schemes 12.4 Operations strategy for HTS system 12.5 Conclusion Acknowledgments References Section 4: New satellite system architectures and components 13 Satellite experiments on direct spectrum division transmission over multiple transponders Fumihiro Yamashita, Daisuke Goto, Yasuyoshi Kojima, Hiroki Shibayama, Hiroyuki Kobashi and Daiki Haraguchi 13.1 Introduction 13.2 Operational principle of DSDT adapter 13.2.1 Spectrum-editing technique in transmitter 13.2.2 Synchronization technique in receiver 13.3 Satellite experiments 13.3.1 Experimental setup 13.3.2 Spectrum division over multiple transponders 13.4 Summary References 14 User terminal wideband modem for very high throughput satellites Steven Kisseleff, Nicola Maturo, Symeon Chatzinotas, Helge Fanebust, Bjarne Rislow, Kimmo Kansanen, Matthieu Arzel and Hans C. Haugli 14.1 Introduction 14.2 System model 14.2.1 Challenges and impairments 14.2.2 Requirements 14.3 Modem design 14.3.1 Architecture 14.3.2 Timing synchronization

161

161 163 163 164 164 165 165 166 166 167 169 169 170 171 173

173 174 174 174 177 177 178 184 184 187

188 189 189 190 191 191 192

Contents 14.3.3 Frequency synchronization 14.3.4 Frame synchronization 14.3.5 Equalization 14.3.6 Demodulation and decoding 14.4 Numerical results 14.5 Conclusion Acknowledgments References 15 Licensed shared access testbed for integrated satellite-terrestrial communications: the ASCENT project Marko Ho¨yhtya¨, Mikko Majanen, Mika Hoppari, Pertti Ja¨rvensivu, Heikki Kokkinen, Arto Reis-Kivinen, Jaakko Ojaniemi, Olivier Pellay and Duc Pham Minh 15.1 Introduction 15.2 Sharing use cases 15.2.1 Sharing the 5G pioneer bands 15.2.2 Sharing terrestrial IMT bands 15.3 Testbed architecture 15.3.1 Architecture for sharing the 5G pioneer bands 15.3.2 Architecture for sharing terrestrial IMT bands 15.4 Performance evaluations 15.4.1 Evacuation and frequency change times 15.4.2 Scalability of the testbed 15.4.3 Proof of concept of IMT frequency bands sharing 15.5 Conclusions Acknowledgments References 16 Cognitive communications for NASA space systems David Chelmins, Janette Briones, Joseph Downey, Gilbert Clark and Adam Gannon 16.1 16.2 16.3 16.4

Introduction Defining cognition Focus areas Cognitive links 16.4.1 Radio frequency interference mitigation 16.4.2 Radio link optimization 16.4.3 Automatic receiver configuration 16.4.4 Deep learning communication links 16.5 Cognitive networks 16.5.1 Delay-tolerant networking 16.5.2 Intelligence in the DTN architecture 16.5.3 Cognition in the DTN protocols

xiii 194 196 198 199 200 202 203 203

205

206 206 206 206 207 207 208 208 210 213 213 217 217 218 219

220 221 222 223 223 224 224 225 225 226 226 227

xiv

Advances in Communications Satellite Systems 2 16.5.4 Legacy, infrastructure, and bootstrapping intelligence 16.5.5 Virtualization in future cognitive networks 16.6 Cognitive systems 16.6.1 User-initiated service 16.6.2 System-wide intelligence 16.7 Enabling technology 16.7.1 Reconfigurable hardware 16.7.2 Cognitive processing challenges 16.8 Conclusion References

17 Supporting NASA Artemis 1 mission with JAXA Uchinoura station Timothy Pham, Hiroshi Takeuchi and Atsushi Tomiki 17.1 Introduction 17.2 Operation concept 17.3 Recording/playback test equipment 17.4 Result of Artemis compatibility test 17.5 Result of LRO spacecraft tracking 17.6 Conclusion Acknowledgments References

228 228 229 229 230 231 231 231 232 232

235 236 236 237 239 240 241 241 242

Section 5: High-speed optical communications and feeder links 1

243

18 Implementation of the method for estimating propagation direction of laser beam transmitted from ground to satellite Hiroki Yamashita and Yoshihisa Takayama

245

18.1 Introduction 18.2 Estimating beam propagation direction using dual boresight camera 18.3 Experimental method to measure the angular error 18.3.1 Experimental method 18.3.2 Geometrical angular error 18.3.3 The procedure of the experiment 18.3.4 Results 18.4 Conclusion References 19 Studies on site diversity to mitigate cloud blockage in satellite-ground optical communications based on long term ground meteorological observation data Yuki Ueda, Tatsuya Mukai and Yoshihisa Takayama 19.1 Introduction

245 246 248 248 249 249 251 253 253

255 255

Contents 19.2 Methodology 19.2.1 Investigation of cloud amount data in Japan 19.2.2 Domestic candidate locations analyzed by ground-based cloud amount data in Japan 19.2.3 Domestic candidate locations further analyzed by the installation conditions 19.2.4 Domestic candidate locations further analyzed by their uncorrelation relations 19.3 Results 19.4 Conclusion Acknowledgments References 20 Overview of optical ground systems developments for network switching controls to avoid cloud blockage in space optical direct communications Tatsuya Mukai, Yoshihisa Takayama and Tomohiro Araki 20.1 Introduction 20.2 Methodology 20.2.1 Laser ground network planning system 20.2.2 Optical ground stations 20.2.3 Infrared cloud monitoring and discrimination system 20.2.4 Network switching testings 20.3 Results 20.4 Conclusion Acknowledgments References 21 Demonstration of high-speed pixelated acquisition and tracking system for optical intersatellite links Alan Scott, Danya Hudson, Hugh Podmore and Elliott Coleshill 21.1 Introduction 21.2 Pixelated versus quad cell trades 21.2.1 Angular resolution versus range 21.2.2 Environmental factors 21.2.3 Trade summary 21.3 Test results 21.3.1 Closed loop testing 21.3.2 Radiation testing 21.4 Conclusions Acknowledgments References

xv 258 258 259 262 263 267 268 268 268

269 270 271 271 273 275 278 281 282 282 282

283 284 285 285 286 288 288 288 288 289 289 290

xvi

Advances in Communications Satellite Systems 2

22 An experimental study of RF optical transformation function Yutaka Oshima, Takuo Goda, Hiroaki Miyoshi, Masahide Hamamoto and Masaki Tanaka 22.1 Introduction 22.2 Methodology 22.2.1 Experimental study plan 22.2.2 Experimental schedule 22.3 Results 22.3.1 Trade-off study of transformation function 22.3.2 Partial model production 22.3.3 Analysis of atmospheric turbulence impact 22.3.4 Partial model evaluation 22.4 Conclusion Acknowledgments References Section 6: Advanced digital payloads and components 23 Beam-hopping system configuration and terminal synchronization schemes Christian Rohde, Doron Rainish, Avraham Freedman, Guy Lesthievent, Nader Alagha, Danielle Delaruelle, Gerhard Mocker and Xavier Giraud

291

291 292 292 294 295 295 296 297 297 297 297 297 299 301

23.1 Introduction 23.2 Beam-hopping system considerations 23.2.1 BH scenarios 23.2.2 Operation strategies 23.2.3 Beam hopping system deployment 23.2.4 Control channel and cell ID considerations 23.2.5 DVB-S2X waveforms 23.3 Terminal synchronization schemes 23.3.1 Bursty data reception 23.3.2 Start of super-frame detection 23.3.3 Enhanced super-frame detection for Format 5 23.3.4 Enhanced super-frame detection for Format 6 23.4 Conclusions References

301 303 303 303 304 306 307 308 308 309 310 311 311 312

24 Adaptive coding and modulation (ACM) and power control scheme for return link of DVB-RCS2 satellite system Dong-Hyun Jung, Min-Su Shin and Joon-Gyu Ryu

315

24.1 Introduction 24.2 System model 24.3 ACM and power control 24.3.1 Requirements

315 316 318 318

Contents 24.3.2 Problem formulation 24.3.3 Proposed scheme 24.4 Simulation results 24.5 Conclusion Acknowledgments References

xvii 318 319 319 323 323 323

25 A study of frequency utilization efficiency of OFDMA adaptive coding and modulation on Ka-band satellite communications system 325 Mitsugu Ohkawa, Hiromitsu Wakana and Amane Miura 25.1 Introduction 25.2 Satellite link model 25.3 OFDMA 25.3.1 Outline of OFDMA 25.3.2 Application to ACM 25.4 DVB-S2X 25.4.1 Outline of DVB-S2(S2X) 25.4.2 Modulation and coding for DVB-S2(S2X) standards 25.5 Improvement of frequency utilization efficiency by ACM 25.5.1 Adaptive algorithm 25.5.2 Improvement of frequency utilization efficiency 25.6 Conclusion Acknowledgments References 26 Antenna pattern evaluation formed by reconfigurable antennas with the configuration of an array-fed reflector Yoshio Inasawa, Hitomi Ono, Masaaki Kusano, Arimasa Kanasashi, Toshiyasu Tsunoda, Terumi Sunaga, Nobuyoshi Horie and Eiichi Sakai 26.1 Introduction 26.2 Antenna pattern evaluation 26.3 Conclusion Acknowledgments References

325 326 328 328 328 330 330 330 330 330 336 340 340 340

343

343 344 346 346 346

27 Gallium nitride MMIC power amplifier for use in Ka-band HTS applications 349 Jim Sowers, Ghislain Turgeon, Rabindra (Rob) Singh and Hampton Chan 27.1 27.2 27.3 27.4

Introduction GaN PA MMICs GaN PA breadboard modules Future efforts

350 350 353 355

xviii

Advances in Communications Satellite Systems 2

27.5 Summary Acknowledgments References

355 358 359

Section 7: High-speed optical communications and feeder links 2

361

28 Technological trends and future prospects of satellite communications for mega-constellations with small satellites Morio Toyoshima

363

28.1 28.2 28.3 28.4

Introduction Trends in mega-constellation programs Frequency map for mega-constellations Frequency allocation for mega-constellations 28.4.1 Definition of a mega-constellation 28.4.2 Date of bringing into use 28.4.3 Short-duration mission 28.5 Conclusion References 29 Commercial communications satellites in the post-2020 era Bruce R. Elbert Nomenclature 29.1 Introduction 29.2 New and evolving GEO systems 29.2.1 High-throughput satellites 29.2.2 GEO system strategies going forward 29.2.3 How these strategies will change 29.3 Non-GEO systems in development 29.3.1 Fully interconnected processor-based LEOs 29.3.2 Issues with broadband LEO constellations 29.4 Assessing the near-term and long-term communications satellites 29.5 Conclusion References 30 5G and beyond for new space: vision and research challenges Marko Ho¨yhtya¨, Marius Corici, Stefan Covaci and Maria Guta 30.1 Introduction 30.2 Use cases and some high-level requirements 30.2.1 Communications on the move 30.2.2 Public safety 30.3 Network architecture 30.3.1 Terrestrial layer 30.3.2 Airborne layer 30.3.3 Space layer

363 364 365 367 367 367 368 369 369 371 372 372 373 373 375 377 377 378 380 383 384 386 387 387 388 388 388 389 389 391 391

Contents 30.4 Research challenges 30.4.1 Physical layer and MAC procedures 30.4.2 Software networks and mobile edge computing 30.4.3 Mobility and routing 30.4.4 High-frequency bands 30.4.5 Spectrum sharing and interference management 30.4.6 Optical communications 30.4.7 Quantum communications 30.4.8 End-to-end cybersecurity 30.5 Autonomous systems as future disruption 30.5.1 Software-defined satellites 30.5.2 Autonomous transport 30.5.3 Autonomous satellites 30.6 Innovative and ambitious missions 30.7 Conclusion Acknowledgments References 31 Direction of Satcom R&D in Japan: WINDS, ETS-IX, and beyond Naoto Kadowaki 31.1 31.2 31.3 31.4

Introduction WINDS ETS-IX Direction of Satcom R&D 31.4.1 Satellite communications in beyond 5G networks 31.4.2 Fundamental technology development for satellite networks in the future 31.5 Conclusion References

xix 392 393 393 394 394 395 396 396 396 398 398 398 398 399 400 400 400 403 403 404 406 407 407 408 409 409

Section 8: Satellite antenna technologies

411

32 Development of highly maintainable and reliable RF transceiver for satellite base stations Munehiro Matsui, Akira Matsushita and Fumihiro Yamashita

413

32.1 Introduction 32.2 Basic configuration of RF transceiver 32.3 C-band amplifier for remote island satellite communications 32.4 Ku-band amplifier for disaster-relief satellite communications 32.5 Performance evaluation 32.6 Conclusions References

413 415 417 419 420 424 424

xx

Advances in Communications Satellite Systems 2

33 Fan-fold Ka-band large reflector and its applications Kentaro Nishi, Satoru Ozawa, Kazuyuki Nakamura and Nobuko Nakamura 33.1 Introduction 33.2 Design concept 33.2.1 Fan-fold deployable structure 33.2.2 Rhombus lattice thin-plate network 33.3 Surface accuracy evaluation of rhombus lattice thin-plate network 33.3.1 Number of division and side length 33.3.2 Surface accuracy estimation by structural analyses 33.4 Applications of fan-fold deployable reflector 33.5 Conclusion References 34 The reduction of measurement point for self-calibration method of systematic errors for DBF antenna using gating process Takuya Okura, Amane Miura, Teruaki Orikasa and Shinji Senba 34.1 Introduction 34.2 Calibration system 34.3 Measurement result using pickup antenna 34.4 Radiation patterns 34.5 Conclusion References 35 Calibration method for array antenna considering mutual coupling in mobile satellite communications Hiroyuki Tsuji, Yuma Abe, Tomotada Kondo, Yuiko Kikuchi and Shuichi Adachi 35.1 35.2 35.3 35.4

Introduction Array antenna model for calibration Array antenna calibration method considering mutual coupling Numerical evaluations 35.4.1 Simulation conditions 35.4.2 Simulation results 35.5 Conclusion Acknowledgments References

427

427 428 430 432 432 432 433 435 438 438

441 442 442 445 446 448 450

451

451 452 454 456 456 456 457 458 458

Contents Section 9: Propagation and modeling for satellite communications 36 Analysis of the impact of turbulence on adaptive optics ground station performance Kenneth J. Grant, Kerry A. Mudge, Bradley A. Clare, Francis Bennet, Brett D. Nener and Dilusha Silva 36.1 Introduction 36.1.1 Atmospheric scintillation 36.1.2 Atmospheric coherence length 36.2 Optical turbulence measurements 36.2.1 Boundary layer Cn2 measurements 36.2.2 Estimated atmospheric coherence length 36.3 Adaptive optics simulations 36.4 Numerical propagation modeling 36.4.1 Numerical phase screen model 36.4.2 PS simulation results 36.5 Conclusion References 37 A 40-year cloud climatological study for Australia: implications for siting of laser communication infrastructure Helen C. Chedzey, David E. Herne, Mervyn J. Lynch, Brett D. Nener, Kenneth J. Grant, Kerry A. Mudge and Bradley A. Clare 37.1 Introduction 37.1.1 Australian cloud climatology studies 37.1.2 The 40-year satellite climatology record 37.1.3 The AVHRR instrument 37.1.4 The cloud detection scheme employed by PATMOS-x 37.2 Climate change and decadal trends in cloud statistics 37.3 Discussion of results 37.4 Conclusions Acknowledgments References 38 Experimental results of seasonal vegetation changes on data transmission for Ka-band mobile satellite communication Tomoshige Kan, Byeongpyo Jeong, Hajime Susukita, Kazuyoshi Kawasaki, Takashi Takahashi and Morio Toyoshima 38.1 Introduction 38.2 Summary of WINDS and the WINDS vehicle earth station 38.2.1 WINDS 38.2.2 WINDS vehicle earth station 38.3 Measurement experiments 38.3.1 Experimental site

xxi 459 461

461 462 462 464 464 464 466 467 468 468 470 472

475

476 477 478 478 478 479 479 487 487 488

491

492 492 492 493 495 495

xxii

Advances in Communications Satellite Systems 2

38.3.2 Measurement system and experimental method 38.4 Results and discussion 38.5 Conclusion References 39 Experimental study of external interference for LEO-based automatic identification system (AIS) Daichi Hirahara, Toshiyuki Nishibori, Toshiyoshi Kimura, Shuji Shimizu, Junichiro Ishizawa, Shinichi Sobue, Satoko Miura and Shinichi Suzuki 39.1 Introduction 39.1.1 SPace-based AIS experiment (SPAISE) 39.2 External interference 39.2.1 Separating AIS messages and external interference 39.3 Experimental results 39.4 Conclusion References Section 10: Future technologies for 5G and beyond 40 Advanced demonstration plans of high-speed laser communication “HICALI” mission onboard the engineering test satellite 9 Yasushi Munemasa, Yoshihiko Saito, Alberto Carrasco-Casado, Dimitar R. Kolev, Phuc V. Trinh, Hideki Takenaka, Kenji Suzuki, Toshihiro Kubo-oka, Tetsuharu Fuse, Hiroo Kunimori, Koichi Shiratama, Yasushiro Takahashi and Morio Toyoshima 40.1 Introduction 40.2 Overview of HICALI project 40.3 Demonstration plans of HICALI project 40.4 Conclusion Acknowledgments References

495 496 501 502

503

503 504 504 506 506 510 511 513 515

516 517 519 520 521 521

41 Optical communication experiment with microsatellite body-pointing using VSOTA on RISESAT 523 Hideki Takenaka, Hiroo Kunimori, Toshinori Kuwahara, Yuji Sakamoto, Shinya Fujita, Homio Tomio, Morokot Sakal, Junichi Kurihara, Toshihiro Kubo-oka, Tetsu Fuse and Morio Toyoshima 41.1 Introduction 41.2 Component of VSOTA 41.3 Initial experiment 41.3.1 Checkout of VSOTA 41.3.2 Tracking accuracy from satellite 41.3.3 Optical tracking from OGS

523 524 525 525 527 527

Contents 41.4 Conclusion Acknowledgments References 42 Research and development of an optical ground station supporting both GEO- and LEO-to-ground links Dimitar R. Kolev, Koichi Shiratama, Hideki Takenaka, Alberto Carrasco-Casado, Yoshihiko Saito and Morio Toyoshima

xxiii 528 528 528

531

42.1 Introduction 42.2 Compatibility with different missions in the proposed solution 42.2.1 Beacons and uplink 42.2.2 Optical bench changes for different missions 42.3 Developed solution 42.3.1 Uplink precompensation for ground-to-GEO missions 42.3.2 Final solution and laboratory test results 42.4 Conclusion References

531 532 532 534 535 535 538 539 539

43 Optical observations of nonoperational satellites in graveyard orbits Michelle K Turberfield, Tetsuharu Fuse and Toshihiro Kubo-oka

541

43.1 Introduction 43.2 Definition of graveyard orbit 43.3 Methodology 43.3.1 Optical system 43.3.2 Selecting satellites 43.4 Image processing 43.4.1 Image processing with IRAF 43.5 Observation 43.5.1 Observation 43.6 Photometry 43.6.1 Photometry with IRAF 43.7 Conclusion Acknowledgments References Section 11: Flexible HTS systems and advanced digital payloads 44 Development of Ka-band digital beam forming antenna payload for the engineering test satellite-9 Eiichi Sakai, Yoshio Inasawa, Masaaki Kusano, Hitomi Ono, Arimasa Kanasash, Nobuyoshi Horie, Terumi Sunaga and Toshiyasu Tsunoda 44.1 Introduction 44.2 R&D activities

541 543 543 543 544 545 545 545 545 546 546 547 549 550 551 553

553 554

xxiv

Advances in Communications Satellite Systems 2

44.2.1 Subject-A: system design and comprehensive evaluation 44.2.2 Subject-C: development of antenna/RF for DBF 44.3 Conclusion Acknowledgments References 45 The initial study of calibrating receiving digital beam forming in engineering test satellite-9 Hitomi Ono, Eiichi Sakai, Yoshio Inasawa, Masaaki Kusano, Arimasa Kanasashi, Nobuyoshi Horie, Terumi Sunaga and Toshiyasu Tsunoda 45.1 Introduction 45.1.1 System configuration of DBF 45.1.2 Difficulty of DBF 45.2 Method of calibration 45.2.1 Detection of gain/phase error between elements 45.2.2 Calibration method with ground station 45.3 Conclusion Acknowledgments References 46 Beam pattern optimization based on up/downlink information for multibeam satellite communication systems Kazuma Kaneko, Shigenori Tani, Shigeru Uchida and Hiroshi Aruga 46.1 Introduction 46.2 Related research on HTS resource management 46.3 Beam pattern optimization 46.3.1 Genetic algorithm 46.3.2 Calculation of throughput 46.4 Performance evaluation 46.4.1 Evaluation model 46.4.2 Evaluation results 46.5 Conclusion References Section 12: Satellite networks design challenges and applications 47 Channel state modeling and performance evaluation of DVB-S2X based broadband land mobile satellite communication systems Burak Unal, Abdulkareem Karasuwa, Ashraf Ali, Nicholas Avlonitis, Jonathan Rodriguez and Ifiok Otung 47.1 Introduction 47.1.1 The DVB-S2 generations

554 557 558 558 558

559

559 560 560 561 562 563 565 565 565

567

567 568 570 570 572 573 573 575 579 579 581 583

584 585

Contents 47.2 Mobile satellite channel 47.3 Simulation scenarios 47.4 Results and discussion 47.5 Conclusion References 48 Impact of antenna and propagation models on coexistence of 5G and fixed satellite services Sahana Raghunandan, Christian Rohde and Jeffrey H. Reed 48.1 Introduction 48.2 System model 48.2.1 Antenna models 48.2.2 Propagation models 48.2.3 Interference calculation 48.3 Simulation setup 48.4 Interference power maps 48.4.1 Single interferer 48.4.2 Multiple BS transmissions 48.4.3 Multiple UE transmissions 48.5 Conclusion References 49 Integrated space-enabled hybrid 5G-V2X communications link modeling Solomon Udeshi, Mfonobong Uko, Muazzam Zafar, Arslan Altaf, Bamidele Adebisi and Sunday Ekpo 49.1 Introduction 49.2 Existing V2X communication networks and architectures 49.2.1 DSRC topologies 49.2.2 C-V2X topologies 49.3 5G infrastructure 49.4 Proposed hybrid DSRC-cellular 5G V2X platform overview 49.5 V2X link budget analysis 49.5.1 Signal attenuations 49.5.2 Noise floor and SNR analysis 49.6 Layer 1: DSRC link analysis 49.6.1 Case 1: V2V stationary 49.6.2 Case 2: V2V stationary and dynamic 49.6.3 Analysis of the effects of vehicle motion on link performance 49.7 Layer 3: Integrated space-enabled vehicle to satellite communication 49.8 Conclusion Acknowledgments References

xxv 587 587 592 594 595

597 597 599 599 601 601 603 606 606 609 613 613 613

615

616 617 617 618 619 619 620 621 621 622 622 622 626 626 628 630 630

xxvi

Advances in Communications Satellite Systems 2

50 K/Ka-band transceiver sensitivity modeling and link characterization for integrated 5G-LEO communication applications Mfonobong Uko, Muazzam Zafar, Arslan Altaf, Solomon Udeshi, Sunday Ekpo and Bamidele Adebisi 50.1 50.2 50.3 50.4

Introduction 5G link characterization 5G mmWave link budget for a K/Ka-band transceiver Sensitivity modeling for integrated 5G-LEO communication applications 50.4.1 Transmitter front-end modeling 50.4.2 Receiver front-end modeling 50.5 Simulation result and analysis 50.5.1 Transmitter front-end analysis 50.5.2 Receiver front-end analysis 50.5.3 5G NR receiver sensitivity modeling 50.6 Conclusion Acknowledgments References 51 Link budget design for integrated 5G-LEO communication applications Mfonobong Uko, Muazzam Zafar, Arslan Altaf, Solomon Udeshi, Sunday Ekpo and Bamidele Adebisi 51.1 Introduction 51.2 5G-LEO RF link budget design and calculation 51.2.1 Received power determination 51.2.2 Path loss modeling 51.3 5G mmWave link budget for a Ka-band satellite 51.4 Simulation result and analysis 51.5 Conclusion Acknowledgments References Section 13: New satellite components and transmitter and modem technologies 52 Secret key agreement for satellite laser communications Hiroyuki Endo and Masahide Sasaki 52.1 Introduction 52.2 Secret key agreement 52.3 Channel model 52.3.1 Generalized on-off keying 52.3.2 Secret key rate for GOOK

633

633 635 636 637 638 638 638 638 639 640 641 642 643

645

645 646 646 647 648 650 652 652 652

655 657 657 659 660 660 660

Contents 52.4 Numerical investigation of secret key rate 52.5 FSO-SKA versus QKD 52.6 Conclusion Acknowledgments References 53 Methods for securing spacecraft tasking and control via an enterprise Ethereum blockchain David Hyland-Wood, Peter Robinson, Roberto Saltini, Sandra Johnson and Christopher Hare 53.1 Introduction 53.2 Literature review 53.3 Methodology 53.4 Results 53.5 Conclusion Acknowledgments References 54 PAPR reduction and digital predistortion for 5G waveforms in digital satellite payloads Ovais Bin Usman, Thomas Delamotte and Andreas Knopp 54.1 Introduction 54.2 System model 54.3 PAPR reduction and predistortion method 54.3.1 HPA model 54.3.2 PAPR reduction 54.3.3 Digital predistortion 54.4 Simulations results 54.4.1 Analysis with only DPD 54.4.2 Analysis with only clipping 54.4.3 Analysis with clipping and DPD 54.4.4 Total degradation analysis 54.5 Conclusion References 55 Effects of differential oscillator phase noise in precoding performance Liz Martı´nez Marrero, Juan C. Merlano Duncan, Jorge Querol, Symeon Chatzinotas, Adriano J. Camps Carmona and Bjo¨rn Ottersten 55.1 Introduction 55.2 Two-state noise oscillator model 55.2.1 Discrete-time implementation 55.3 Satellite precoding system with different clock references

xxvii 661 663 665 665 665

669

669 670 674 675 681 682 682

685 686 687 688 688 688 689 690 691 691 693 695 696 696

699

700 702 703 704

xxviii

Advances in Communications Satellite Systems 2

55.4 System implementation 55.5 Simulations results 55.6 Conclusion Acknowledgments References 56 GNSS-assisted acquisition technique for LTE over satellite Xiangdong Liu and Dilip Gokhale 56.1 Introduction 56.2 LTE acquisition and synchronization background 56.2.1 LTE acquisition method overview 56.2.2 Need for modification to operate over a satellite 56.3 Review of prior work in the literature 56.4 A GNSS-assisted method for LTE acquisition and synchronization over satellite 56.5 Summary References

706 708 711 712 712 715 715 716 716 719 720 720 723 723

Section 14: NGSO constellations and 5G integration

725

57 Information rate and quality of service guarantees for end-to-end data flows in an NGSO satellite network Dilip S. Gokhale, Anshul Kantawala and Piya Bhaskar

727

57.1 57.2 57.3 57.4

Introduction NGSO constellations for broadband connectivity Issues with QoS and SLAs in NGSO networks Proposed approach 57.4.1 247 flow admission control 57.4.2 Edge-based NGSO satellite network QoS enforcement 57.4.3 Precision handover management 57.4.4 SDN-based QoS flow tables at user terminals, gateways, and satellite 57.5 Conclusion References

727 728 729 729 729 730 732 734 734 734

58 A new optimization tool for mega-constellation design and its application to trunking systems 737 Steven Kisseleff, Bhavani Shankar, Danilo Spano and Jean-Didier Gayrard 58.1 Introduction 58.2 System modeling and requirements 58.2.1 Link budget 58.2.2 Traffic demand 58.3 System optimization 58.3.1 Optimization parameters

737 739 739 740 741 742

Contents 58.3.2 Methodology 58.4 Numerical results 58.5 Limitations and future enhancements 58.6 Conclusion Acknowledgments References 59 Estimation and compensation of timing drift for NR-based NTN system Jianwei Zhou, Hejia Luo, Chenlei Xu, Zhenjun Jiang, Jun Wang, Bin Wang, Liang Hu and Rong Li 59.1 Introduction 59.2 Methodology 59.2.1 Timing drift compensation method 59.2.2 Calculating timing drift rate based on timing tracking 59.2.3 Calculating timing drift rate based on frequency offset tracking 59.3 Results 59.4 Conclusion References

xxix 743 746 748 749 750 750

753

753 756 756 758 760 762 763 764

60 Spectrum sharing schemes in integrated satellite-terrestrial network 765 Yang Mingchuan, Guan Xin and Miao Xinxin 60.1 Introduction 60.2 System model 60.2.1 Satellite system 60.2.2 Terrestrial system 60.3 Spectrum sharing schemes 60.4 Simulation results and analysis 60.5 Bandwidth estimation method based on protected area spectrum sharing scheme 60.6 Conclusion References

765 767 767 768 770 772

61 Hybrid analog–digital precoding design for satellite systems Aakash Arora, Christos G. Tsinos, R. Bhavani Shankar, Symeon Chatzinotas and Bjo¨rn Ottersten

779

61.1 Introduction 61.2 System model 61.2.1 System description 61.2.2 Architectures 61.2.3 Performance metrics 61.2.4 Channel

775 777 778

779 782 782 783 785 786

xxx

Advances in Communications Satellite Systems 2

61.3 Problem formulation, solutions, and sample performance 61.3.1 Sample performance 61.4 Conclusion Acknowledgments References Section 15: NGSO and GEO system issues and interference mitigation techniques 62 Carrier phase recovery for DVB-S2x standard in VL SNR channel Pansoo Kim and Joon-Gyu Ryu

787 788 789 789 789

793 795

62.1 Introduction 62.2 System description 62.2.1 Transmitter 62.2.2 Signal and channel model 62.3 Carrier phase synchronization 62.3.1 Overall demodulator architecture 62.3.2 Conventional carrier phase synchronization 62.3.3 Proposed approach 62.4 Numerical results and analysis 62.5 Conclusion Acknowledgments References

795 796 796 797 797 797 798 799 800 802 802 802

63 Spectrum prediction and interference detection for satellite communications Lissy Pellaco, Nirankar Singh and Joakim Jalde´n

803

63.1 Introduction 63.2 Proposed approach 63.2.1 Notation and assumptions 63.2.2 Method 63.2.3 Long short-term memory 63.3 Experimental results 63.3.1 Dataset 63.3.2 Architecture and training 63.3.3 Results 63.4 Comparison with a model-based approach 63.4.1 Notation 63.4.2 Method 63.4.3 Experimental results 63.4.4 Comparison 63.5 Conclusion Acknowledgments References

803 805 805 805 806 807 807 809 811 814 814 815 816 816 818 819 819

Contents 64 Channel capacity analysis of satellite MIMO system depending on the orbital altitude Chihaya Kato, Mitsuhiro Nakadai, Daisuke Goto, Hiroki Shibayama and Fumihiro Yamashita 64.1 Introduction 64.2 Proposal of LEO-MIMO channel 64.2.1 Channel model 64.2.2 Coordinate transformation 64.3 Parametric analysis of LEO-MIMO channel capacity 64.3.1 Parameters of analysis 64.3.2 Results and discussion 64.4 LEO-MIMO channel capacity analysis using actual satellite orbital and attitude data 64.4.1 Parameters of analysis 64.4.2 Results and discussion 64.5 Conclusion References

xxxi

821

821 823 823 825 826 826 827 831 831 833 835 835

65 Effects of channel phase in multibeam multicast satellite precoding systems Xavier Artiga and Miguel A´ngel Va´zquez

837

65.1 Introduction 65.2 System model 65.2.1 Channel model 65.2.2 Precoding strategy 65.2.3 Multibeam system and performance metrics 65.3 Unicast 65.4 Multicast 65.4.1 Comparison of clustering techniques 65.4.2 Sensitivity to phase estimation errors 65.5 Conclusions and discussion Acknowledgments References

837 838 839 839 840 841 842 843 844 846 847 847

66 Hardware precoding demonstration in multibeam UHTS communications under realistic payload characteristics Juan Duncan, Jorge Querol, Nicola Maturo, Jevgenij Krivochiza, Danilo Spano, Norshahida Saba, Liz Marrero, Symeon Chatzinotas and Bjo¨rn Ottersten 66.1 Introduction 66.2 Hardware demonstrator 66.2.1 System model 66.2.2 Gateway

849

849 851 852 853

xxxii

Advances in Communications Satellite Systems 2

66.2.3 Channel emulator 66.2.4 User terminal 66.2.5 Resource occupation in FPGAs 66.3 Conclusion Acknowledgments References Index

855 858 861 861 862 863 867

About the editors

Ifiok Otung is a chartered engineer and professor of satellite communications at the University of South Wales (USW), UK. He earned a PhD in Satellite Communications from the University of Surrey, UK in 1995. His areas of expertise include mobile and satellite communication systems and radio wave propagation. He has supervised many PhDs and contributed to numerous publications in technical journals and international conference proceedings. Professor Otung is also the author of several major textbooks including Communication Engineering Principles (Wiley, 2020) and Digital Communications Principles and Systems (IET, 2014). He is a member of the IET and AIAA. Thomas Butash is the founder (2011) and principal of Innovative Aerospace Information Systems (IS), a company providing consulting services on state-of-theart aerospace IS with a focus on communications satellite systems design. Previously, Dr. Butash was a technical director and engineering fellow at BAE Systems (formerly Lockheed Martin, Loral and IBM) Space Systems & Electronics, with more than 30 years’ experience in communications satellite systems development. Dr. Butash is an AIAA Fellow. Tetsushi Ikegami is a professor of telecommunication in the Department of Electronics and Bioinformatics, Meiji University, Kawasaki, Japan. Before his professorship, he led ETS-V, ETS-VI, COMETS Mobile and Ka and mm-waveband satellite experimental teams at the Kashima Space Research Center of the Communications Research Laboratory (CRL, now NICT). He has served as Technical Program and General Chairs of several international conferences, including IWUWBS2005, ISCIT2012, ISSSTA2012, ISMICT2013, ISMICT2020, etc. He is an IEICE fellow and a member of IEEE.

This page intentionally left blank

Acknowledgments

We wish to thank all the contributing authors for the stellar effort they put into writing up their ICSSC 2019 presentations as full technical articles and for responding positively to queries and comments from reviewers and editors. We acknowledge the IET team of Olivia Wilkins and Valerie Moliere for their patient and professional work in managing this selection of the proceedings of ICSSC 2019 through to publication. Organizing a successful not-for-profit international conference such as ICSSC 2019 takes a large team of competent volunteers. We were fortunate to have been part of a highly experienced and dedicated conference organizing committee which ensured the smooth running of all aspects of ICSSC 2019, including one colloquium, keynote speech, four plenary sessions, three social events, 15 technical sessions, and general conference administration. Our special mention goes to Hiroyuki Tsuji, Morio Toyoshima, Yosuke Takahara, and the entire local organizing committee for their untiring work in supporting all aspects of the conference from administration and finance to the colloquium, social, and technical activities. We also thank all the paper reviewers, technical session chairs, panel chairs, panelists, speakers, and conference participants who contributed immeasurably to make ICSSC 2019 an exciting and stimulating forum for exchanging and refining the latest ideas in integrated terrestrial and satellite communications as well as space technologies. We hope that through this volume we have been able to capture and preserve some of these innovative ideas for the benefit of the global satellite communications community. Ifiok Otung, United Kingdom Thomas Butash, United States Tetsushi Ikegami, Japan May 2020

This page intentionally left blank

Communications satellite systems: retrospect and prospect Ifiok Otung1, Thomas Butash2 and Tetsushi Ikegami3

The International Communications Satellite Systems Conference (ICSSC) is widely regarded as the world’s leading technical conference in the field of communication satellite systems. It is certainly the oldest conference in the field, the maiden edition of ICSSC having taken place in 1966 to commemorate the first anniversary of the launch of the world’s first commercial communications satellite, Early Bird (later renamed Intelsat I) on April 06, 1965. Held biennially in even numbered years from 1966 through 2000 and annually since 2001, the 2019 event in Okinawa Japan was therefore the 37th edition of the conference. It brought together researchers, practitioners and experts in academia, industry, space agencies, and regulatory organizations around the world to formulate and exchange the latest ideas and to strengthen alliances and establish new partnerships and collaborations.

Japan and Space The ICSSC has been held in Japan on four occasions, namely in 1998 and 2003 at Yokohama, in 2011 at Nara, and in 2019 at Okinawa. This is largely in recognition of Japan’s significant role in the development of satellite communications. The era of artificial satellites began on October 04, 1957 with the launch of the first artificial earth satellite, SPUTNIK 1, into low earth orbit. Within 13 years, on February 11, 1970, with the launch of Ohsumi satellite into a 144-min elliptical orbit of respective perigee and apogee altitudes 350 km and 5140 km, Japan became only the 4th nation—after the Soviet Union (USSR, 1957), USA (1958) and France (1965)—to successfully place a satellite into orbit on its own. It is worth noting that in addition to USSR, USA, and France, a few other nations did precede Japan in operating a satellite. This includes the UK (April 1962), Canada (September 1962), Italy (December 1964), Australia (November 1967), and 1

School of Engineering, University of South Wales, Pontypridd, UK Innovative Aerospace Information Systems, Centreville Virginia, USA 3 Department of Electronics and Bioinformatics, Meiji University, Kawasaki, Japan 2

2

Advances in communications satellite systems 2: ICSSC-2019

West Germany (November 1969). However, Japan’s contribution to Space and the development of earth station technologies began in 1963 when its first earth station was established at Ibaraki. This was quickly followed by a second earth station at Kashima from which the first ever sustained transoceanic television transmission was relayed to the rest of the world during live coverage of the 1964 Tokyo Summer Olympic Games using the world’s first geostationary satellite, Syncom-3 (launched by NASA on August 19, 1964). In the 55 years since the start of commercial satellite communications, Japan has been among the leading nations in many aspects of Space and has cutting-edge capabilities in rocket and satellite launch technologies, manned and unmanned space explorations, ballistic missile defense and military space programs, earth observation, navigation and spy satellite systems, space science missions, and satellite communications. The National Space Development Agency of Japan (NASDA) was established on October 01, 1969 and launched its first geostationary orbit (GEO) satellite, known as Kiku 2, on February 23, 1977. The first Japanese experimental communications satellite, Sakura 1, was built by Mitsubishi Electric and Ford Aerospace and launched on December 15, 1977 on a United States Delta2914 rocket from the Kennedy Space Center into geostationary orbit (GEO). This was followed by an experimental direct broadcasting satellite (DBS), Yuri 1, built by Toshiba in cooperation with RCA Astro and launched on April 07, 1978 on the same rocket. Two other experimental DBS satellites, Yuri 2a and Yuri 2b, were launched on January 23, 1984 and February 12, 1986, but this time from the Tanegashima Space Center using the Japanese N-2 Star-37E rocket. These satellites supported a wide range of important studies such as propagation measurements, frequency sharing and satellite control techniques, ground and satellite terminals performance, satellite broadcasting operations, TV signal characteristics, etc. To provide communication services for voice, facsimile, high-speed data for business and television distribution direct to homes, meet the demands of Japan’s many islands, and maintain vital connection in emergencies and natural disasters, a domestic satellite telecommunications network was established. This network involved various GEO satellites, including the Sakura (2a, 2b, 3a, 3b) communication satellites series and the broadcasting satellite series (BSat 1a, 1b, 2a, 2c, 3a, 3b, 3c, 4a) both of which are Government controlled, and the commercial communications satellite series JCSat (1, 2, . . . , 18) which are privately owned by the Japanese Communications Satellite Company (JCSAT) created in 1985 by Hughes, Mitsui, and C. Itoh. These domestic satellites have grown in capacity and technological sophistication from Sakura 2a, launched on February 04, 1983 carrying four Ka-band transponders, to JCSat 18—a high-throughput satellite (HTS) launched on December 16, 2019 with Ku band payload and 56 geographically tailorable Ka-band spot beams, each having a capacity up to 1.25 Gb/s. The coverage areas of these satellites have also been expanded to reach many Pacific and South East Asian countries with communication services that include high-speed Internet access, mobile telephony, and broadband data. Japan’s active role in the development of advanced communications satellite technologies has continued in

Communications satellite systems: retrospect and prospect

3

more recent years with the launch of the Wideband InterNetworking engineering test and Demonstration Satellite (WINDS, also known as Kizuna) into GEO on February 23, 2008 from the Tanegashima Space Center and the planned launch of Kiku-9 or ETS-9 (Engineering Test Satellite) in 2021. WINDS was operational until February 2019 and supported download speeds up to 155 Mb/s and 1.2 Gb/s to terminals with 45 cm and 5 m aperture antennas, respectively. Using airborne, seaborne, and land-vehicle-borne earth stations, WINDS enabled a wide range of high data rate transmission experiments looking into network architectures, disaster and fade countermeasures, dynamic communication resource sharing, regenerative transponder operation, protocol evaluation, etc. ETS-9 is an HTS that will feature GPS-aided labor-saving orbital maneuvers as well as an all-electric-propulsion bus system based on high-power Hall thrusters.

The 37th International Communications Satellite Systems Conference (ICSSC 2019) In recent years, the pace of growth in Satellite Communications has been accelerating not just in Japan but globally. The ICSSC 2019 conference was held at the Okinawa Municipal Autonomous Center in Naha City and its program showcased these developments under an overarching theme: “Space Communications, Navigation and Earth Observation in the 5G Era: Ensuring a realistic migration, harmonization and integration of space-based communications infrastructure within the 5G network.” Some of the key topics covered at the conference over a period of three days, from October 30 to November 01, 2019, included: ● ● ● ● ● ● ● ● ● ● ● ● ●

Interoperable networks: Quantum leap into the future Satellite industries and related activities in Asia-Pacific region Smart mobility and GNSS Satcom in 5G and digital disruption era Broadband satellite communications architectures and applications Flexible high-throughput satellite systems New satellite components and transmitter and modem technologies High speed optical communications and feeder links Satellite antenna technologies and advanced digital payloads and components Future technologies for 5G and beyond Propagation and modeling for satellite communications NGSO constellations and 5G integration NGSO and GEO system issues and interference mitigation techniques

In addition, a full day workshop or colloquium was held on October 29, 2019, a day prior to the opening of the conference, on the subject of “Disruptive expansion of space laser communications.” The colloquium explored the current state of the art in a range of areas including optical LEO-LEO and LEO-GEO intersatellite links, space-to-earth laser links, optical payloads, and in-orbit verification of various optical technologies. The day of the workshop was divided into four sessions.

4

Advances in communications satellite systems 2: ICSSC-2019

The first morning session featured presentations from various national space agencies, including the US, Japan, ESA, and China, on their space laser communications programs. A second morning session was devoted to various presentations covering optical ground stations, technologies, operations, and services. In the afternoon, the first session examined laser communications for constellations and small satellites, whereas the second session explored emerging ideas and applications of space quantum communications. An introductory plenary session of the conference featured a keynote address by Badri Younes of NASA on interoperable networks, which was followed by a panel discussion. The recent annual economic growth of around 6% in the AsiaPacific region and rapid increase in demand for communication services have made the region very attractive for commercial satellite industries. The panel discussed the growing satellite industry and related economic activities in the Asia-Pacific region. A second plenary session on the opening day of the conference focused on smart mobility and enabling Global Navigation Satellite System (GNSS) technologies. Michibiki, the Japanese GNSS system which began service in November 2018 using quasi-zenith satellites, was discussed along with other global and regional GNSS and mobility systems. The last day of the conference saw two more plenary sessions dedicated to satellite communications in the 5G and digital disruption era. The first of these sessions explored the roles of satellite communications in 5G deployments, supporting applications such as backhaul links and coverage expansion to oceans, air space, and remote regions. This session also featured a talk by Junichi Miyakawa, President of the HAPS Mobile Inc., on the development situation of the stratospheric platform communications system (HAPS). The second session, enriched by critical audience participation, addressed the technical and business challenges leading to industry winners and losers, presented a lively chronicle of the rapidly changing satellite industry landscape and identified some of the key emerging trends, including ground-breaking concepts such as software defined digital payloads. Panelists for all the plenary sessions were drawn from leading industrial authorities in both services and manufacturing sectors. Sandwiched between the opening-day and closing-day plenary sessions were 15 technical sessions in which 70 accepted papers were presented within 20-min time slots by academic and industry-based researchers. To accommodate all the technical papers, sessions were scheduled in two parallel streams. Each technical session was managed by two co-chairs to cover an initial talk by the speaker followed by a lively exchange of ideas involving audience contributions. The papers covered a range of topics in satellite systems (HTS, NGSO, and GEO), technologies (digital payloads, optical communications, and IoT), components (antennas, modems, amplifiers, and terminals), architectures (5G integration, broadband networks, and satellite constellations), and techniques (propagation modeling, fade and interference mitigation, and adaptive coding and modulation). In addition, an entire technical session was devoted to papers that attempted to lay a roadmap for future trends in terrestrial and satellite communication technologies.

Communications satellite systems: retrospect and prospect

5

An exciting component of the ICSSC and one that contributes to a continued understanding across the global community, is its social program. The conference offers a spousal attendance package and is for many an annual opportunity to renew long distance friendships. Apart from the splendid moments of informal interactions late into the pleasant Okinawa nights, the official ICSSC 2019 social program included a cocktail evening, an awards luncheon, and a conference dinner. The cocktail evening was a welcome reception party on the first day of the conference held at the “Pine Tree Bless in T Galleria by DFS” restaurant located next to Okinawa’s Omoromachi station. Each year, the American Institute of Aeronautics and Astronautics (AIAA) prestigious Aerospace Communications Award is presented by the AIAA Communications Systems Technical Committee at the ICSSC to the individual(s) deemed to have made outstanding contributions to advancing the field of communications satellite systems. This year’s awards luncheon, held on the second day of the conference, saw the AIAA Aerospace Communications Award presented to Professor Michel Bousquet for his outstanding contribution and promotion of education and proliferation of knowledge on aerospace communication and navigation. Finally, the conference banquet, held at the Double Tree by Hilton Hotel Naha Shuri Castle, was a grand and memorable event at which conference participants were treated to excellent food and freeflowing drinks in a jolly atmosphere spiced with exciting Japanese dance entertainments. In all, there were 150 participants at ICSSC 2019 drawn from 16 countries, including Japan, USA, Germany, Luxembourg, United Kingdom, Australia, Canada, UAE, Italy, China, Korea, Finland, Netherland, Singapore, Spain and Sweden. At the close of the conference, the organizers were able to announce that, as in 2005 (in Rome, Italy), 2012 (in Ottawa, Canada), 2013 (in Florence, Italy), 2016 (in Cleveland, USA), 2017 (in Trieste, Italy), and 2018 (in Niagara Falls, Canada), the next year’s event (ICSSC 2020) would be held jointly with the Ka and Broadband Communications Conference in Arlington, Virginia, USA on October 13–16, 2020.

About this volume This book is based on the presentations given in the various technical sessions of ICSSC 2019. A selection of 66 of the presentations covering significant and hitherto unpublished contributions to advances in communications satellite systems have been written up and edited as full articles and are presented as separate chapters of the present volume. The articles cover a broad range of space and terrestrial communication technologies and are grouped into 15 sections: ● ● ● ● ●

Broadband Satellite Communication Architectures and Applications Integrated Applications and Architectures for Vessels and IoT DTN and HTS Technologies New Satellite System Architectures and Components High Speed Optical Communications and Feeder Links 1

6 ● ● ● ● ● ● ● ● ● ●

Advances in communications satellite systems 2: ICSSC-2019 Advanced Digital Payloads and Components High Speed Optical Communications and Feeder Links 2 Satellite Antenna Technologies Propagation and Modeling for Satellite Communications Future Technologies for 5G and Beyond Flexible HTS Systems and Advanced Digital Payloads Satellite Networks Design Challenges and Applications New Satellite Components and Transmitter and Modem Technologies NGSO Constellations and 5G Integration NGSO and GEO System Issues and Interference Mitigation Techniques

Each chapter is a self-contained technical article with its own references. No attempt was made to integrate the contributions to eliminate any repetitive referencing or overlap in the treatment of some concepts. It is hoped that this approach preserves the individual style and perspective of the various authors and will allow readers to dip into any topic or chapter in any order without loss of continuity. The contributing authors come from a wide range of backgrounds in academia, industry, government, and regulatory bodies. The book is therefore a multisectoral collection of research advances which should be of interest and great benefit to satellite industry practitioners, academic researchers, and other technical personnel engaged in telecommunications in general.

Looking ahead Research and Development results presented at the 37th ICSSC in Okinawa and published in this volume clearly advanced communications satellite system stateof-the-art architectures, technologies, performance, and feasible applications. These advances establish new approaches to enhance flexibility, support autonomous mobile platforms, Internet of things (IoT), and machine to machine (M2M) communications, and integration of satellite and terrestrial wireless networks. GEO and NGSO systems integration with and support of 5G networks, including backhaul, tower feed, and 5G protocols were presented, suggesting an inevitable convergence of terrestrial and satellite broadband networks. Technologies to implement these advanced architectures, and realize increased HTS and VHTS throughput, including new transmitter and receiver component technologies, high speed optical communications, and advanced digital payloads for increased flexibility and frequency utilization efficiency—through digital channelization and beamforming or beam hopping—were presented. Finally, advanced interference mitigation techniques were proposed to support ongoing increases in the number of deployed NGSO megaconstellation satellites and spectrum sharing between GEO and NGSO systems. However, as noted in these proceedings and during plenaries held at the conference, continuing exponential increases in terrestrial fiber-optic and advanced 4G and 5G wireless broadband networks’ geographic coverage and data access rates will require broadband communications satellite networks having

Communications satellite systems: retrospect and prospect

7

even higher flexibility, capacity, user access rates, and affordability if they are to remain competitive with and become integral parts of much larger terrestrial broadband access networks. Sub-6 GHz band 5G wireless access currently achieves download speeds between 450 Mbps and 1 Gbps, with mmWave band 5G user download speeds running between 1 Gbps and 10 Gbps. Thus, communications satellite system capacities, user access speeds, and service affordability will have to see quantum increases for these systems to remain an integral part of global broadband communications network solutions. This, in turn, will further accelerate the roll-out of terrestrial broadband networks by providing essentially ubiquitous coverage for their more latency-tolerant and bandwidth-limited applications. Advances in communications satellite systems technologies that will contribute to realizing these quantum increases, and hence which deserve further research, include: ●

●

●

●

●

●

More efficient, higher performance RF front- and back-end technologies for implementing Ka, Q, and V band satellite payloads. Lower power dissipation, higher-performance analog to digital converter (ADC) and digital to analog converter (DAC) devices to facilitate direct digital conversion at Ka-band and higher frequencies. Space flight qualified, radiation tolerant digital ASIC and FPGA technologies fabricated at CMOS and SOI process nodes closer to those employed in stateof-the-art commercial production lines (currently 7 nm). Hyper-efficient digital signal processing algorithms and architectures for broadband channelization (frequency division demultiplexing/multiplexing) and beamforming to effect increased flexibility (in-orbit adaptability) and decrease the finance, size, weight, and power ($SWaP) costs of broadband satellite access capacity. Further reductions in launch costs, most likely to be realized with fully reusable vehicles. Currently, the lowest launch cost of US$5K/kg approximately doubles the in-orbit cost of a LEO broadband megaconstellation communications satellite which is production line manufactured. To render NGSO broadband megaconstellation systems more cost competitive and responsive relative to terrestrial broadband wireless networks, launch costs must become a small fraction of the satellites’ costs. Ka-band electronically steerable array (ESA) antennas capable of scan angles and beam directivities compatible with megaconstellation link performance requirements, like those of Starlink and/or OneWeb, at an OEM cost on the order of US$30, to enable affordable (i.e., US$250) consumer terminals for rapid widescale adoption of direct-to-home broadband access services from NGSO megaconstellations.

In as much as terrestrial wireless broadband access systems are able to employ lower cost, higher performance commercial-off-the-shelf (COTS) transmitter and receiver technologies, and enjoy propagation losses less than those of satellite broadband access systems by at least 50 dB, it is imperative that the

8

Advances in communications satellite systems 2: ICSSC-2019

communications satellite industry pursue these existential research goals. The 37th ICSSC results published in these proceedings clearly point the way forward. The solutions discovered by following these research and development paths will surely provide excellent advances for presentation at future International Communications Satellite Systems Conferences.

Section 1

Broadband satellite communication architectures and applications

This page intentionally left blank

Chapter 1

The results of WINDS experiments of NICT Tajkashi Takahashi1, Naoko Yoshimura2, Kaszuyoshi Kawasaki1, Tomoshige Kan1, Byeongpyo Jeong3 and Morio Toyoshima2

The Wideband InterNetworking engineering test and Demonstration Satellite (WINDS) was developed for the research of high-data-rate satellite communication technologies using Ka-band. WINDS was launched on February 23, 2008 and completed its operation on February 27, 2019. The National Institute of Information and Communications Technology (NICT) planned and conducted various experiments, such as satellite communication experiments for disaster countermeasures, an orthogonal frequency-division multiplexing transmission experiment, and mobile satellite communication experiments, using a small vehicle earth station, an airborne earth station, and a seaborne earth station. Key Words: WINDS; Ka-band; NICT fundamental experiments

1.1 Introduction The Wideband InterNetworking engineering test and Demonstration Satellite (WINDS) [1] was developed by the National Institute of Information and Communications Technology (NICT) and the Japan Aerospace Exploration Agency (JAXA) for high-data-rate satellite communication technology. WINDS was launched by the H-IIA rocket from the Tanegashima Space Center on February 23, 2008. After the initial on-orbit checkout, the regular operational phase was initiated, and the NICT and JAXA began conducting fundamental experiments. In addition, the Association for Application Experiments of WINDS (AAEW) conducted application experiments that had been widely recruited and adopted by the Ministry of Internal Affairs and Communications (MIC). 1

National Institute of Information and Communications Technology, Kashima, Japan National Institute of Information and Communications Technology, Koganei, Japan 3 National Institute of Information and Communications Technology, Aobaku, Sendai, Japan 2

12

Advances in communications satellite systems 2: ICSSC-2019

The NICT conducted experiments, including onboard equipment performance experiments and transmission experiments, according to a fundamental experiment plan. WINDS then began the postoperational phase in April 2013. The NICT also created a fundamental experiment plan in the postoperational phase and conducted experiments. WINDS completed its operation on February 27, 2019. In this chapter, the WINDS and NICT experiments in the postoperational phase are introduced.

1.2 WINDS WINDS [1] is an engineering test satellite used to establish high-data-rate satellite communication technologies using Ka-band. The configuration of WINDS is presented in Figure 1.1.

LNA

U/C

D/C

RX APAA

TX SEL (8×19)

ABS

MPA (8×8)

ATMS

MOD

BPF-U BPF-U BPF-W BPF-W Bent pipe DEM

LNA

RX IFS (11×13)

LNA

RX SEL & D/C

LNA

U/C (8)

MBA: domestic beam

TX IFS (7×10)

MBA: Asian beam

BPF-W BPF-W Bent pipe

LNA : Low noise amplifier D/C : Down converter SEL : Selector IFS : Intermediate frequency switch BPF-U : Band pass filter (upper band: 550 MHz) BPF-W : Band pass filter (wide band: 1.1 GHz) ABS : ATM baseband switch

TX APAA DEM ATMS MOD U/C APAA MPA MBA

: Demodulator : ATM switch : Modulator : Up converter : Active phased array antenna : Multi-port amplifier : Multi-beam antenna

Figure 1.1 Configuration of the Wideband InterNetworking engineering test and Demonstration Satellite (WINDS) transponder

The results of WINDS experiments of NICT

13

The NICT developed the advanced baseband switch (ABS) subsystem and installed it on WINDS. The ABS consisted of three demodulators, two switches including a redundant one, and three modulators. WINDS had two relay modes: a bent-pipe relay and a regenerative relay. The regenerative mode used ABS. In the bent-pipe mode which bypassed the ABS, a transmission signal from the earth station was received, the frequency was converted and amplified to be sent to the earth station. WINDS consisted of two fixed multibeam antennas (MBAs) and an active phased array antenna (APAA). The MBAs covered Japan and 10 major cities in Southeast Asia, whereas the APAA could be controlled electronically and covered the Asia Pacific region. The coverage of WINDS is presented in Figure 1.2, whereas the major specifications of the communication mission are presented in Table 1.1.

1.3 WINDS experiments After WINDS was launched on February 23, 2008, it was placed into a geostationary orbit of 143
east, and an initial checkout was performed, which confirmed its soundness. Regular WINDS operation began in July 2008, and the NICT and JAXA, which developed WINDS, began to conduct experiments. The MIC recruited application experiments using WINDS; many universities and companies applied, and 53 were selected. They constituted the AAEW,

Multi beam antenna (MBA) Active phased array antenna (APAA)

Coverage by MBA

Coverage by APAA

Figure 1.2 Coverage of the Wideband InterNetworking engineering test and Demonstration Satellite (WINDS)

14

Advances in communications satellite systems 2: ICSSC-2019

Table 1.1 Major specifications of communication mission MBA Frequency band Antenna type Number of beams G/T EIRP Regenerative mode Tx: 155 Mbps

Tx: 17.8–18.8 GHz Rx: 27.5–28.6 GHz Offset Cassegrain (2.4 m) Nine Japan and 10 Southeast Asian cities >18 dB/K >68 dBW

APAA

Direct radiating phased array antenna Tx: 2 Rx: 2 >7 dB/K >55 dBW Rx: 1.5, 6, 24, 51, 51  3 Mbps

APAA, active phased array antenna; EIRP, effective isotropically radiated power; MBA, multibeam antenna.

adjusted the schedule, and promoted the application experiments. Figure 1.3 presents the course of the WINDS operation and experiments. Five years from the launch, which was the design lifetime of the satellite, the postoperational phase of WINDS began in April 2013. The application experiments had completed within the regular operational phase. In the postoperational phase, JAXA planned utilization demonstration experiments, whereas the NICT planned fundamental experiments.

1.4 NICT fundamental experiments in the regular operational phase As presented in Table 1.2, the NICT conducted fundamental experiments [1] in the regular operational phase, including experiments to verify the performance of onboard equipment, fundamental transmission experiments, high-data-rate satellite network experiments, and network application experiments.

1.5 NICT fundamental experiments in postoperational phase NICT planned eight types of experiments in the postoperational phase, as listed in Table 1.3 and introduced in the following subsections 1.5.1–1.5.5.

1.5.1

Function verification experiment of fully automatic transportable earth station

After the Great East Japan Earthquake that occurred in March 2011, the NICT developed a fully automatic transportable earth station for WINDS, as presented in Figure 1.4. It was simple to set up and acquire WINDS without a specialist. The major specifications listed in Table 1.4 were verified.

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

WINDS

Launch C/O

Regular operation phase

NICT

Fundamental experiments

Fundamental experiments

JAXA

Post–operation phase

Fundamental experiments in post–operation phase

Utilization demonstration

Application experiments

AAEW

C/O: Check out AAEW: Association for Application Experiments of WINDS

Figure 1.3 Course of operation and experiments of the Wideband InterNetworking engineering test and Demonstration Satellite (WINDS)

16

Advances in communications satellite systems 2: ICSSC-2019

Table 1.2 List of National Institute of Information and Communications Technology (NICT) fundamental experiments No.

Experimental item

N-A-01 N-A-02 N-A-03 N-A-04 N-C-01 N-C-02 N-C-03 N-C-04 N-C-05 N-C-06 N-D-01 N-D-02 N-D-03 N-D-04 N-D-05 N-E-01 N-E-02

Level diagram verification experiment Frequency characteristics verification experiment APAA performance evaluation Regenerative transponder function verification experiment TDMA synchronization experiment Rain attenuation compensation experiment Bent-pipe mode transmission characteristics experiment Regenerative mode transmission characteristics experiment ABS congestion experiment 1.2-Gbps transmission experiment Star network experiment Mesh network experiment Protocol evaluation experiment Dynamic demand assignment experiment SHV transmission experiment Connection experiment with ground network Medical ICT satellite communications experiment

ABS, advanced baseband switch; APAA, active phased array antenna; ICT, information and communications technology; SHV, super high definition video; TDMA, time division multiple access.

Table 1.3 List of National Institute of Information and Communications Technology (NICT) fundamental experiments in the postoperational phase No.

Experimental item

N-L-01 N-L-02 N-L-03 N-L-04 N-L-05 N-L-06 N-L-07 N-L-08

Function verification experiment of fully automatic transportable earth station Function verification experiment of small-sized vehicle earth station Function verification experiment of small station on vessel Satellite network experiment Satellite communication experiments for disaster countermeasures Connection experiment with ground network Medical ICT satellite communication experiment Basic experiment for future satellite technology

ICT, information and communications technology.

1.5.2

Function verification experiment of small-sized vehicle transportable earth station

Drawing on lessons from the Great East Japan Earthquake, the NICT developed a small-sized vehicle earth station to be able to communicate with the dispatching agency while moving to a disaster area, as presented in Figure 1.5. This station can track the satellite and use the satellite to communicate while traveling [2].

The results of WINDS experiments of NICT

17

Figure 1.4 Fully automatic transportable earth station Table 1.4 Major specifications of fully automatic transportable earth station Tx frequency Rx frequency Polarization Antenna EIRP G/T Antenna drive range Data rate (regenerative mode) User interface

27.5–28.6 GHz 17.7–18.8 GHz Linear 1-m offset parabola >64.9 dBW >18.7 dB/K El: 15º–75º Az: þ/95º Tx: 1.5, 6, 24, 51 Mbps Rx: 155 Mbps Ethernet (1000 Base-T)

The major specifications listed in Table 1.5 were verified. The NICT also used the earth station to measure the propagation characteristics.

18

Advances in communications satellite systems 2: ICSSC-2019

Figure 1.5 Small-sized vehicle earth station Table 1.5 Major specifications of small-sized vehicle earth station Tx frequency Rx frequency Polarization Antenna EIRP G/T Drive range Tracking accuracy Data rate (regenerative mode) User interface

27.5–28.6 GHz 17.7–18.8 GHz Linear 0.65-m Cassegrain antenna >55.5 dBW >13.5 dB/K El: 20
–90
Az: 360
(continuous) 35.4 dBW >9.0 dB/K El: 25
–65
Az: 360
(continuous) 1: 2: 3:

50.00 40.00

47.500 GHz 48.000 GHz 48.500 GHz

–1.47 dB –1.44 dB –1.49 dB

30.00 20.00 10.00 1

0.00

2

3

–10.00 –20.00 –30.00 –40.00 –50.00 1 >Ch1: Start 40.0000 GHz

Stop 50.0000 GHz

Figure 11.9 Insertion losses of the back-to-back module

High-performance V-band GAN MMIC HPA for the QVlift project Tr 1 S11 LogM 10.00 dB/ 0.00 dB Tr 3 S22 LogM 10.00 dB/ 0.00 dB 50.00 40.00

155

Tr 2 S21 LogM 10.00 dB/ 0.00 dB > 1: 2: 3:

47.500 GHz 48.000 GHz 48.500 GHz

–2.47 dB –2.44 dB –2.64 dB

30.00 20.00 10.00 0.00 –10.00

1 2

3

–20.00 –30.00 –40.00 –50.00 1 >Ch1: Start 40.0000 GHz

Stop 50.0000 GHz

Figure 11.10 Insertion losses of 50-ohm line mounted inside a module

Finally, a 50-ohm line provided by OMMIC was mounted and measured inside one of the modules (see Figure 11.10). Losses of the 50-ohm line, including bondings (this measurement minus losses of back-to-back module with adapters), are 1 dB at 48 GHz. An electromagnetic simulation of the assembled 50-ohm line was performed by OMMIC. Simulated values are usually optimistic, but it allows estimating the losses of the bonding wires: – –

Losses of 50-ohm line: @ 48 GHz: 0.5 dB Losses of bonding wires (three wires in parallel): @ 48 GHz: 0.25 dB

In order to properly compare the MMICs on-wafer results with the mounted module performances, the following piece of information should be considered: 1.

2.

From the measurements shown in Figures 11.8–11.10, the back-to-back module insertion loss is around 1 dB for output power results while 2 dB for gain levels. Due to thermal aspects, CW mode operations show a lower output power and gain than the pulsed mode. From previous experience with the same technology, we expect a reduction of around 0.5 dB for power results and 1 dB for gain levels.

156

Advances in communications satellite systems 2: ICSSC-2019 On-wafer pout (dBm ) ( @Vd=12 V) 38

Pout (dBm)

37 36 35 34 33 46

46.5

(a)

47 Frequency (GHz)

47.5

48

On-wafer linear gain (dB) (@Vd=12 V) 19 18

Gain (dB)

17 16 15 14 13 (b)

46

46.5

47 Frequency (GHz)

47.5

48

Figure 11.11 (a) On-wafer output power and (b) linear gain versus frequency

11.3.3 HPA on-wafer measurements The on-wafer measurements of the output power and linear gain were performed by OMMIC on several dies. Results are shown in Figures 11.11 and 11.12 for one representative HPA. Depending on the performance level, the best chips were selected and mounted in HPA modules.

11.3.4 HPA preliminary power measurements A set of 10 units of HPAs was assembled and characterized at 48 GHz. Preliminary measurements were performed in terms of functional direct current (DC) tests and output power versus input power. Results of output power and gain compression are showed in Figures 11.13 and 11.14 for one HPA as an instance.

Pout (dBm)

High-performance V-band GAN MMIC HPA for the QVlift project 37 35 33 31 29 27 25 23 21 19 17 15

157

On-wafer pout (dBm) (@Vd=12 V) @ 48 GHz

0

5

15

10

20

25

Pin (dBm)

Figure 11.12 On-wafer output power as a function of input power

Pout (dBm)

Pout (dBm) @ 48 GHz 35 33 31 29 27 25 23 21 19 17 15 –45

–40

–30

–35

–25

–20

Pin (dBm)

Figure 11.13 Output power as a function of input power Table 11.1 summarizes the measured values of output power, gain, and compression level obtained in CW conditions at 24 dBm of input power at HPA level, for the entire set of HPAs. These HPAs were selected among the best performing chipset provided by OMMIC with the first run of the MMIC fabrication. According to the project work plan, two BUCs will be integrated and delivered for the final system test campaign of QVLIFT and each BUC will use one HPA as a driver and four combined HPAs for the power stage.

158

Advances in communications satellite systems 2: ICSSC-2019

Gain compression (dB)

Gain compression of HPA (dB) @ 48 GHz 0 –1 –2 –3 –4 –5 –6 –7 –8 –9 –10 –45

–40

–35

–30

–25

–20

Pin (dBm)

Figure 11.14 Gain compression as a function of input power

Table 11.1 HPAs preliminary CW measurements at 48 GHz (at 24 dBm of input power) HPA HPA HPA HPA HPA HPA HPA HPA HPA HPA HPA

Pout (dBm) Gain (dB) Compression (dB) #1 #2 #3 #4 #5 #6 #7 #8 #9 #10

32.6 33.7 33.7 33.4 33.4 31.4 32.7 33.4 33.2 33.4

8 9.2 10.9 10.7 10.7 6.8 9.9 10.7 10.5 10.7

6.8 6.3 5.4 7.1 5.5 5.1 5.2 6.1 6 5.9

11.4 Conclusions In this chapter, we have shown the integration and functional test of a V-band GaN MMIC high power amplifier, with the development of a dedicated waveguide module and the performance analysis in CW conditions. As presented, the developed HPA was able to provide up to 2 W at 48 GHz. Its behavior in terms of gain and power consumption is coherent with expected results. Achieved results validate the operation of the HPA with respect to the specifications dictated by the QVLift system requirements. Ongoing activities are focused on preliminary reliability tests (including 500 h DC Life Test) of the HPA, assembly, and characterization of the final BUC units, which will be part of the RF chain of the QVLIFT ground terminals.

High-performance V-band GAN MMIC HPA for the QVlift project

159

Acknowledgments The authors acknowledge the QVLIFT project funded by the European Commission under the Horizon 2020 program (Grant Agreement No. 730104 H2020-COMPET-2016).

References [1] Q/V-band earth segment link for future high throughput space systems. H2020- COMPET-2-2016. Available from https://www.qvlift.eu/ [Accessed September 19, 2019]. [2] Codispoti, G., Parca, G., and Amendola, G., et al. “RF technologies for the ground segment of future Q/V band satellite systems.” 23rd Ka and Broadband Communications Conference. Trieste, Italy. 2017. [3] Codispoti, G., Parca, G., and Amendola, G., et al. “Validation of ground technologies for future Q/V band satellite systems: the QV – LIFT project.” IEEE Aerospace Conference. Big Sky, Montana. 2018. [4] Moron, J., Leblanc, R., Frijlink, P., et al. “A novel high-performance V-band GaN MMIC HPA for the QV-LIFT project.” 24th Ka and Broadband Communications Conference. Canada. 2018.

This page intentionally left blank

Chapter 12

Performance study of frequency flexibility in high throughput satellites and its contribution to operations strategy Yuma Abe1, Mitsugu Okawa1, Amane Miura1, Kazunori Okada1, Maki Akioka2 and Morio Toyoshima1

In this chapter, we describe a plan for comprehensive evaluation of a satellite communications (SATCOM) system with frequency flexibility. To cope with the more and more increasing demand of the SATCOM system, nextgeneration high-throughput satellites (HTS) will install a frequency flexibility function that can change the frequency assignment flexibly. To control this HTS, the flexible channel assignment method has been proposed. Under the time-varying traffic, this method performs higher throughput and reduces the number of control actions that change the frequency assignment. In the next step, comprehensive evaluation of the HTS system with the frequency flexibility is required. We summarize our previous results and describe a plan for the comprehensive evaluation of the HTS system in terms of performance index, communication traffic, and link assignment schemes. Finally, we suggest operations strategy of the HTS system including the predictive control. Key Words: satellite communications; frequency flexibility; time-varying traffic

12.1 Introduction In recent years, the demand for satellite communications (SATCOM) has been more and more increasing. Use cases of the SATCOM have spread over a wide variety of areas: aeronautical, maritime, emergency, and Internet of things (IoT) communications [1]. 1 Space Communications Laboratory, Wireless Networks Research Center, National Institute of Information and Communications Technology (NICT), Tokyo, Japan 2 Planning Office, Wireless Networks Research Center, National Institute of Information and Communications Technology (NICT), Tokyo, Japan

162

Advances in communications satellite systems 2: ICSSC-2019

To improve frequency utilization efficiency of satellite resources such as bandwidth, a frequency flexibility function that can flexibly change frequency assignment of each spot beam is implemented in next-generation high-throughput satellites (HTSs) [2,3] by installing a digital channelizer (DC) [4]. Figure 12.1 shows a satellite communications system for the HTS with the frequency flexibility. In this system, the HTS is connected to user terminals such as aircraft, ships, and very small aperture terminals (VSATs) in user links by multiple spot beams. In a feeder link, the HTS is connected to a ground station connected to the terrestrial network. This system should be operated automatically because multiple user terminals try to request the resources of the HTS. Thus, a network operations center (NOC), which works as a controller is implemented in this system. User information such as communication traffic from the user terminals and Ka-band link status are aggregated to the NOC and the NOC determines an efficient frequency assignment for each user terminal. Then, a satellite operations center (SOC) sends a control command to the HTS and the frequency assignment is controlled. By using the functions of the NOC and SOC, the effective assignment is achieved so that the total capacity of each beam exceeds the traffic of each beam. We focus on the frequency assignment method, which is implemented in the NOC. We proposed a flexible channel assignment method [5], which is combined with conventional fixed and dynamic assignment methods. This method assigns fixed bandwidth and variable bandwidth separately. By using this method, the number of control actions can be decreased when the assigned fixed bandwidth Next-generation HTS with frequency flexibility User links

Feeder link

Frequency flexibility

Ground station

Multiple spot beams

Satellite operations center (SOC)

Terrestrial network

Time variation of traffic

MT position MT traffic

Link status Network operations center (NOC)

Figure 12.1 HTS system with frequency flexibility

Frequency flexibility in high throughput satellites

163

50 45

Latitude (˚)

40 35 30 Beam cluster Assigned from

25 A

20 120

130

140 Longitude (˚)

C 150 LHCP

B

Lower limit frequency

Upper limit frequency RHCP

D

Figure 12.2 Beam arrangement in the numerical simulation (45 beams) exceeds the total traffic in each beam. Thus, the energy consumption of the HTS also can be decreased because it is unnecessary to control the HTS. In this chapter, we summarize our previous results such as system configuration, frequency assignment methods, and previous evaluation results showing the effectiveness of the proposed flexible channel assignment method. We further describe a plan for comprehensive evaluation of the HTS system with frequency flexibility in terms of performance index, communication traffic, and link assignment schemes. Finally, we suggest operations strategy of the HTS system.

12.2 Summary of previous results In this section, we summarize our previous results of the performance study of the HTS with the frequency flexibility [5,6].

12.2.1 System configuration The HTS has a multibeam antenna and is installed in the DC as described in Figure 12.2. An example of beam arrangement is shown in Figure 12.1. Four beams having different properties consist one cluster and this cluster is repeatedly placed to cover Japan. For pairs of beams (A, B) and (C, D), the bandwidth is assigned from lower and upper limit frequency, respectively. On the other hand, for pairs of beams (A, C) and (B, D), the left- and right-hand circularly polarized wave (LHCP and RHCP) are used, respectively. The maximum bandwidth is shared with adjacent beams exhibiting the same polarization. Thus, this beam arrangement avoids the interference between the adjacent beams.

164

Advances in communications satellite systems 2: ICSSC-2019

12.2.2 Frequency assignment method We proposed the flexible channel assignment method [5]. We have verified that this proposed method has outperformed the conventional fixed and dynamic assignment methods in terms of throughput and number of control actions which change the frequency assignment. The features of the conventional methods are as follows: ●

●

For the fixed assignment method, bandwidths are assigned fixedly for all beams and this cannot cope with the time-varying traffic. For the dynamic assignment method, bandwidths can be changed according to the traffic and can cope with the time-varying traffic.

The proposed flexible channel assignment method divides the maximum bandwidth into two: fixed and dynamic channel bandwidths as shown in Figure 12.3 [6]. The fixed channel band-width is fixedly assigned to each beam; the dynamic channel bandwidth constitutes the remaining maximum bandwidth, which can be changed according to the traffic of each beam at each time step. This method can cope well with time-varying traffic, similar to the dynamic assignment method. Furthermore, if the traffic does not exceed the fixed channel bandwidth, the frequency assignment does not need to be changed; hence, the number of control actions is reduced.

12.2.3 Previous evaluation results In this section, we show a previous evaluation result [6] of the numerical simulation in a disaster situation. We set the beam arrangement of the HTS as illustrated in Figure 12.2. Figure 12.4(a) and (b) shows the results comparing the three methods: the fixed, dynamic, and flexible channel assignment (as proposed) methods in terms of the throughput and the number of control actions. In Figure 12.4(a), the results of the dynamic and proposed assignments are overlapped. In Figure 12.4(b), the number of control actions of the fixed assignment is zero because the fixed Upper limit frequency

500

100

100

100

500 MHz

250

150

150

150

0 Lower limit frequency

Fixed channel

Dynamic channel

15

21

28

34

Beam number

Figure 12.3 Conceptual diagram of the flexible channel assignment method. The orange blocks represent the fixed channel bandwidths, which remain fixed and the blue and green blocks with arrows represent the dynamic channel bandwidths, which are for the time-varying traffic

Throughput (Mbps)

Frequency flexibility in high throughput satellites 1000

Fixed assignment Dynamic assignment Proposed assignment

500

0 0:00

4:00

8:00

12:00

16:00

20:00

24:00

20:00

24:00

Time

(a) Number of control actions

165

30

20

10

0 0:00

4:00

(b)

8:00

12:00

16:00

Time

Figure 12.4 Simulation results for the disaster traffic: (a) throughput and (b) number of control actions assignment method does not change the frequency assignment. Our studies show the number of control actions by the proposed method is less than the others and the throughput of the proposed method is almost the same as the dynamic method.

12.3 Comprehensive evaluation plan In comprehensive evaluation of the HTS with the frequency flexibility, here, we mainly focus on the performance index, communication traffic, and link assignment methods.

12.3.1 Performance index In the comprehensive evaluation, we plan to calculate and compare the following performance indices: ● ● ● ●

Throughput Call loss rate Number of control actions (changing the frequency assignment) Frequency utilization efficiency of the total HTS system (for the four indices) and each user (for the first two indices).

166

Advances in communications satellite systems 2: ICSSC-2019

We have already confirmed that the proposed method has performed almost the same throughput as the dynamic method and the smaller number of control actions than the dynamic method. In the next step, we try to find an index and traffic features that show the high performance by using the conventional methods, not only for the proposed method.

12.3.2 Communication traffic In the comprehensive evaluation, we assume that there are various types of traffic in the HTS system, such as aircraft traffic, ships traffic, and very small aperture terminals (VSAT) traffic in a disaster. First, we assume that there are aircraft and disaster traffic in the HTS system and analyze time-varying traffic of the sum of them (the traffic of ships will be analyzed). For the aircraft traffic, we analyze the Collaborative Actions for Renovation of Air Traffic Systems (CARATS) Open Data [7] of November 9, 2015. For the disaster traffic, we analyze the SATCOM traffic data of the Great East Japan Earthquake provided by the Local Authorities Satellite Communications Organization of Japan. Note that the Great East Japan Earthquake occurred at 14:46 on March 11, 2011. Figure 12.5 shows an example of the time-varying traffic in the Kanto beam and Tohoku beam (beam numbers are 27 and 33, respectively, in Figure 12.2) obtained by a numerical simulation. In this graph, the two types of traffic are stacked and displayed. We note that in this analysis, the raw data of the aircraft and disaster traffic is adjusted multiplying some constants so that the traffic changes over time greatly. The amount of the aircraft traffic is more than that of the disaster traffic in the Kanto beam and vice versa in the Tohoku beam. Especially, the amount of disaster traffic rapidly increased when the earthquake occurred at 14:46. Thus, the HTS may not be able to accommodate all users’ traffic during the disaster. This rapid increase is also related to operations strategy described in Section 12.4.

12.3.3 Link assignment schemes In the comprehensive evaluation, three types of link assignment schemes are considered. Figure 12.6 shows these three schemes. In this figure, columns represent capacity and requested bandwidth. Each feature is as follows: ●

●

Ideal scheme In this scheme, DC can be controlled when the capacity is not enough to accommodate all users’ traffic. The DC is controlled every time according to users’ traffic; thus, the power consumption of the whole system including the DC and ground system is large. Immediate response scheme In this scheme, DC can be controlled periodically at the DC control timing determined in advance. In between the DC control period, link of traffic can be assigned immediately if the capacity is enough. Thus, some users may have to be kept waiting when the total request bandwidths exceed the capacity.

Frequency flexibility in high throughput satellites

Traffic (MHz)

250

167

Aircraft Disaster

200 150 100 50 0 0:00

4:00

8:00

12:00 Time

16:00

20:00

24:00

8:00

12:00

16:00

20:00

24:00

(a)

Traffic (MHz)

250

Aircraft Disaster

200 150 100 50 0 0:00

(b)

4:00

Time

Figure 12.5 Time-varying traffic of the sum of the aircraft and disaster. Blue and red columns represent the aircraft and disaster traffic of 24 h, respectively: (a) Kanto beam and (b) Tohoku beam ●

Waiting time scheme

In this scheme, all users have to be kept waiting until the DC control timing. Thus, the call loss of the traffic may be relatively large compared to the other two schemes.

12.4 Operations strategy for HTS system By utilizing a comprehensive evaluation result, we further consider operations strategy for the HTS system. We assume that we can estimate the traffic of each beam in advance. This information enables us to design the beam arrangement in the design phase of the HTS. We can also assign the appropriate bandwidth to each beam in advance and reduce the call loss of the traffic [8]. If the control period of the HTS is longer than the change period of the time-varying traffic and we can estimate the future traffic precisely, the predictive control of the frequency assignment based on the predicted traffic should work well. Furthermore, the scenario-based strategy is important for operating the HTS system. If the large disaster occurs, the traffic from a disaster area should have

168

Advances in communications satellite systems 2: ICSSC-2019 DC control timing

Link assigned!

Link assigned!

Link assigned!

Time User 1

User 3

User 2

Maximum bandwidth (capacity) Requested bandwidth

DC is controlled.

User 1

User 3 User 2 User 1

User 2 User 1

(a)

Figure 12.6 Three types of link assignment schemes: (a) ideal scheme, (b) immediate response scheme, and (c) waiting time scheme

DC control timing DC control period Waiting time for DC control User 3 is kept waiting.

Link assigned!

Link assigned!

Link assign failed.

Link assigned!

Time User 1

User 2

Maximum bandwidth (capacity) Requested bandwidth

User 3

DC is controlled.

User 1

User 2 User 1

User 3 User 2 User 1

(b)

Figure 12.6 (Continued )

User 3 User 2 User 1

Frequency flexibility in high throughput satellites

169

DC control timing DC control period

Waiting time for DC control Users are kept waiting. Request is accepted.

Request is accepted.

Request is accepted.

Link assigned!

Time User 1

User 2

User 3

Maximum bandwidth (capacity)

DC is controlled. User 3 User 2

Requested bandwidth

User 1

(c)

Figure 12.6

(Continued )

priority over the others. In this situation, the NOC needs to aggregate types of traffic and prioritize each traffic. By this strategy, the disaster traffic is processed with a high priority and the HTS reduces the bandwidth allocated to traffic other than the disaster one.

12.5 Conclusion In this chapter, we summarized our previous study and described the comprehensive evaluation plan of the HTS with the frequency flexibility. This evaluation plan will be performed soon. Our proposed method will also be demonstrated by using the Engineering Test Satellite-9 (ETS-9), which is planned to be launched in the fiscal year of 2021.

Acknowledgments This study was conducted under the commissioned research of the “Research and Development of Bandwidth-on-Demand High Throughput Satellite Communications System” by the Ministry of Internal Affairs and Communications of Japan. Communication traffic data in the Great East Japan Earth-quake was provided by the Local Authorities Satellite Communications Organization of Japan. CARATS Open Data were provided by the Ministry of Land, Infrastructure, Transport and Tourism of Japan.

170

Advances in communications satellite systems 2: ICSSC-2019

References [1] Elbert, B.R. Introduction to Satellite Communication. Norwood, MA: Artech House; 2008. [2] Hasan, M. and Bianchi, C. “Ka band enabling technologies for high throughput satellite (HTS) communications.” International Journal of Satellite Communications and Networking. 2016; 34(4); 483–501. [3] Vasavada, Y., Gopal, R., Ravishankar, C., Zakaria, G., and Ben Ammar, N. “Architectures for next generation high throughput satellite systems.” International Journal of Satellite Communications and Networking. 2016; 34(4): 523–546. [4] Kaneko, K., Nishiyama, H., Kato, N., Miura, A., and Toyoshima, M. “Construction of a flexibility analysis model for flexible high throughput satellite communication systems with a digital channelizer.” IEEE Transactions on Vehicular Technology. 2018; 67(3): 2097–2107. [5] Abe, Y., Okawa, M., Okada, K., Miura, A., and Toyoshima, M. “Performance study of frequency assignment with flexibility in high throughput satellites.” Proceedings of the 24th Ka and Broadband Communications Conference. 2018. [6] Abe, Y., Okawa, M., Miura, A., Okada, K., Akioka, M., and Toyoshima, M. “Performance evaluation of frequency flexibility in high throughput satellites: An application to time-varying communication traffic.” Proceedings of the 32nd International Symposium on Space Technology and Science (ISTS). 2019. [7] Ministry of Land, Infrastructure, Transport and Tourism of Japan. “Collaborative actions for renovation of air traffic systems (CARATS).” Available from http://www.mlit.go.jp/en/koku/koku_fr13_000000.html [Accessed September 27, 2019]. [8] Abe, Y., Tsuji, H., Miura, A., and Adachi, S. “Frequency resource management based on model predictive control for satellite communications system.” IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences. 2018; E101-A(12): 2434–2445.

Section 4

New satellite system architectures and components

This page intentionally left blank

Chapter 13

Satellite experiments on direct spectrum division transmission over multiple transponders Fumihiro Yamashita1, Daisuke Goto1, Yasuyoshi Kojima1, Hiroki Shibayama1, Hiroyuki Kobashi2 and Daiki Haraguchi2

We have been studying a direct spectrum division transmission (DSDT) technique that can divide a single carrier signal into multiple subspectra and assign them to dispersed frequency resources of the satellite transponder to improve the spectrum efficiency of the whole system. In a past study, we carried out fundamental satellite experiments on DSDT over a single transponder. This time, we conduct satellite experiments on DSDT over multiple transponders by using the latest DVB-S2X format signal. We expect using DSDT over multiple transponders will provide a new type of satellite communication service by raking unused frequency resources dispersed through multiple transponders. Key Words: satellite experiments; direct spectrum division transmission

13.1 Introduction The rapid adoption of network services requires more efficient wireless network infrastructures. Since satellite communications are the sole access network for aircrafts, vessels, and disaster-struck areas, broadband satellite links are needed to establish IP access networks or mobile backhaul systems [1,2]. In typical satellite communication systems, demand assigned multiple access (DAMA) [3] assigns the frequency resources of the satellite transponder to user earth stations (UESs) independently, so the total usage of the satellite transponder changes momentarily. The repeated acquisition and release of frequency resources among various services scatters the unused frequency resources of the satellite transponder and makes them 1

NTT Access Network Service Systems Laboratories, NTT Corporation, Yokosuka-shi, Kanagawa, Japan 2 SKY Perfect JSAT Corporation, Minato-ku, Tokyo, Japan

174

Advances in communications satellite systems 2: ICSSC-2019

individually insufficient to accommodate new users. This fragmentation of the frequency resources degrades frequency utilization efficiency. To tackle this problem, we propose the direct spectrum division transmission (DSDT) technique, as it better utilizes unused frequency resources of the satellite transponder [4]. Figure 13.1 shows the DSDT concept. On the transmitter (Tx) side, the DSDT divides the single carrier modulated signal into multiple subspectra in the frequency domain and assigns them to the unused frequency resources of the satellite transponder. On the receiver (Rx) side, the divided subspectra are recombined and the original signal is regenerated.

13.2 Operational principle of DSDT adapter 13.2.1 Spectrum-editing technique in transmitter Due to the band overlap between the dividing filters, the total bandwidth after spectrum division is greater than the original bandwidth [5]. Dividing filters have a special frequency response to suppress the bandwidth increment caused by spectrum division. Figure 13.2 shows an example applying the spectrum-editing technique in the transmitter. The dividing filter reduces the transition band of the modulated signal by compensating the amplitude characteristics of its spectrum to satisfy X ð1 þ bÞ  Bkd (13.1) ð1 þ aÞ  B ¼ k¼1

where B is symbol rate before spectrum division, a is its roll-off ratio, Bkd is the –3 dB bandwidth of kth subspectra, and b is the roll-off ratio of subspectra.

13.2.2 Synchronization technique in receiver The automatic phase control (APC), the automatic frequency offset compensation scheme (AFC), and the automatic gain control (AGC) are the key technologies that combine the divided subspectra and regenerate the original single carrier signal in the receiver.

Frequency band on satellite transponder Tx

Rx Spectrum combination

Spectrum division Satellite modem

Modulated signal

Divided sub-spectra

f Received signal

Figure 13.1 Overview of DSDT

Regenerated modulated signal

f

Satellite modem

Satellite experiments on direct spectrum division transmission

175

Bd = B/2

Bd

B= Bd + Bd Spectrum division

f

f B

(1 + β)Bd

(1 + β)Bd

BWorig = (1+α) B

(1 + α)Bd

(1 + α)Bd

B: Symbol rate Bd: –3dB bandwidth of sub-spectra, β: Roll-off factor of sub-spectrum

Figure 13.2 Spectrum-editing technique W1

(a)

W2

Before spectrum division

f

s1,2 Subspectrum1 f s2,1

Subspectrum2 f

s2,2

Subs spectrum3 3,1 (b)

Divided spectrum

θ

f

f1 f f2 (c)

Phase characteristics in Rx side

Figure 13.3 APC

13.2.2.1 Auto phase control The DSDT adapter estimates the phase differences between Rx subspectra from the transition bands between adjacent subspectra and compensates them, making the phase characteristics become continuous. The phase synchronization of the combined signal is established in the existing modem. Figure 13.3 shows the operational concept of the phase offset compensator. On the Tx side, subspectra k and k þ 1 originate from the

176

Advances in communications satellite systems 2: ICSSC-2019

same signal components in their transition band, Wk, due to band overlap between dividing filters k and k þ 1, as shown in Figure 13.3(a) and (b). Thus, no phase offset is present between adjacent subspectra in Wk on the Tx side. However, phase differences between the subspectra occur in transition band Wk on the Rx side. The phase compensator detects and compensates the averaged phase differences jk ¼ argðsðk; 2Þ; sðk þ 1; 1ÞÞ

(13.2)

where s(k,m) is generated in the transition band Wk from sub-spectrum k and * is the complex conjugate. In addition, m ¼ 1 is the higher band and m ¼ 2 is the lower band of the transition band. After phase compensation, the combined signal is converted into the time domain by IFFT.

13.2.2.2

Auto frequency control

Figure 13.4 shows the DSDT frequency offset compensation scheme. Our approach divides transition band Wk evenly into two bandwidths and calculates the averaged correlation value as follows: b¼

1 Xn1   ðUk;a Ukþ1;b  Uk;c Ukþ1;d Þ k¼1 n1

(13.3)

where n is the spectrum division number, Uk,a is the lower band, and Uk,b is the higher band of transition band Wk in subspectrum k, and Ukþ1,c is the lower band and Ukþ1,d is the offset (Df), the combined filter reduces the signal power of the Rx subspectra. If Df is positive, the signal component of Uk,c decreases while that of Uk,d increases, so b is a positive value. Similarly, if Df is negative, then b is a negative value. Our scheme calculates b before the combining filters in the frequency domain and compensates the frequency offset before the FFT circuit in the time domain if b is not zero. When frequency offset Df ¼ 0, the averaged correlation value b becomes zero.

13.2.2.3

Auto gain control

If the spectral power density of the received subspectra is not equal in the frequency domain, the algorithm of APC/AFC will not work as expected. Thus, the spectral density of the received subspectra must be equalized before spectrum combination. We implement auto gain control (AGC) in the frequency domain for this purpose. Figure 13.5 shows its operating principle. The averaged power density of all subspectra Pall is calculated as Pall ¼

1 Xn P i¼1 i n

(13.4)

where Pi is the averaged spectrum power density at the pass-band of subspectrum Si. To equalize the spectral density in the frequency domain, the control value of AGC gi to the subspectrum Si is calculated as gi ¼

Pall Pi

(13.5)

Satellite experiments on direct spectrum division transmission W1

177

W2

f (a)

Before spectrum division U1,c SubU1,d spectrum1 f U2,b

U2,c

SubS2,d spectrum2

U2,a

f U3,b Subspectrum3

U3,a f

(b)

After spectrum division Δf U1,c SubU1,d spectrum1 Δf U2,b

f U2,c U2,d Subspectrum2

U2,a

Δf

f

U3,b Subspectrum3 U 3,a f (c)

Rx sub-spectra (Δf >0)

Figure 13.4 AFC

13.3 Satellite experiments 13.3.1 Experimental setup We carried out satellite experiments to confirm the practical performance of DSDT. Figure 13.6 shows the experimental setup. Figure 13.7 shows the simple graphical user interface (GUI) for setting the parameters of spectrum division/combination. The firmware automatically calculates the spectrum division/combination parameters and sets them to the FPGA once the operator sets the signal parameters of the symbol rate and the roll-off ratio of the input single carrier signal, available

178

Advances in communications satellite systems 2: ICSSC-2019 (1 - α)B1 (1 - α)B2 p11 p12 p13 p14 (1 - α)B4 (1 - α)B3 p21 p22 p23 p31 p32 p33 p41 p42 p43 S1

S2

4 Pall = 1 Pk 4 k=1 Pall P1 S1

S3

Pall

S4

Pall P3

Pall

P2

S2

S3

S4

P4

f

f

Figure 13.5 AGC

Ku band (14/12GHz)

Satellite modem

DSDT adapter

Conv.

TWTA

Satellite modem

DSDT adapter

Conv.

LNA

BER tester

Noise generator

Figure 13.6 Configuration of satellite experiments using JCSAT-3A [6] system bandwidth, the spectrum division number, each subspectrum symbol rate, and the center frequencies of output divided signals.

13.3.2 Spectrum division over multiple transponders Figure 13.8 shows the experimental setup. We mainly measured bit error ratio (BER) performance while changing various parameters. We also checked if the 4K movie content could be delivered using the DSDT technique as expected. Figure 13.9 shows example spectra that are arranged over multiple satellite transponders. Figure 13.9(a) is the spectrum at the output of the Tx modem. Figure 13.9(b) is the divided spectrum at the output of the Tx DSDT adapter. Conversely, Figure 13.9(c) is the received signal at the input of Rx DSDT adapter. Figure 13.9(d) is the combined signal at the input of the Rx modem. Figure 13.9(c) shows the guard band

Satellite experiments on direct spectrum division transmission

179

Figure 13.7 GUI for setting parameters

RF Unit GUI

4K Movie

4K Movie

Modem

Modem DSDT adapter

DSDT adapter

(a)

Tx side

GUI

(b)

Rx side

Figure 13.8 Experimental setup between the two transponders is 4 MHz. Due to the influence of the on-board analogfilters, the right edge of the second subspectrum and the left edge of the third subspectrum are slightly distorted, partially affecting the combined spectrum (d). In practice, on-board analog equipment performance and amplification gain differed transponder to transponder. Assuming this practical situation, we measured BER performance while experimentally changing the spectrum density difference DG (dB, see Figure 13.10) between two transponders. Figure 13.11 shows BER performance with 20 M-baud/QPSK modulation/LDPC R¼3/4; the performance degradation corresponds to the increments of DG because the AGC in the DSDT

180

Advances in communications satellite systems 2: ICSSC-2019

20 MHz (symbol rate)

(a)

Tx Signal (Txmodem out)

(b)

Tx signal (TxDSDT adapter out)

(d)

Rx signal (Rx modem in)

4MHz (Guard band) Transponder 2 Transponder 1

(c)

Rx signal (Rx DSDT adapter in)

Figure 13.9 Spectra over multiple satellite transponders

G (dB)

f Figure 13.10 Spectrum density difference DG (dB) between adjacent transponders adapter amplified the level of noise in the smaller subspectrum so that Eb/N0 degraded. Figure 13.12 shows Eb/N0 degradation (¼ DEb/N0) versus DG. As Figure 13.12 illustrates, to suppress D Eb/N0, Tx spectral density must be initially controlled subspectrum by subspectrum to become almost DG ¼ 0 dB. We measured BER performance after controlling Tx spectral density subspectrum by subspectrum for DG in the receiver to be 0 dB. At first, we measured BER performance assuming the fundamental three types of spectrum arrangement over satellite transponders, as shown in Figure 13.13. Figure 13.13(a) is an example

Satellite experiments on direct spectrum division transmission

181

10–3 0 dB 3 dB 6 dB 9 dB 12 dB 15 dB

10–4

BER

10–5 10–6 10–7 1.8dB

10–8

Eb/N0 (dB)

0

2

4

6 8 10 Eb/N0 (dB)

12

14

Figure 13.11 BER performances while changing DG (dB)

10

Eb/N0 (dB)

8 6 QPSK 8PSK 16APSK 32APSK 64APSK

4 2 0 0

2

4

6

8 10 G (dB)

12

14

16

Figure 13.12 DG (dB) versus DEb/N0 (dB) of a signal arrangement without a spectrum division. Figure 13.13(b) shows a signal arrangement with spectrum division over a single transponder. Figure 13.13 (c) shows a signal arrangement with spectrum division over multiple transponders. Detailed signal conditions are shown in Table 13.1. The symbol rate was set at 20 M-bauds. Its roll-off ratio was 0.05. DSDT divided a 20-Mbaud single carrier signal into two 10-Mbaud sub-spectra, as shown in Figure 13.13(b) and (c). QPSK, 8PSK, 16APSK, 32APSK, and 64APSK modulation types were changed. LDPC (R ¼ 3/4) was applied for forward error correction (FEC). Figure 13.14 shows the representative BER performances in the satellite experiments with the parameters

182

Advances in communications satellite systems 2: ICSSC-2019

f (a)

Spectrum arrangement without spectrum division

(b)

Spectrum arrangement over a single transponder

f

f (c)

Spectrum arrangement over multiple transponders

Figure 13.13 Spectrum arrangements for satellite experiments

Table 13.1 Experimental parameters. Adapted from [7] No. of div. Mod. type Symbol rate Spectrum arrangement

1 (no division), 2 QPSK/8PSK/16APSK/32APSK/64APSK (DVB-S2X) 20 MHz (roll-off ratio: 0.05) (a), (b), or (c) in Figure 13.13

listed in Table 13.1. BER performances over a single transponder or multiple transponders were almost the same as that of a single-carrier. Next, we measured BER performance assuming the other three types of spectrum arrangement over satellite transponders, as shown in Figure 13.15. Figure 13.15(a) is an example of signal arrangement with spectrum evenly divided into three subspectra. Figure 13.15(b) shows a signal arrangement with spectrum evenly divided into four subspectra. In addition, Figure 13.15(c) shows a signal arrangement with spectrum unevenly divided into four subspectra: 1:2:1:4. Detailed signal conditions are shown in Table 13.2. The symbol rate was set at 30 M-bauds. Its roll-off ratio was 0.2. DSDT divided a 30-Mbaud single carrier signal and arranged subspectra over multiple transponders, as shown in Figure 13.15(a)– (c). QPSK, 8PSK, 16APSK, and 32APSK modulation types were changed. LDPC (R ¼ 3/4) was applied for FEC. Figure 13.16 shows the representative BER performances in the satellite experiments with the parameters tabulated in Table 13.2. BER performances of Figure 13.16 (5)–(16) over multiple transponders were almost the same as that of a single-carrier as shown in Figure 13.16 (1)–(4).

Satellite experiments on direct spectrum division transmission (1)QPSK (a) (2)8PSK (a) (3)16APSK (a) (4)32APSK (a) (5)64APSK (a) (6)QPSK (b) (7)8PSK (b) (8)16APSK (b) (9)32APSK (b) (10)64APSK (b) (11)QPSK (c) (12)8PSK (c) (13)16APSK (c) (14)32APSK (c) (15)64APSK (c)

10–3 QPSK 8PSK

BER

10–4

32APSK

10–5

10–6

16APSK

10–7

10–8

0

2

4

6

64APSK

8

10

12

14

Eb/N0 (dB)

Figure 13.14 BER performances

f (a)

Spectrum evenly divided into 3 sub-spectra

f (b)

Spectrum evenly divided into 4 sub-spectra

f (c)

Spectrum unevenly divided into 4 sub-spectra

Figure 13.15 Spectrum arrangements for satellite experiments

183

184

Advances in communications satellite systems 2: ICSSC-2019

Table 13.2 Experimental parameters No. of div. Mod. type Symbol rate Spectrum arrangement

3, 4 QPSK/8PSK/16APSK/32APSK (DVB-S2X) 30 MHz (roll-off ratio: 0.2) (a), (b), or (c) in Figure 13.15

10–3

(1) QPSK ref. (2) 8PSK ref. (3)16APSK ref. (4) 32APSK ref. (5) QPSK (a) (6) 8PSK (b) (7) 16APSK (a) (8) 32APSK (a) (9) QPSK (b) (10) 8PSK (b) (11) 16APSK (b) (12) 32APSK (b) (13) QPSK (c) (14) 8PSK (c) (15) 16APSK (c) (16) 32APSK (c)

8PSK QPSK

10–4

32APSK

BER

10–5

10–6

10–7 16APSK 10–8

0

2

4

6 8 Eb/N0 (dB)

10

12

14

Figure 13.16 BER performances

13.4 Summary We carried out satellite experiments on DSDT over multiple transponders by using the latest DVB-S2X format signal. The satellite experiments confirmed that BER performed the same with or without spectrum division and the DSDT technique is practical. In the near future, using DSDT over multiple transponders will provide a new type of satellite communication service by raking unused frequency resources dispersed over multiple transponders.

References [1] Wu, W.W., Miller, E.F., Pritchard, W.L., and Pickholtz, R.L. “Mobile satellite communications.” Proceedings of the IEEE. 1994; 82(9)1431–1448. [2] Casoni, M., Grazia, C.A., Klapez, M., Patriciello, N., Amditis, A., and Sdongos, E. “Integration of satellite and LTE for disaster recovery.” IEEE Communications Magazine. 2015; 53(3): 47–53.

Satellite experiments on direct spectrum division transmission

185

[3] Barberis, G. and Brignolo, R. “Capacity allocation in a DAMA satellite system.” IEEE Transactions on Communications. 1982; 30(7): 1750–1757. [4] Abe, J., Yamashita, F., Nakahira, K., and Kobayashi, K. “Direct spectrum division transmission for highly efficient frequency utilization in satellite communications.” IEICE Transactions on Communications. 2012; E95.B (2): 563–571. [5] Suzuki, Y, Taromaru, M., Yano K., and Ueba. M. “Proposal of band-limited divided-spectrum single carrier transmission for dynamic spectrum controlled access in ISM band.” IEEE 20th International Symposium on Personal, Indoor and Mobile Radio Communications. 2009 [6] See https://www.jsat.net/jp/contour/jcsat-3a.html [Accessed November 7, 2020]. [7] Ugolini, A., Modenini, A., Colavolpe, G., Picchi, G., Mignone, V., and Morello, A. “Advanced techniques for spectrally efficient DVB-S2X systems.” 7th Advanced Satellite Multimedia Systems Conference and the 13th Signal Processing for Space Communications Workshop (ASMS/SPSC). 2014

This page intentionally left blank

Chapter 14

User terminal wideband modem for very high throughput satellites Steven Kisseleff1, Nicola Maturo1, Symeon Chatzinotas1, Helge Fanebust2, Bjarne Rislow2, Kimmo Kansanen3, Matthieu Arzel4 and Hans C. Haugli5

The continuous increase of the demand for high data rate satellite services has triggered the development of new high-end satellite modems, which are capable of supporting a bandwidth of up to 500 MHz. For commercial application, the downlink from low Earth orbit (LEO) sensors and observation satellites is of a special interest. Such satellites should be capable of recording gigabytes of data and transferring it to the ground stations within a few minutes since the satellite is only visible for a short time at such low altitudes. This implies a very fast and reliable information processing at the terminal. For this, it would be beneficial to utilize the entire 1500 MHz spectrum of the extended Ka-band. In this context, the design of the modem architecture is very challenging. This problem is addressed in this chapter for the first time. We develop a new terminal modem architecture, which is expected to support a data rate in the range between 25 Msps and 1400 Msps. Through this, the receiver can easily adapt to changes in the data rate according to the traffic requirements. Furthermore, a simulator tool is developed, which is used for a numerical performance evaluation of the individual components and the whole system. Key Words: wideband modem; high throughput satellite; MODCOD; equalisation; synchronisation

1

Interdisciplinary Centre for Security, Reliability and Trust, University of Luxembourg, Luxembourg WideNorth, Lysaker, Norway 3 Department of Electronic Systems, Norwegian University of Science and Technology, Trondheim, Norway 4 Departement Electronique, IMT-Atlantique Bretagne-Pays de la Loire, Brest, France 5 Space Norway, Oslo, Norway 2

188

Advances in communications satellite systems 2: ICSSC-2019

14.1 Introduction The increasing demand for high data rates in commercial applications of satellite communication, such as the downlink from low Earth orbit (LEO) sensors and observation satellites, motivates the design of novel terminal modems, which are capable of efficiently operating in the extended Ka-band and reliably processing the received data. Here, the main challenges for the modem design appear in the context of extremely large signal spectrum, which can be up to 1.5 GHz. The development of the terminal modem that supports up to 500 MHz for commercial high data rates has been addressed in various projects, for example, refer to [1–3]. In contrast, we impose even more challenging and future-oriented requirements in this work by assuming a substantially wider target signal bandwidth, that is 1.5 GHz, and much higher user data rate of up to 1.4 Gsps (with a root-raised cosine filter and excess bandwidth of at most 5%). In this context, the design of the modem architecture faces the following challenges: ●

●

●

●

●

Parallel processing is required in order to support high baud rates. However, this may lead to large and expensive FPGAs. It may not be possible to utilize the optimum signal processing and synchronization algorithms due to the computational complexity and corresponding delays. Hence, a careful selection of the algorithms is required, which implies frequent trade-offs between the performance and the complexity. The modem should be able to operate at a low signal-to-noise ratio (SNR), for example, at
2 dB. The modem operation in such noisy environments is very challenging for the considered scenario since there are additional limitations of the processing power due to the high baud rates. The modem should support high MODCODs based on high order symbol constellations, for example, 16APSK, 32APSK, and 64APSK. In order to avoid packet loss, the terminal synchronization needs to be very accurate and fast [4,5]. However, due to the computational complexity, it may not be possible to apply the conventional high-performance synchronization methods here. Hence, less complex and correspondingly less accurate methods are preferable, which implies that a new receiver architecture needs to be specifically designed in order to take into account the individual constraints related to these low-complexity methods.

In this chapter, we propose a new modem architecture, which is expected to support a data rate in the range between 25 Msps and 1400 Msps. This architecture enables a quick adaptation of the receiver to changes of the data rate according to the traffic requirements. In particular, we address the selection and analysis of the following signal processing components: ● ● ● ● ●

timing synchronization, equalization, frequency offset compensation and tracking, frame synchronization, and demodulation/decoding.

User terminal wideband modem for very high throughput satellites

189

Furthermore, we develop a simulator tool, which is used for a numerical performance evaluation of the individual components and of the whole system. This chapter is organized as follows. In Section 14.2, the system model is described and the implications for the design of a wideband modem are explained. The design of the individual modem components including frequency, timing, and frame synchronization, equalization, and demodulation is discussed in Section 14.3. The numerical results based on the developed simulator tool are presented in Section 14.4. Finally, Section 14.5 concludes the chapter.

14.2 System model The very large bandwidth and baud rate pose many challenges for the design of the terminal modem. In addition, some of the less-known effects and hardware impairments become substantial and make it difficult to fulfill the requirements of the system. In the following, we address these challenges, impairments, and requirements.

14.2.1 Challenges and impairments The following challenges are especially crucial for the design of the modem: 1. 2. 3.

parallel signal processing due to extremely high baud rates, which requires large FPGAs; trade-offs between performance, latency, and complexity; high-frequency selectivity due to varying amplitude and phase responses over the large signal bandwidth.

In addition, the system performance can degrade due to the following impairments: 1. 2. 3. 4.

carrier frequency offset and drift, timing offset, clock frequency offset and drift, and frequency selectivity of the cables.

For the carrier frequency offset and drift, we assume 3 MHz and 5 kHz/s, respectively. Depending on the selected baud rate, which is between 25 Msps and 1400 Msps in this work, the maximum offset can be up to 12% and 0.2%, respectively. These values indicate that in absence of an accurate frequency synchronization, the frequency error would be too large for a feasible frame lock. The constant timing offset is modeled as a random variable, which is uniformly distributed between 0.5 T and 0.5 T , where T is the symbol duration, since this offset mostly depends on the switching time, which is uniform in this range. The impairments (1), (2), and (4) are well-known problems addressed by various works in the past. In general, most of these impairments can be dealt with using the traditional methods of receiver synchronization or equalization. However, a clock frequency offset has rarely been considered. In fact, this offset may not only

190

Advances in communications satellite systems 2: ICSSC-2019

result from the hardware impairment of the onboard clock but from the spectrum spreading as a side effect of the Doppler shift. This imperfection has rarely been addressed in the literature due to a typically assumed much lower signal bandwidth, so that the spectrum spreading is not crucial even with a very heavy Doppler shift. However, in the considered scenario, the resulting clock frequency offset leads to a timing drift, which may lead to a significant performance degradation. According to our estimations, this clock frequency offset due to the Doppler shift is at most 30 ppm for the target application, that is, LEO satellites and maximum baud rate of 1400 Msps. The respective change of the Doppler shift, which comes along with the motion of the satellite, lead to a clock frequency drift, which corresponds to a second-order timing drift. This drift is expected to be below 1 ppm/s. Furthermore, the frequency selectivity of the cables is not negligible due to the large bandwidth and the corresponding difference between the lowest and the highest signal frequency within the frequency band. We employ a cable slope model in order to take this effect into account. A cable carries the received downmixed signal to the baseband unit. This process can be viewed as an additional frequency-selective filtering. Here, we assume that the cable slope can be up to
10 dB over the whole frequency band. Note that the received noise undergoes this filtering as well. In addition, a slight performance degradation can occur due to a possible I-Q imbalance in the hardware of the terminal. However, this effect can be precompensated before the main operation of the terminal. Correspondingly, we omit this effect from our consideration.

14.2.2 Requirements As mentioned earlier, we aim at providing the service at a wide range symbol rates, that is, between 25 Msps and 1400 Msps. For a higher flexibility of the system, we need to ensure the possibility of changing the symbol rate “on the fly.” Hence, the system should be able to work with any symbol rate using the same methods and possibly the same parameters. For the air interface, we assume the DVB-S2X standard (cf. [6]) with the superframing format IV [7]. This format enables some advanced functionalities, for example, beam hopping, which can be useful in future scenarios and applications. The main properties of this superframe format, that are important for our application, are the equidistant position of the pilot fields and an extended header. The header contains in total 1,440 known symbols including an indicator of the superframe format, which can be assumed to be known as well. The pilot fields contain 36 known symbols repeated after every 1,440 data symbols. Both header and pilots are modulated via QPSK modulation. In this work, we focus on very harsh signal propagation conditions. Correspondingly, the system should in principle operate in a wide range of signalto-noise ratios, for example, Es/N0 between
10 dB and 30 dB. However, for a more practical application, we stick to the DVB-S2X standard [6] and employ the MODCODs accordingly together with the target Es/N0 values, which correspond to each of these MODCODs. In particular, we select the following schemes of phase-

User terminal wideband modem for very high throughput satellites

191

shift keying (PSK) modulation: QPSK, 8PSK, and 16 APSK. The target Es/N0 is selected between
2 dB and 13 dB. These values are assumed in order to guarantee a sufficient reliability of symbol detection, that is, very low frame error rates (FER) below 10
5, in the case of perfect synchronization. The residual frequency error after the offset compensation is expected to be below 5 kHz. This corresponds to 2 10
4 and 3.6 10
6 of the respective baud rates above. Hence, an extremely accurate frequency synchronization is required in order to almost completely mitigate the frequency offset. For the timing error correction, there are no specific requirements. However, in order to enable both frequency and frame synchronization, the timing offset and drift compensation should be sufficiently accurate. Hence, we assume that the residual timing offset should be below 10% of the symbol interval. In this case, the impact of the symbol misalignment on the frequency and frame synchronization is relatively low. Although the 10% of the symbol interval can potentially lead to a significant performance degradation in terms of symbol detection due to intersymbol interference and magnitude decrease, this effect can be mostly compensated by introducing an equalization block.

14.3 Modem design 14.3.1 Architecture In this work, we consider both nondata-aided (NDA) methods for the initial (coarse) synchronization and data-aided (DA) methods for the fine synchronization. For the DA methods, we rely on a successful frame lock, so that the pilot symbols transmitted along with the data payload can be utilized for a more reliable parameter estimation. In order to cope with the mentioned difficulties and guarantee the fulfillment of the requirements, we propose the following modem architecture, see Figure 14.1. After the analog-to-digital converter (ADC), the signal is fed into the adaptive matched filter, which is connected to the timing error detector, so that the estimated timing error is used in order to adjust the coefficients of the matched filter. This method is preferable since no interpolation of the consecutive samples is

Timing control

TED Frame sync

ADC

×

Matched filter

ALC

Equalize

Phase track

Deformat & de-scramble

NCO/DDS Coarse Frequency estimation Fine, Pilots

Figure 14.1 Proposed modem architecture

Soft decision LLR

FEC decoder

192

Advances in communications satellite systems 2: ICSSC-2019

needed in order to account for the timing error. After the timing error correction, the signal is fed into the equalizer, which mitigates the distortion imposed by the frequency-selective cables, etc. During the initialization phase, the equalizer is in the NDA mode and has to be switched off, since it is typically very vulnerable to the frequency offset, which is not yet compensated at this point. Hence, the signal is guided to the coarse frequency estimator, which provides an estimate of the frequency offset. This estimate is used in order to compensate (at least) a part of the frequency offset for the subsequent symbols. After a successful frame synchronization, it is possible to use the structure of the frame in order to enable DA synchronization. As mentioned earlier, we utilize the superframe format IV. Correspondingly, the structure of the frame is known to the receiver and incoming symbols can be de-formated and de-scrambled. Through this, the data are separated from the pilots and guided to further processing into the FEC decoder via soft log-likelihood calculation. After the decoding, the decoded bits are assembled into a bit stream.

14.3.2 Timing synchronization As described in Section 14.3.1, the first block of the synchronization chain is the time synchronization. Because of the very high symbol rate and the much lower clock rate employed in the potential hardware implementation, the matched filter is implemented in a parallel fashion. The output of the matched filter is then passed to the timing error detector (TED) block that will evaluate the timing error. The TED is an NDA algorithm, which means that the algorithm can be executed without any knowledge of the transmitted signal. Among the various NDA TED methods, we select the Gardner algorithm [8], which is known to perform well even in case of a relatively large frequency offset. The expression applied by Gardner TED algorithm is 

1 eðkÞ ¼ x k
T þ t ½xððk
1ÞT þ tÞ
xðkT þ tÞ 2  (14.1)

1 þy k
T þ t ½yððk
1ÞT þ tÞ
yðkT þ tÞ: 2 In this expression, x (kT þ t ) and y (kT þ t ) are in-phase and quadrature components of the input signal to the timing error detector and t is the estimated timing error. Differently from the classical scheme, where the timing error detection is typically followed by the interpolation of the subsequent symbols, we combine the timing error detection and the matched filter in one signal processing block. The functionality of this block is as follows. Since the filter taps can be stored with a very high oversampling factor, it is possible to update the matched filter according to the estimated timing offset. For this, the matched filter is sampled with various timing offsets and the resulting filter taps are stored in the memory. Then, the index of the best set of filter coefficients is determined using the TED output. This set is loaded from the memory and used for filtering. Hence, the interpolation of the subsequent symbols is implicitly taken care of by adapting the matched filter. Correspondingly, significant complexity savings are achieved.

User terminal wideband modem for very high throughput satellites

193

In order to account for the processing delay within the timing detection loop, we introduce a buffer of 16 symbols, so that the output of the TED algorithm is delayed by 16 symbols before it reaches the loop filter. Through this, we avoid overestimating the performance of the TED. The main drawback of this approach is the limited resolution of the matched filter, so that the accuracy of the TED is bounded by the number of possible sets of coefficients. In particular, the set, which corresponds to the closest timing offset, is selected, which provides the lower bound for the performance of the error compensation. The worst case of this effect clearly happens when the output value of the TED falls exactly in between two closest sets of the stored filter taps. In this case, the residual uncompensated time offset scales with the reciprocal oversampling factor of the stored filter, for example, the maximum offset is 1/4 with oversampling factor 2 and 1/8 with oversampling factor 4, etc. Hence, this imperfection can be mitigated by a sufficient oversampling, which leads to a trade-off between the accuracy and the available storage size. According to our investigations, the degradation of the signal quality in terms of SNR becomes negligible even with an oversampling factor as low as 8. The convergence of the TED algorithm is shown for an example frame with QPSK modulation in Figure 14.2. Here, the initial timing offset is set to zero and we utilize the following parameters: damping factor is 3, detector gain is 4, and loop bandwidth is 10
3. We observe that after some time the algorithm starts to track the timing drift, which results from the clock frequency offset, as explained earlier. Nevertheless, there are still some deviations from the true timing offset,

1 0.9

Relative timing offset

0.8 0.7 0.6 0.5 0.4 0.3 0.2 Estimated offset True offset

0.1 0

0

1

2

3

4

5 6 Sample index

7

8

9

Figure 14.2 Example of timing drift tracking using TED

10 x105

194

Advances in communications satellite systems 2: ICSSC-2019

which produces the mentioned degradation of the signal quality. This degradation is typically below 1 dB.

14.3.3 Frequency synchronization 1.

2.

Coarse frequency synchronizers: The traditional method of coarse frequency synchronization is a quadricorrelator (QC, [9]). QC comprises two analog filters and a subsequent correlation of the outputs of the two filters. The so-called balanced (or optimized) QC performs much better than the unbalanced QC (i.e., without optimization of the filters’ coefficients). The resulting filters of the balanced QC are matched filter and its derivative. The implementation of two filters imposes a high hardware complexity, which makes the use of the QC impractical for the considered wideband scenario. In addition, QC has been shown to provide a sufficient independency of the timing error only with oversampled signals. Another method of coarse frequency synchronization is delay-and-multiply (D&M, [10]). Despite a very low complexity of this method, it has a drawback of performing well only in case of oversampling similarly to the QC. Hence, in the considered architecture the coarse frequency synchronizer needs to be deployed after the equalizer since the output of the equalizer is already in symbol domain. In addition, the performance of both methods is not sufficient, since a very large size of the observation window is required in order to satisfy the system requirements. This leads to a long delay. Since running the coarse frequency synchronization after the equalizer would guarantee that the frequency synch algorithm wouldn’t be subject to the cable slope, we have also investigated an alternative method of coarse frequency synch using the Rife&Boorstyn algorithm, which can operate at the symbol level as well (cf. [11]). This algorithm is based on a peak search in the frequency domain. For this, a fast Fourier transform (FFT) method with a sufficiently fine frequency spacing is utilized. This method is substantially more complex than D&M. However, it does not require oversampling and the sufficient accuracy can be achieved with substantially lower sizes of the observation window so that both processing delay and complexity can be kept low with this method. Also, it is possible to achieve very accurate estimation results already at the coarse synchronization stage, so that a fine frequency synchronization may even be skipped in some cases. This method shows a clear advantage compared to the other methods so that we select the R&B method for the coarse frequency synchronization in our architecture. Fine frequency synchronizers: D&M can be easily adjusted to a DA frequency estimator. Such a DA estimator would be applicable for a fine frequency estimation. However, according to our observation, D&M in fine frequency synchronization mode with a given number of pilots does not reach the performance requirements. Instead, the following methods have been considered first: Mengali&Morelli, Fitz, Luise&Reggiannini, and Lovell&Williamson methods [10]. Among these methods, a trade-off between complexity and accuracy has

User terminal wideband modem for very high throughput satellites

195

10-2

Normalized frequency variance

After coarse freq synch After fine freq synch

10-4

10-6

10-8

10-10

10-12 -2

0

2

4 6 ES/N0 (dB)

8

10

12

Figure 14.3 Frequency error variance after coarse and fine frequency synchronization using R&B method, respectively. A cable slope of
10 dB has been assumed been made so that Luise&Reggiannini (L&R) method has been selected for a deeper analysis. As known from the literature, L&R method can reach the CRB if the number of inner summations is at least half of the window size. Unfortunately, this number of inner summations also impacts the maximum offset range, which can be compensated by the algorithm. Due to a larger residual offset after methods like D&M and QC, this parameter corresponding to the number of summations needs to be selected sufficiently low, so that the CRB cannot be reached and the performance is typically relatively bad. On the other hand, when using R&B for coarse frequency synchronization, the residual offset is usually very low, which makes L&R applicable. However, R&B can be used in DA mode as well and provides an even better performance compared to L&R especially in presence of frequency-selective channels. A distinct advantage of this strategy is that no additional complexity for the implementation of the fine frequency synchronization is required since R&B is already assumed for the coarse frequency synchronization. The normalized residual frequency variance after the R&B method in coarse and fine synchronization mode is shown in Figure 14.3. Here, we assume a cable slop of
10 dB over the whole frequency band of 1.4 GHz. For the coarse frequency synchronization, we consider a window of 213 symbols, whereas for the fine synchronization, we assume a window size of 1,440 symbols, that are associated with the

196

Advances in communications satellite systems 2: ICSSC-2019

known symbols of the superframe header*. Obviously, a large residual frequency variance occurs for very low SNR values, which is due to a wrong peak selection in the frequency domain. Furthermore, we observe that the coarse estimation achieves a better performance at high SNR compared to the pilot-based fine estimation. This is due to a larger window size utilized by the coarse estimation as mentioned earlier. Correspondingly, in the higher SNR regime, the use of a fine frequency synchronization is not beneficial. In addition, we observe that the frequency variance with coarse frequency synchronization is approx. 210
12. Hence, the residual frequency offset is in the range of to a few kHz in case of 1.4 GHz. After the fine synchronization, we employ a phase tracker in order to compensate the phase offset and possible phase noise. For the phase tracker, we employ a secondorder phase-locked loop (PLL). This phase tracker can be employed further in order to fine-tune the frequency estimation by calculating the difference between the estimated phase offsets of the consecutive samples and deducing the frequency offset, which caused this relative phase difference. With this strategy, the maximum residual frequency offset reduces to extremely low values below 10
8 corresponding to less than 14 Hz at 1.4 GHz, so that the frequency variance is in the order of 10
15
10
16.

14.3.4 Frame synchronization The frame synchronization is usually done via correlation of the available pilot symbols with the received signal. In presence of a frequency offset, each symbol experiences a phase rotation, which prevents from applying a straightforward correlation. However, a small frequency offset may not significantly affect the correlation performance. Hence, a noncoherent postdetection integration (NCPDI) approach can be employed (cf. [12,13]), where a sequence of pilot symbols is split into small blocks. Each block is correlated with the received signal at the respective position of the block. Then, the results of all blocks are combined coherently. Hence, the phase rotation from block to block is eliminated, which leads to the noncoherent integration. This method has been thoroughly investigated in the past. For our design purpose, we utilize the expressions for the false alarm and missed detection probabilities, that is, Prfa and Prmd, respectively, as well as the so-called CHILD’s rule [14], which indicates the maximum number of pilot symbols per block: dKL

Prfa ¼ e
Pr

Prmd

1 ¼1
2Df

1K

*

  L
1 X 1 dKL m m! Pr m¼0 ð Df
Df

QL

  pffiffiffiffiffiffi pffiffiffi! sinc ðKDf Þ KL d    sinc ðDf Þ  s ; s df

3 ðCHILD0 s ruleÞ 8DfT

(14.2)

(14.3) (14.4)

At the input of the FFT, we apply also zero-padding so that the total number of input symbols including zeros is a power of 2. Through this, a better signal resolution in the frequency domain is obtained.

User terminal wideband modem for very high throughput satellites

Number of pilot symbols

103

197

25 Msps 50 Msps 100 Msps 200 Msps 500 Msps 1000 Msps 1400 Msps

102

101 –2

0

2

4 6 ES/N0 (dB)

8

10

12

Figure 14.4 The total required number of pilots for a frame lock using NCPDI under assumption of 3 MHz frequency offset where K and L are the number of pilots per block and the number of blocks, respectively. Pr stands for the received signal power including the noise with the variance s2, and 2s2 ¼ N0 is the noise power density. The threshold for the decision making of the frame acquisition is denoted d. In addition, QL( ) is the Marcum-Q function of order L and f is the maximum absolute frequency offset. Using these equations, it is possible to determine the optimal threshold for the frame lock and the optimal number of blocks, that is, total required number of pilots so that the target requirements on Prfa and Prmd are satisfied. Since the design of the frame synchronization can be done offline, we find the optimal set of parameters, that is, K, L, and d, via full search. The results for the total required number of pilots, that is, K L, is depicted in Figure 14.4 for various SNR values and baud rates. We observe, that the required number of pilots reduces with increasing baud rate. This is due to a constant frequency offset of 3 MHz so that the relative offset reduces. Correspondingly, the maximum value of K increases according to the CHILD’s rule and the correlation of each window becomes more and more accurate. Also, the number of pilots reduces with increasing SNR since the correlation of each window becomes more accurate and reliable. For
2 dB, the total number of pilots is in the range between 96 and 1338 symbols depending on the baud rate. Fortunately, these values are lower than the number of symbols in the header of a superframe, which is 1,440, so that NCPDI can be applied exclusively to the header without taking into account the pilot fields spread across the superframe. Hence, the frame synchronization is straightforward and has a relatively low complexity.

198

Advances in communications satellite systems 2: ICSSC-2019

After the frame lock has been acquired, most of the NDA methods can be switched off, that is, coarse frequency synchronization and blind equalization. Instead, the DA methods are utilized, which make use of the pilots and are therefore more reliable. In particular, fine frequency synchronization and DA equalization are switched on (if needed)

14.3.5 Equalization The main task of the equalization for the envisioned application is to reduce the signal distortion imposed by imperfect transmit/receive filters and cable slope. In this work, the equalization component is placed after the timing error correction and before the frequency synchronization. On the one hand, the equalizer would mitigate the intersymbol interference which results from the residual timing offset. On the other hand, the equalization is needed for an accurate operation of the frequency synchronization since the peak selection in frequency domain is sufficiently accurate only in presence of a reasonably equalized received signal. We consider the following methods: ● ● ● ●

Normalized Normalized Normalized Normalized of NCMA.

Least Mean Square Algorithm (NLMS) [15], Block Least Mean Square Algorithm (NBLMS) [15], Constant Modulus Algorithm (NCMA) [16], and Block Constant Modulus Algorithm (NBCMA) as an extension

All four methods are based on a gradient descend method, that is, the filter coefficients fn are updated iteratively in each step n in the direction of the gradient rn, which is calculated from the observed samples. These samples are stored in the signal vector xn. The update is done via f nþ1 ¼ f n þ mrn

(14.5)

where m is a step size in the range between 0 and 1. The first two methods (NLMS and NBLMS) employ pilot symbols, whereas the last two methods (NCMA and NBCMA) are blind. Furthermore, the symbol-based algorithms (NCMA and NLMS) utilize a symbol-wise calculation of the stochastic gradient, whereas the block-based algorithms (NBCMA and NBLMS) employ an approximate gradient calculation, which is done over a block of symbols. Through this, a better stability and performance can be achieved, especially in the low SNR regime due to additional averaging. The NCMA algorithm utilizes the following gradient in the nth step:   x yn 1 (14.6) rn ¼
n 2 1
jyn j jjxn jj

User terminal wideband modem for very high throughput satellites

199

For the NLMS, the gradient is given by rn ¼


xn yn jjxn jj2

ðyn
pn Þ

(14.7)

where pn denotes the reference signal obtained from the pilot signal. For the two block-based algorithms NBCMA and NBLMS, we obtain   n 1 X xk yk 1 (14.8) 1
rn ¼
L k¼n
L jjxn jj2 jyk j and rn ¼


n 1 X xk yk ðyk
pk Þ L k¼n
L jjxn jj2

(14.9)

respectively. Among these four algorithms, the block-based blind algorithm can be employed during the coarse synchronization phase, that is, before the frame lock. After the successful frame synchronization, pilot fields can be utilized in order to further improve the performance of equalization. Here, the block-based equalization is preferred again, as explained before, and we select NBLMS. Depending on the symbol rate, the assumed cable slope has a stronger or weaker effect on the length of the impulse response and correspondingly the equalization performance, which is shown in Figure 14.5. Here, we show only the pilot-based equalization via NBLMS method after the convergence of the algorithm. Apparently, the performance degradation in terms of signal quality can be observed, if no equalization is applied, especially with a large symbol rate. Using the NBLMS equalizer, it is possible to compensate the degradation almost completely for the input SNR (before the cable) of up to 12 dB. With larger input SNR, the equalizer is capable of dramatically improving the performance compared to the case with no equalization in the loop. However, it is not possible to compensate the losses completely. Fortunately, we focus on relatively low SNR values in this work, so that this performance of the equalization is sufficient.

14.3.6 Demodulation and decoding The design of an LDPC decoder for the DVB-S2 and DVB-S2X standards is challenging especially in case of high throughput and very low bit error rate. This is a nontrivial task, especially for an implementation on a reconfigurable circuit. Even though manufacturers such as Xilinx and Intel Programmable Solutions provide more and more processing power with every generation of their chips, which enables higher and higher processing parallelism, the hardware resources still remain limited. In particular, the parallel processing may lead to frequent access conflicts, which pose the main difficulty in current FEC decoder implementations that rely on massive exchanges of information between many thousands of parallel processing units. Access conflicts are known to generate wrong or deprecated

200

Advances in communications satellite systems 2: ICSSC-2019 20 18 16 14 SINR (dB)

12

Unequalized SINR 35 Msps NBLMS 35Msps Unequalized SINR 500 Msps NBLMS 500 Msps Unequalized SINR 1 Gsps NBLMS 1 Gsps Unequalized SINR 1.4 Gsps NBLMS 1.4 Gsps

10 8 6 4 2 0 –2 –2

0

2

4

6

8 10 12 ES/N0 (dB) After MF

14

16

18

20

Figure 14.5 Performance of NBLMS equalization with FIR filter of length 5 and cable slope of
10 dB messages between processing units which produce altered messages thus letting the FEC decoder diverge from its expected correction capability. In this work, we employ the methods proposed in [17] in order to resolve the memory update conflicts while fulfilling the very low bit error rate performance requirement of the standards. Another difficulty when targeting high throughput with parallel architectures is the routing congestion on reconfigurable targets. All the messages have to be sized so that they require as less bits as possible without degrading the correction performance (cf. [18]). Through this, a fine matching between code properties and FPGA characteristics can be established and performance degradation due to routing congestion is avoided.

14.4 Numerical results In this section, the numerical results for the system performance. We assume a cable slope of 0 dB so that no equalizer is required during the acquisition phase. However, in the proposed architecture, the equalizer can be switched on, if the performance deviates substantially from that one shown below. We start with the frame lock acquisition. For this, we assume that the frame lock has to be acquired within three consecutive superframes, after which the system remains locked for the next five consecutive superframes. The relative number of successful acquisitions is shown in Figure 14.6 for the symbol rates of

User terminal wideband modem for very high throughput satellites

201

Relative number of successful acquisitions (%)

100 90 80 70 60 50 25 Msps 100 Msps 300 Msps

40 30 –2

–1

0

1 ES/N0 (dB)

2

3

4

Figure 14.6 Relative number of successful frame lock acquisitions 25 Msps, 100 Msps, and 300 Msps, respectively. We can observe a monotonic increase in the number of acquisitions. With a very low signal quality and a relatively large initial frequency offset, for example, 12% in case of 25 Msps, the timing error detection does not work efficiently, so that the peak of the NCPDI either falls below the assumed threshold or even corresponds to a wrong start of the superframe. In the latter case, we detect the wrong frame lock during the FEC decoding. We observe that the frame lock is obtained with 300 Msps in more than 90% of cases even at Es/N0 ¼ 2 dB. On the other hand, for Es/N0 ¼ 2 dB, no frame lock can be obtained in 40% of cases with 100 Msps and in 60% of cases with 25 Msps, that is, a new acquisition attempt is required. Typically, the signal quality improves substantially after the frame lock since fine frequency synchronization and phase tracking can be applied. In order to get insight into the signal quality after the phase tracker, an estimate of the SNR is obtained using the squared signal-to-noise variance (SNV) estimator described in [19]. In our investigations, we observed a proportional increase of the SNR estimate with increasing Es/N0 with small fluctuations. The difference between the SNR estimate and the Es/N0 typically varies between 0.4 dB and 0.6 dB. After the acquisition, the system remains locked even in case of sudden variations of the system parameters. In particular, a change of the modulation scheme, for example, from QPSK to 8PSK or 16APSK does not impact the timing or frequency estimation. Also, we observe that Es/N0 needs to drop below
4 dB after the acquisition for a loss of frame lock. In case of carrier frequency instability, the relative offset needs to change by at least 12% of the symbol rate. With a symbol

202

Advances in communications satellite systems 2: ICSSC-2019 100

10–1

BER,FER

10–2

10–3

10–4

QPSK 1/4, BER QPSK 1/3, BER QPSK 1/4, FER QPSK 1/3, FER QPSK 1/3, BER, no impairments QPSK 1/3, FER, no impairments

10–5

10–6 –2

–1.8

–1.6

–1.2 –1.4 ES/N0 [dB]

–1

–0.8

–0.6

Figure 14.7 BER and FER with two MODCODs in low SNR regime rate of 25 Msps, this corresponds to 3 MHz, which is equal to the initially assumed maximum carrier frequency offset. The decoding performance of the end-to-end superframe transmission in terms of frame error rate (FER) or bit error rate (BER) depends on the selected MODCOD. In Figure 14.7, we show the performance of two MODCODs according to [20] during the tracking phase. The first MODCOD corresponds to QPSK modulation with code rate 1/ 4, whereas the second MODCOD has code rate 1/3. In addition, we show the performance of the second MODCOD in absence of any impairments, that is, only the received noise has an impact on the signal quality and error rate. Apparently, in absence of impairments, we obtain extremely low BER and FER for Es/N0 1.3 dB, which is wellaligned with the suggested region of operation for this MODCOD (cf. [20]). In presence of both timing and frequency impairments, the required Es/N0 to reach the same performance is 0.5 dB, which is 0.8 dB higher than without impairments. With the first MOD-COD, we obtain a significantly better performance in terms of BER and FER, so that the error probability becomes negligible for Es/N0 
1.7 dB. Without impairments, this MODCOD has an extremely low BER and FER for Es/N0 
2.5 dB.

14.5 Conclusion In this chapter, we design a terminal modem for very high throughput satellites and focus especially on the receiver synchronization and signal processing. Due to very harsh conditions for the information recovery and due to envisioned low complexity solution, the problem of designing a practical modem architecture appears

User terminal wideband modem for very high throughput satellites

203

to be very challenging. Hence, we propose a new architecture, which is capable of dealing with the assumed hardware impairments and satisfying the imposed system requirements. Our simulations have shown that a quick and reliable synchronization can be reached using the proposed architecture. Furthermore, the end-to-end losses in signal quality are relatively low ( 0.8 dB), which indicates a very good performance of synchronization and information detection.

Acknowledgments This work has been supported by European Space Agency (ESA) in the framework of project AO8908. Opinions, interpretations, recommendations, and conclusions presented in this chapter are those of the authors and are not necessarily endorsed by the European Space Agency.

References [1] “Product leaflet: Newtec MDM5000 satellite modem.” Technical Reports. June 2017, rev. 3.2. [Online]. Available from http://www.newtec.eu/frontend/files/ leaflet/newtec-mdm5000-satellite-modem-r3.2.pdf. [Accessed September 19, 2019] [2] “RT logic HDRM high data rate modem.” Technical Reports. [Online]. Available from http://www.rtlogic.com/media/datasheets/rtl dst hdrm.pdf. [Accessed September 19, 2019] [3] “Comtech EF data CDM-760 advanced high-speed trunking and broadcast modem.” Technical Reports. June 2017. [Online]. Available from http:// www.comtechefdata.com/products/satellite-modem/cdm-760/. [Accessed September 19, 2019] [4] Lee, L., Eroz, M., and Becker, N. “Modulation, coding, and synchronization for mobile and small satellite terminals: An update of the DVB- S2 standard.” IEEE 80th Vehicular Technology Conference (VTC2014- Fall). 2014. pp. 1–6. [5] Casini, E., Gaudenzi, R.D., and Ginesi, A. “DVB-S2 modem algorithms design and performance over typical satellite channels.” International Journal of Satellite Communications and Networking. 2004; 22(3): 281–318. [6] ETSI EN 302 307-2. “Digital video broadcasting (DVB): Second generation framing structure, channel coding and modulation systems for broadcasting, interactive services, news gathering and other broadband satellite applications - Part 2: DVB-S2 eXtensions (DVB-S2X) .” February 2015. [7] Rohde, C., Stadali, H., Perez-Trufero, J., Watts, S., Alagha, N., and De Gaudenzi, R. “Implementation of DVB-S2X super-frame format 4 for wideband transmission.” International Conference on Wireless and Satellite Systems. Springer. 2015. pp. 373–387. [8] Gardner, F. “A BPSK/QPSK timing-error detector for sampled receivers.” IEEE Transactions on Communications. 1986; 34(5): 423– 429.

204

Advances in communications satellite systems 2: ICSSC-2019

[9] D’Andrea, A.N. and Mengali, U. “Design of quadricorrelators for automatic frequency control systems.” IEEE Transactions on Communications 1993; 41(6): 988–997. [10] Mengali, U. Synchronization techniques for digital receivers. New York, NY: Springer Science & Business Media; 2013. [11] Rife, D. and Boorstyn, R. “Single tone parameter estimation from discretetime observations.” IEEE Transactions on Information Theory. 1974; 20(5): 591–598. [12] Corazza, G.E. and Pedone, R. “Generalized and average likelihood ratio testing for post-detection integration.” IEEE Transactions on Communications. 2007; 55(11): 2159–2171. [13] Mazzali, N., Stante, G., Bhavani, S.M.R.R., and Ottersten, B. “Performance analysis of noncoherent frame synchronization in satellite communications with frequency uncertainty.” 2015 IEEE Symposium on Communications and Vehicular Technology in the Benelux (SCVT). IEEE. 2015. pp. 1–6. [14] Corazza, G., Pedone, R., and Villanti, M. “Frame acquisition for continuous and discontinuous transmission in the forward link of satellite systems.” International Journal of Satellite Communications and Networking. 2006; 24(2): 185–201. [15] Haykin, S.S. Adaptive filter theory. India: Pearson Education; 2005. [16] Papadias, C.B. and Slock, D.T. “On the convergence of normalized constant modulus algorithms for blind equalization.” Proceedings of DSP International Conference on Digital Signal Processing. 1993. [17] Marchand, C. and Boutillon, E. “LDPC decoder architecture for DVB- S2 and DVB-S2X standards.” Proceedings of IEEE Workshop on Signal Processing Systems (SiPS). 2015. pp. 1–5. [18] Marchand, C., Conde-Canencia, L., and Boutillon, E. “Architecture and finite precision optimization for layered LDPC decoders.” Journal of Signal Processing Systems. 2011; 65(2): 185. [19] Middlestead, R. Digital communications with emphasis on data modems: Theory, analysis, design, simulation, testing, and applications. Hoboken, NJ, USA: John Wiley & Sons; 2017. [20] ETSI EN 302 307, v1.3.1. “Digital video broadcasting (DVB): Second generation framing structure, channel coding and modulation systems for broadcasting, interactive services, news gathering and other broadband satellite applications (DVB-S2) .” 2013.

Chapter 15

Licensed shared access testbed for integrated satellite-terrestrial communications: the ASCENT project Marko Ho¨yhtya¨1, Mikko Majanen1, Mika Hoppari1, Pertti Ja¨rvensivu1, Heikki Kokkinen2, Arto Reis-Kivinen2, Jaakko Ojaniemi2, Olivier Pellay3 and Duc Pham Minh3

The ASCENT project studies spectrum sharing between satellite and cellular networks using a licensed shared access (LSA) approach, in which information stored in a database is exploited to enable spectrum sharing between users. The project focuses on one side 5G pioneer bands 3.4–3.8 GHz and 24.25–27.5 GHz, which are well-recognized as satellite frequency bands, and on the other side on legacy international mobile telecommunication (IMT) frequency bands. We identified different spectrum sharing scenarios and defined four use cases. We have developed an LSA system testbed, which is used to study and validate the interest of the LSA approach in satellite/ terrestrial sharing scenarios. The results of our simulations and the tests realized on the testbed show that the LSA approach could be used to improve spectrum sharing between satellite and terrestrial networks. This approach could be envisaged to facilitate spectrum sharing between terrestrial mobile network such as international mobile telecommunications and satellite networks operating in a same frequency band. Such an approach could offer new spectrum sharing opportunities which could benefit to both mobile and satellite community. The LSA testbed and the results of the project will be further utilized to develop and assess new sharing scenarios that could lead to creating new spectrum opportunities for different radiocommunication systems. Key Words: cognitive radios; spectrum sharing; 5G; satellites systems

1

VTT Technical Research Centre of Finland Ltd., Oulu, Finland Fairspectrum Oy, Turku, Finland 3 Airbus Defence and Space, Toulouse, France 2

206

Advances in communications satellite systems 2: ICSSC-2019

15.1 Introduction Spectrum sharing between terrestrial international mobile telecommunication (IMT) systems and satellite communication systems could be performed in different manners [1–7]. Secondary terrestrial systems could operate in the satellite bands using cognitive principles to avoid interference to the primary satellite system. It is also possible for the satellite systems to use a frequency band used by cellular networks and to operate as a secondary user in that band. Finally, two satellite systems may share the same spectrum assuming coordination and implementation of proper sharing techniques for interference protection. The aim of the ASCENT project is twofold: (1) to study spectrum sharing techniques in order to understand their usability, performance, and limitations in defined Satcom scenarios considering 5G and satellite spectrum sharing; and (2) to define, demonstrate, and validate a mechanism that allows spectrum sharing for identified use cases relying on LSA sharing mechanisms. As a prerequisite to this project, the study initially focused on sharing in the 3.4–3.8 GHz and 24.25–27.5 GHz frequency bands but the project also proposed to consider legacy IMT frequency bands. In this chapter, we advance state-of-art by introducing novel use cases for satellite/terrestrial sharing. We present the testbed architecture that has been implemented in the project. We then detail the results of measurements done in the testbed on 5G pioneer bands and the simulations done on IMT bands.

15.2 Sharing use cases 15.2.1 Sharing the 5G pioneer bands 3.4–3.8 GHz and 24.25–27.5 GHz have been identified as 5G pioneer bands in Europe [8]. These bands are currently allocated for satellite radiocommunications services and used for different types of satellite applications including broadcasting, broadband applications, and inter-satellite communication. The LSA system controls the operating parameters of the cellular base stations (BSs) in order to protect satellite applications focusing on the protection of receiving satellite earth stations from harmful interference.

15.2.2 Sharing terrestrial IMT bands Terrestrial IMT networks are deployed in ultra-high frequency (UHF) bands, for example, at 450 MHz, 900 MHz, 1,800 MHz, and 2,100 MHz. Three different technologies are deployed in Europe, namely Global System for Mobile communications (GSM), Universal Mobile Telecommunication System (UMTS), and 3GPP Long Term Evolution (LTE). The second objective of the project is also to investigate LSA-based solution which could enable satellite systems to reuse these bands to offer enhanced satellite communication services for land, maritime, and aeronautical applications [9]. The LSA system provides protection zone and criteria

LSA testbed for integrated satellite-terrestrial communications

207

that are used by satellite operators to control their emissions and thus to prevent creating harmful interference on operating IMT network in the considered frequency bands.

15.3 Testbed architecture 15.3.1 Architecture for sharing the 5G pioneer bands From the dynamic spectrum management viewpoint, the system is divided into two main subsystems including their respective functionalities as follows: 1.

The LSA system which (i) assesses the aggregate interference generated by the cellular system on satellite receivers, (ii) iteratively adjusts the transmission power, and (iii) manages the dynamic frequency assignments.

2.

Real hardware tests with the BS controller which (i) (ii)

includes capabilities of adaptive power and frequency control and manages evacuation and frequency change times.

The testbed architecture of the LSA system for sharing in the 3.5 GHz and 26 GHz downlink satellite band is shown in Figure 15.1. It includes an LSA repository and the LSA controller. The LSA repository contains information about satellite earth stations, satellite space stations, and terrestrial IMT base stations’ characteristics as well as protection criteria and associated technical rules for sharing. The incumbent data are entered into the repository through a user interface (UI). In addition, some characteristics such as the location and characteristics of the licensed radio transmitters can also be uploaded from the National Regulatory Authority (NRA) license database, for example, as a CSV-file.

Remote access UI

Base station controller

Internet

LSA system Frequency

UI for entering use data incumbent data (e.g., FSS)

LSA controller LSA repository

National regulatory authority

BS/vendor specific commands

Lock/unlock Set freq/power Get status BS2 BS1 Aggregated interference

Policy data

Not part of the current testbed

Measurement equipment

Figure 15.1 Implemented testbed for spectrum sharing between satellite and cellular systems in 5G pioneer bands

208

Advances in communications satellite systems 2: ICSSC-2019

The project has developed a generic protocol for the LSA controller to control the operating parameters of the terrestrial cellular BSs. We have also developed a BS controller that maps the calls to vendor-specific commands. In the testbed, the LSA system is used to estimate the interference level received by the satellite earth stations and to manage the emission of two commercial BSs in such a way that interference will not occur on satellite system. Further performance assessments of the LSA approach on large networks using a mix of real and emulated BSs is also investigated. The algorithms of the LSA controller are validated by measurements. Finally, remote access UI allows control and connection to the testbed anywhere over the Internet. The basic action that could be achieved through this interface is just to monitor what is going on in the testbed. The UI could also be used to perform network configuration such as adding new FSS earth stations to the satellite system, see Figure 15.2. It can be also used to change operation and parameters of the testbed to run different kinds of tests remotely.

15.3.2 Architecture for sharing terrestrial IMT bands On the other part of the project, when considering the operation of a satellite network over IMT terrestrial communications systems spectrum, the information related to the terrestrial networks with associated protection criteria are included in the LSA repository. Based on this information, the satellite network operator could adjust the operating parameters of the satellite communications so that no harmful interference is caused to the terrestrial systems. This approach can be used to study opportunities of using cellular frequencies by the low Earth orbit (LEO) or geostationary orbit (GEO) satellite systems to provide satellite services to land, maritime, and aeronautical users. The LSA system could be used to calculate areas where satellite system is able to operate to provide connectivity to aeronautical and maritime users. To protect the cellular network from interference, protection zones are calculated by the LSA system that uses the cellular network system parameters in the calculation. The operating area or power restriction information are used by the satellite network operator in order to define operating area of its network and/or to define the adequate emitted power. For maritime application, such an operation could be easily managed by the satellite network operator who is able to manage their antennas footprint when approaching the protection zone boundary to accommodate their coverage area in order to comply with emission restriction zones. For the aviation domain, the satellite network operator will be able to adjust the transmitted power level so that the power level received at the ground level by the terrestrial IMT network does not exceed the protection criteria.

15.4 Performance evaluations The key performance indicators (KPI) of the described testbeds are shown in Table 15.1. We evaluate the testbed in two main categories using the selected KPIs and assess also scalability of the testbed.

LSA testbed for integrated satellite-terrestrial communications

209

Figure 15.2 Web UI interface for entering satellite data We have analyzed the results obtained with our testbed through simulation and measurements for the selected KPIs. The analysis and tests show the applicability of the LSA approach for the different sharing cases. Testbed design and implementation provides scalability and the LSA computations can be done to any number of controlled BSs. The testbed can control the BSs and change iteratively their power and frequency allocations to match interference requirements.

210

Advances in communications satellite systems 2: ICSSC-2019

Table 15.1 KPIs of the testbed ID

Name

Category

Description

Assessment method

KPI-1

Registration time

KPI-2

Frequency handover time

Latency metric Latency metric

Inspection and analysis Inspection/ measurements

KPI-3

Frequency Latency evacuation time metric

KPI-4

System modifications

Time for registering relevant data and setting up the sharing system How fast the testbed can change the operational frequency and continue on another band How fast the testbed can evacuate the transmission on a current band when the incumbent user appears/needs it Modifications required compared to the current systems. This will include the LSA as such and also additional techniques to be used jointly

Complexity metric

Inspection/ measurements Analysis and inspection

Regarding the registration time (KPI-1), analysis in Finland showed that we are able to acquire up-to-date regulatory data regarding the use of certain frequency band and use that in the LSA system. The data include, for example, locations of IMT base stations and operational parameters. Obtaining the incumbent data typically differ between countries, incumbent types and frequency bands. The process of handling the incumbent data from the incumbent notification, possibly processing it at the regulator and inserting it to the LSA system is taken into account in communicating the system performance to incumbents and licensees. In this study, we analyze only the delay caused by the LSA system (Figure 15.3). The delay can be significant for two reasons: (1) the amount of incumbent data are huge requiring a long time for data transfer and possible for insertion into the database, (2) the incumbent data need preprocessing, for example, if part of processing is done for all incumbent and geographic data before any requests come to the LSA system. Here, we did not have any preprocessing and the amount of data was not causing any significant delays. The incumbent data were entered through a web page. For an experienced user, entering the incumbent information took typically 30–60 s. The licensee data were provided through a protocol interface and the delays caused by the licensee data acquisition are incorporated in registration and spectrum allocation times shown in Figures 15.4 and 15.5.

15.4.1 Evacuation and frequency change times The measurement setup is depicted in Figure 15.3 showing the LSA system, the BSC, and the controlled BSs, including commercial and research BSs. The commercial small cell BSs are 3GPP Release 12 BSs, operating in the lower C band (3.4–3.6 GHz) (Table 15.2). The research BSs are universal software radio peripheral (USRP) devices. They are equipped with the controller PC and an LTE framework [10], which is a

LSA testbed for integrated satellite-terrestrial communications Inquiring frequency availability frequently

LSA system

LSA DB

Running base station controller (BSC) software

211

BSC server

National instruments LTE framework with VTT modifications

USRP controller PC eNB_1

eNB_2

Controlled commercial base stations

USRP eNB_1

USRP Controlled USRP eNB_2 research base stations

Figure 15.3 Measurement setup for evacuation and frequency changes

Execution time (ms)

Mean registration time 2,000 1,500 1,000 500 0 1

10

50 eNB amount

100

1,000

Mean registration time

Figure 15.4 Registration time for increasing number of BSs software add-on that provides a real-time physical layer LTE implementation. Thus, we have a real-time prototyping setup that enables very dynamic operations in contrast to commercial systems that are not yet designed for fast frequency changes [11].

Advances in communications satellite systems 2: ICSSC-2019 Mean execution time (ms)

212

Mean spectrum inquiry time 140,000 120,000 100,000 80,000 60,000 40,000 20,000 0 1

10

50

100

1,000

eNB amount Mean spectrum inquiry time

Figure 15.5 Spectrum allocation calculation times for increasing number of BSs Table 15.2 Configuration times with the used C band research BSs (white columns) and commercial BSs (gray columns) Configuration process/command

Mean time duration (s)

Standard deviation (s)

Mean time duration (s)

Standard deviation (s)

LOCK UNLOCK SET FREQUENCY GET STATUS SET POWER

0.746 0.746 0.246 0.208 0.245

0.028 0.025 0.023 0.023 0.023

8.78 8.97 7.97 12.53 7.73

0.39 0.37 0.37 0.47 0.45

The main configuration commands from the BSC to control the BS are as follows: ●

●

●

●

●

LOCK: The command used to turn off the air interface at the BS in order to release the current LSA frequency under use. UNLOCK: The command is used to open the air interface at the BS to let the BS start its services. The used frequency needs to be decided by the LSA controller based on the information in the LSA repository. SET FREQUENCY: The command is used to change frequency at the BS to an available LSA frequency provided by LSA repository. SET POWER: The command is used to change transmission power at the BS to avoid interference with incumbent users. GET STATUS: The request message to the BS to update current statuses such as air interface parameters, frequency used, and transmission power.

Evacuation time: The evacuation process is defined as the necessary time duration to complete the LOCK process. The average result from the measurements regarding the research BSs is roughly 0.75 s whereas with the commercial BSs the result is close to 9 s. The evacuation time defines, for example, how much earlier one needs to know the appearance of a nomadic satellite station to avoid interfering with it.

LSA testbed for integrated satellite-terrestrial communications

213

The frequency change time is defined as the time when the BS unlocks itself and opens the air interface in the new frequency. The total frequency change time is combined UNLOCK time and SET FREQUENCY time. When the research BSs are used the mean frequency change time is very close to 1 s. When the commercial BSs are used, the result is much higher, in the order of 17 s. Thus, the results show clearly that the SDR platforms provide better dynamic capability. Commercial BSs should be developed to support faster frequency changes in order to minimize the likelihood of interference and to facilitate spectrum sharing. This is one of the identified system modifications related to KPI-4. In general, the commercial 3GPP devices can be connected to the LSA system without any modifications which is clearly one of the advantages of the approach. The main modifications to current cellular systems come with (1) inclusion of the LSA system and its interfaces, (2) developing vendor-specific control interface in the system, and (3) adding advanced spectrum sharing mechanisms such as smart antennas or layer 1 filtering for interference management.

15.4.2 Scalability of the testbed The implemented LSA system can handle a large amount of real and simulated BSs. To study the scalability of the testbed, we have made performance measurements considering that the FSS Earth station in the C band would be located in the Paris area. We registered up to 1,000 BSs in the LSA system using the real network parameters obtained from the French regulatory authority registry using locations of 2.6 GHz BSs and converting them to 3.5 GHz devices. The average results over multiple experiments regarding the registration of the BSs to the LSA system and the spectrum inquiry times are shown in Figures 15.4 and 15.5. Registration of a single BS takes 275 ms. It is easy to register simultaneously multiple BSs using the same registration file where locations and operational parameters are stored in the system. Registration of 100 BSs increases the time to 515 ms and further increase to 1,000 devices takes 1,851 ms which is less than seven times the registration time of a single BS. Then, we measured spectrum inquiry times that include power and frequency allocation calculations of the LSA system. Aggregated interference to the FSS Earth station is included in the calculations. The results shown in Figure 15.5 reveal that calculation for a single station takes 245 ms, 10 stations takes 1 s, 50 stations takes 5 s, 100 stations takes 10 s, and 1,000 stations more than 100 s. Thus, the LSA system can handle a large number of controlled BSs and their resource allocations. Computation times can be reduced by leasing more computation power, that is, it can be also scaled down to the level required by the dynamic spectrum access system and the selected use case.

15.4.3 Proof of concept of IMT frequency bands sharing IMT spectrum is currently allocated in several UHF bands, from 450 to 2,690 MHz. Examples of allocations are illustrated in Figure 15.6 indicating the nature of the technology as for GSM, UMTS, and LTE systems.

214

Advances in communications satellite systems 2: ICSSC-2019 Allocated IMT bands 880

1,915

2,500 7

915 925

900 MHz (GSM) 1,965 2,110 2,100 MHz (UMTS) 2,570 2,620

960

1,710

1,785

2,170

1,915

1,800 MHz (GSM) 1,965 2,110 2,170

2,690

1,805

1,880

2,300 MHz (LTE/LTE-A TDD)

2,600 MHz (LTE/LTE-A )

Figure 15.6 Example of IMT spectrum identification In the framework of the proof of concept, we considered to reuse the IMT 900 MHz band where the amount of spectrum is 2
35 MHz linearly polarized without prejudging of the suitability of other IMT-identified UHF frequency bands. Both interferences at the mobile terminals and at the BSs are analyzed with Systems Tool Kit [12]. The interferer is a large multibeam GEO satellite with 121 beams providing voice and data communication using a second-generation Digital Video Broadcasting – Satellite (DVB-S2) waveform. The reuse factor is 4 and each beam uses 8.75 MHz of the band. The satellite polarization is either circular (cases 1 and 2) or linear (case 3). We evaluated the maximum possible Equivalent Isotropic Radiated Power (EIRP) from the satellite that would not cause harmful interference to the cellular system (i.e., respect the threshold of I/N below than 6 dB). Case 1: Satellite interfering with mobile terminals (circular polarization of satellite emissions) The mobile terminal has an omnidirectional antenna with a 0 dBi gain. The assessment of the maximum permissible satellite EIRP without causing interference on IMT terminals is relatively constant and varies between 12 and 12.5 dBW on land. Over sea, it is possible to increase the EIRP by 3 to 9 dB without interfering with IMT systems. The link budget analysis shows that quasi-error free reception is possible with omnidirectional satellite terminals on all sea areas as presented in Figure 15.7. Quasi-error free reception means that the rate of errors is not yet visible in video reception. The modulation is a quadrature phase shift keying (QPSK) modulation providing around 5 Mbps of aggregated throughput. The use of a small parabolic antenna of 45 cm diameter provides enough gain to have a quasi-error free reception even over land. Case 2: Satellite interfering with a BS (circular polarization for satellite emissions) The BS has a directive antenna providing 13 dBi of gain which is a typical value for a macro BS. The 3 dB beamwidth is 10 in vertical and 120 in horizontal. The antenna height is 40 m. A down tilt of the antenna is 6 , corresponding to a cell radius of 22 km.

LSA testbed for integrated satellite-terrestrial communications

215

Figure 15.7 I/N at user terminal (considering circular-polarized satellite emissions) Because of the directivity of the antenna of the BS and due to its low elevation, the worst case corresponding to the maximum interference from the satellite to the BSs of the IMT terrestrial network is cell edges, when satellite emissions are received on the main lobe of the BS. The situation is depicted in Figure 15.8. The permissible satellite EIRP varies from 44 dBW at the border of the satellite coverage to 74 dBW at the nadir of the satellite. The link budget analysis shows that quasi-error free reception is possible with omnidirectional satellite terminals except at the edge of coverage (dark and blue cells). The modulation is a QPSK modulation providing around 5 Mbps of aggregated throughput. The use of a small parabolic antenna of 70 cm diameter would provide enough gain to have a quasi-error free reception over the whole coverage area. Case 3: Satellite interfering with mobile terminals (considering linearpolarized satellite emissions) Due to the fact that the terrestrial and the satellite system use the same polarization, the satellite transmission power must be reduced to comply with the

216

Advances in communications satellite systems 2: ICSSC-2019

Figure 15.8 Max permissible satellite EIRP without causing interference on IMT base station (considering circular-polarized satellite emissions) I/N criteria of 6 dB. The permissible satellite EIRP must then be reduced by 10 dB (10 dB have initially been taken into account as a polarization mismatch). Despite lower EIRP, this case is still viable and could be of interest to facilitate the development of lower performance application such as Internet-of-things (IoT). In addition, this particular sharing scenario implies that similar signals are considered to achieve, both terrestrial and satellite communication (e.g., same polarization) and therefore this is of particular interest to consider dual-mode mobile terminals and for the development of integrated satellite and terrestrial systems. Finally, it should be emphasized that power limitations could be overcome by the use of spread spectrum technologies or more complex modulations to enable good transmission despite the low signal power. Case 4: Satellite interfering with a BS (considering linear-polarized satellite emissions) Similar to use Case 3, the EIRP values are 10 dB lower than in Case 2 due to the polarization “alignment.”

LSA testbed for integrated satellite-terrestrial communications

217

The main interest of using a linear polarization is that this facilitates the development of dual-mode terminals (satellite þ terrestrial) and will foster the emergence of new integrated applications. However, due to the fact that the bands are allocated in opposite directions, the development of terminals will be more difficult than in Case 3. More detailed analysis of this case remains as future work.

15.5 Conclusions We have developed an LSA testbed to study spectrum sharing between satellite and terrestrial systems. The results on studies related to 5G pioneer bands reveal that the implemented system is scalable and able to control hundreds of BSs simultaneously. In addition, achieved evacuation times using our research BSs outperform commercial BSs and state-of-the-art values of previous studies in the C-band. Regarding the sharing of IMT bands by satellites, the simulations confirmed the possibility of the sharing. We have shown that the direction of the bands allocated to the satellite has a very important impact on the design and performance of the satellite system: ●

●

When both satellite and cellular systems operate in the downlink direction (Cases 1 and 3), the maximum EIRP is flat around 44 dBW above land and 63 dBW over sea. With linear polarization, these values must be reduced by 10 dB. The simulations show that quasi-error free reception could be possible with omnidirectional satellite terminals on all sea areas. With opposite direction (Case 2), the maximum EIRP varies from 44 dBW (at low elevations) to 74 dBW (at the subsatellite point). The worst case from interference point of view is on the cell edges because at low elevation, the satellite emissions are received on the main lobe of the BS.

Further investigations will be conducted to evaluate the possible optimization of the space segment (reduced coverage, position of the satellites, and use of LEO constellations) and the service offer. Of particular interest are maritime and aeronautical applications which are not well served by terrestrial systems; but also new integrated satellite and terrestrial services, including IoT, could benefit from the new opportunities offered by the sharing of IMT bands.

Acknowledgments The work has been funded by the European Space Agency ARTES Project “ASCENT: Demonstrator for license assisted spectrum access satellite networks, Contract No. 4000123000/18/NL/WE.” The views expressed herein do not reflect the official opinion of the European Space Agency.

218

Advances in communications satellite systems 2: ICSSC-2019

References [1] 3GPP. “3rd Generation Partnership Project; Technical Specification Group Services and System Aspects; Study on architecture aspects for using satellite access in 5G.” TR 23.737 V0.3.0. Technical Report. October 2018. [2] ECC Report 254. “Operational guidelines for spectrum sharing to support the implementation of the current ECC framework in the 3600–3800 MHz range.” November 2016. Available from http://www.erodocdb.dk/Docs/ doc98/official/pdf/ECCREP254.PDF [Accessed September 19, 2019]. [3] Ho¨yhtya¨, M., Ma¨mmela¨, A., Chen, X., et al. “Database-assisted spectrum sharing in satellite communications: A survey.” IEEE Access. 2017; 5: 25322–25341. [4] Gozupek, D., Bayhan, S., and Alago¨z, F. “A novel handover protocol to prevent hidden node problem in satellite assisted cognitive radio networks.” Proceedings of the ISWPC. May 2008. [5] Tehrani, R.H., Vahid, S., Triantafyllopolou, D., Lee, H., and Moessner, K. “Licensed spectrum sharing schemes for mobile operators: A survey and outlook.” IEEE Communications Surveys & Tutorials. 2016; (18): 2591–2623. [6] ITU-R M.2109. “Sharing studies between IMT-advanced systems and geostationary satellite networks in the fixed-satellite service in the 3 400-4 200 and 4 500-4 800 MHz frequency bands.” October 2007. [7] Ho¨yhtya¨, M., Hoppari, M., and Majanen, M. “Validation framework for building a spectrum sharing testbed for integrated satellite-terrestrial systems.” Proceedings of EUSIPCO. September 2019. [8] RSPG 16-032, Radio Spectrum Policy Group opinion on spectrum related aspects for next-generation wireless systems (5G). “Strategic roadmap towards 5g for Europe.” November 9, 2016. Available from http://rspg-spectrum.eu/wp-content/ uploads/2013/05/RPSG16-032-Opinion_5G.pdf [Accessed September 19, 2019]. [9] Ilcˆev, S.D. Global Mobile Satellite Communications Theory for Maritime, Land, and Aeronautical Applications, 2nd ed. Springer, 2017. [10] National Instruments. “Overview of the LabVIEW Communications Application Frameworks.” IEEE Spectrum. Available from https://spectrum.ieee. org/computing/networks/overview-of-the-labview-communications-applicationframeworks [Accessed September 19, 2019]. [11] Ho¨yhtya¨, M., Ma¨mmela¨, A., Chiumento, A., Pollin, S., Forsell, M., and Cabric, D. “Database-assisted spectrum prediction in 5G networks and beyond: A review and future challenges.” IEEE Circuits & Systems Magazine. 2019; 19(Third quarter): 34–45. [12] Analytical Graphics, Inc. Available from https://www.agi.com/home [Accessed September 19, 2019].

Chapter 16

Cognitive communications for NASA space systems David Chelmins1, Janette Briones1, Joseph Downey1, Gilbert Clark1 and Adam Gannon1

The growing complexity of spacecraft constellations, communication relay offerings, and mission architectures drives the need for the development of autonomous communication systems. The National Aeronautics and Space Administration (NASA) has traditionally launched single spacecraft missions that are served by the Space Communication and Navigation (SCaN) program. Operations on SCaN networks are typically scheduled weeks in advance, and often each asset serves a single user spacecraft at a time. Recent movement towards swarm missions could make the current approach unsustainable. Additionally, the integration of commercial communication service providers will substantially increase the data transfer options available to new missions. NASA science missions have found benefit in launching swarms of spacecraft, allowing coordinated simultaneous observations from different perspectives. Inter-spacecraft communication (mesh networking) is an enabler for this architecture, as are CubeSats that allow cost-effective provisioning of distributed mission assets. As more complex swarm missions launch, one challenge is coordinating communication within the swarm and choosing the appropriate mechanism for telemetry, tracking, control, and data services to and from Earth. Cognitive communications research conducted by SCaN aims to mitigate the increasing communication complexity for mission users by increasing the autonomy of links, networks, and service scheduling. By considering automation techniques including recent advances in artificial intelligence and machine learning, cognitive algorithms and related approaches enable increased mission science return, improved resource utilization for service provider networks, and resiliency in unpredictable or unplanned environments. The Cognitive Communications Project at the NASA Glenn Research Center develops applications of data-driven, nondeterministic methods to improve the 1

NASA Glenn Research Center, Cleveland, Ohio, USA

220

Advances in communications satellite systems 2: ICSSC-2019

autonomy of space communication. The project emphasizes the development of decentralized space networks with artificial intelligence agents optimizing communication link throughput, data routing, and system-wide asset management. This chapter discusses the objectives, approaches, and opportunities of the research to address growing needs of the space communications community. Key Words: cognitive; autonomy; delay-tolerant networking

16.1 Introduction Mission needs typically spur innovation in the field of space communication. The National Aeronautics and Space Administration (NASA)’s communication architecture is no exception: from the early days of the Apollo lunar missions to the remarkable construction and operation of the International Space Station, and to the more recent InSight mission to Mars leveraging a small spacecraft communications relay (MarCO) [1]. In each case, communications system engineers match the mission needs to an appropriate communications architecture and capability, delivering the correct data throughput, with the correct latency, with the required reliability, at an acceptable cost. Traditionally, NASA’s space communication infrastructure has been government-owned or contracted. Early dedicated networks including the Spacecraft Tracking and Data Acquisition Network and Manned Space Flight Network consisted of ground stations that provided limited but acceptable data communications capability for uncrewed and crewed spacecraft, respectively, in the 1960s. NASA added geostationary relay satellites to the infrastructure beginning in the 1980s, providing continuous coverage of low Earth orbit through the tracking and data relay satellite system (TDRSS) [2]. Today, NASA operates three communication networks: the space network (SN), the deep space network (DSN), and the near Earth network (NEN). The modern space communications marketplace continues to evolve. While commercial satellite relays and ground stations have existed since the earliest days of spaceflight, only recently has NASA considered the regular use of these systems to meet its mission needs [3]. Multiple companies have proposed new megaconstellations of satellites in low Earth orbit, providing high-rate and low-latency communication capability [4]. NASA is developing cognitive communications technology to reduce the burden of bridging its legacy government-owned/operated communications systems with the use of commercial systems. The growing number of operational spacecraft requires more sophisticated techniques to cooperatively share spectrum as well as mitigate intra- and internetwork interference when necessary. Regular use of increasingly high frequencies including Ka-band (26.5–40.0 GHz) and beyond necessitate more intelligent techniques to adapt to time-varying atmospheric conditions [5,6]. More complex multihop network topologies envisioned around other bodies such as the Moon and Mars provide a challenge for optimal in-space data routing [7]. Software-defined radio provides a platform to address many of these challenges.

Cognitive communications for NASA space systems

221

The space communication and navigation (SCaN) program office sponsors the Cognitive Communications project at the NASA Glenn Research Center. The purpose of the project’s research is to leverage the flexibility and adaptability of software-defined radio within the context of autonomy, providing an overall benefit to the mission without increasing the cost of the operations.

16.2 Defining cognition Merriam-Webster defines “cognitive” as relating to, being, or involving conscious intellectual activity such as thinking, reasoning, or remembering [8]. While many of the best examples of cognitive systems are biological, a significant effort in artificial intelligence (AI) has been exerted to develop artificial systems that exhibit abilities resembling thinking, goal-oriented reasoning, and remembering. A cognitive radio, as defined by seminal works in the field, is a “brain empowered” wireless device capable of reacting to and learning from its environment to enhance communications [9,10]. Thus, cognitive radios should be aware of the operational environment, capable of adapting their operational parameters, and able to improve from past actions to enhance future performance [11]. The principles of cognitive radio are broadly applicable across the protocol stack to cognitive networking and even entire cognitive communication systems [12,13]. NASA’s Cognitive Communications project adapts the formal definition of “cognitive” as follows: any system, or part of a system, that is able to mitigate obstacles, respond to and learn from its environment, and achieve beneficial goals to the completion of its primary mission. Such a cognitive system can perform these activities with minimal to no human interaction. Finally, the cognitive system must have the ability to adapt to changing conditions by producing reasonable outcomes in scenarios that extend beyond the preprogrammed knowledge of its original inception. The Cognitive Communications project defines a cognitive engine (CE) as a decision-making algorithm that enables part of a cognitive system. Multiple CEs can apply to various levels of the communications protocol stack, from a single radio frequency (RF) or optical link to complex distributed applications. A system designer could implement each CE in many different ways utilizing different decision-making methods, including those based on machine learning (ML), so long as these methods align to the goals of the overall cognitive system. In general, CEs must rely on multiple inputs and process data in different ways to come up with a usable solution. One CE design approach aggregates various ML and deterministic algorithms into a single framework, evaluates performance in parallel, eliminates poor-performing algorithms, and aggregates the remaining algorithms to deliver an optimal solution for the particular problem at hand. This approach requires environmental feedback to optimize toward particular objectives. The decision process (Figure 16.1) provides a general approach on how CEs interact with the outside world and behave intelligently in that environment to maximize their objectives.

222

Advances in communications satellite systems 2: ICSSC-2019 Initialization parameters

Initialization: Preprocessing tuples

Tuple x[features] y[labels] or d[reward function]

Error value e’ Feedback: Error function Gradient control Target function (x)

Prediction (y)

Environment changes

Environment descriptor

Action tuple: Action based on predicted value changes the environment descriptor Adaptation: Weight tuning Hyper–parameter tuning Ketnel function tuning

Machine learning Statistical inference Algorithms

is e’ > [e(t – 1) = (y – x)]? List: x [features] y [labels] or d [reward function] E [environment function]

Figure 16.1 Interaction of cognitive engines with their environment

16.3 Focus areas The Cognitive Communications project performs research in four distinct but intertwined areas: ● ●

●

●

Links – concerning point-to-point connections between two devices; Networks – concerning multiple devices routing information among multiple links; Systems – concerning the interaction among devices and supporting groundand space-based infrastructure; and Enabling technology – concerning the on-board processing, sensing, and adaptation capability of a device that allows it to participate in cognitive links, networks, or systems.

These areas broadly resemble the open systems interconnection (OSI) model [14] with some notable differences. The highly directional links common to space communications necessitate an approach to medium access control, typically a function of the data link layer, different from that of terrestrial wireless where transmissions from user equipment are generally omnidirectional. System infrastructure (ground stations or relay satellites) featuring directional antennas are generally incapable of supporting many simultaneous users. CEs applied broadly across all levels of the protocol stack will determine link optimization, network routing, and system management. While each of these focus areas can mature independently, the end goal is to transition toward an overall cognitive system-of-systems, optimized across all OSI layers. The spacecraft itself and the communication provider networks must perform joint cross-layer, distributed decision making that conforms to the mission objectives and network

Cognitive communications for NASA space systems

223

capabilities. In Sections 16.4–16.7, this chapter will discuss each focus area, its potential optimizations, and benefits in the context of a cognitive space communications system.

16.4 Cognitive links A space communication link is a wireless connection between two radios over distance with at least one of the radios in space. Currently, NASA missions determine the exact communication system configuration of the radio prior to launch. In the case of an RF system, the mission designer typically allocates a single frequency and bandwidth to the radio, applies for the corresponding spectrum license, and negotiates service with a communications service provider (through spacebased relays or ground stations). The traditional approach is robust and proven, yet inflexible to real-time changes. The communication system can fail to communicate under several plausible scenarios including: ● ●

●

The receiver encounters interference resulting in loss of lock; Mission hardware degrades, reducing the transmit power or increasing receiver noise figure; and The communications service provider cannot schedule a sufficient number of contacts.

In each of these cases, the only remedy in a traditional, inflexible communications approach is to keep transmitting and hope the result improves over time. Using a software-defined radio, a mission operations team can program the system to adapt to predictable or gradual failures. However, most real-time issues (especially those that are unlikely or unpredictable) can result in loss of mission data. Cognitive link capabilities include technologies, algorithms, and protocols applicable to the physical and data link layers. The prime benefit of a cognitive link approach is on-board, autonomous mitigation of real-time issues. A second, significant benefit is the ability to improve performance and efficiency of the communication link.

16.4.1 Radio frequency interference mitigation One example of a cognitive link capability is RF interference mitigation, which automatically senses and avoids spectrum interference by changing frequency, bandwidth, data rate, and antenna pointing. An automated approach was developed [15] that significantly reduces data loss from RF interference while increasing throughput. Figure 16.2 shows the RF interference mitigation concept, where the space-based transceiver is located on the International Space Station. Previous space-based testing using NASA’s SCaN Testbed has shown that RF interference is most likely encountered on the ground. That is, transmitters local to the ground station are more likely to introduce interference in a link than (1) a space- or ground-based transmitter pointing at the spacecraft or (2) a space-based transmitter

224

Advances in communications satellite systems 2: ICSSC-2019

Spectrum analysis

Link controller

Figure 16.2 RF interference mitigation concept pointing at the ground station. The authors in [8] describe a cognitive engine that optimizes: ● ● ●

Throughput (maximized) – the number of data bits transferred per second; Occupied bandwidth (minimized) – occupied spectrum; and Transmit power (minimized) – the amount of communications power on the spacecraft.

16.4.2 Radio link optimization Prior work has shown adaptive coding and modulation (ACM) successfully optimizing throughput over a communication link with varying margin [16]. Typically, ACM uses fixed signal-to-noise ratio thresholds based on theoretical characterization. A new approach is to implement a cognitive engine that decides when to change modulation, coding, and transmission power based on observed channel conditions and mission platform constraints. Results have demonstrated that a neural network-based reinforcement learning algorithm performing multiobjective optimization is feasible for satellite communication [17]. The authors of the referenced work developed a cognitive engine that tested multiple radio settings so that the system could learn how to adapt to achieve multiple goals for satellite communication in a dynamically changing environment.

16.4.3 Automatic receiver configuration In an effort to reduce operator burden when switching between communication relay providers, self-configuration of a software-defined radio may be possible by sensing the inbound signal to perform signal recognition. This technique relies on signal processing mechanisms to recognize signal waveform parameters such as modulation scheme [18], coding, and data rate. Using these parameters facilitates

Receiver Channel

Input bit stream

Transmitter

225

Output bit stream

Cognitive communications for NASA space systems

Optimizer

Figure 16.3 High-level view of a learned communication system system self-configuration and link acquisition even in the presence of noise or weak signals. For deep space systems that have a significant round-trip time (RTT) and multiple possible waveform configurations, such a method could save one RTT or more.

16.4.4 Deep learning communication links One intriguing development in the cognitive links area is the creation of learned communication systems, which use deep learning (AI) techniques to minimize endto-end message reconstruction. This method offers potential improvements to traditional systems in three key areas: jointly optimizing modulation and encoding, utilizing the nonlinear function approximation capabilities of a neural network to account for power amplifier distortions, and a relaxation on the assumption that system noise follows a Gaussian distribution. In [19,20], the authors introduce an autoencoder model (see Figure 16.3) that performs physical layer optimization. This work was extended in [21] where a generative adversarial network (GAN) was used to learn an arbitrary channel model that includes nonlinearities and memory effects. Although this approach shows promise, a real-time adaptation will impose a significant computational burden for today’s space processors. Additionally, due to the amorphic nature of the self-learned system, new protocols must be implemented to account for symbol timing acquisition and transmitter/receiver coordination during training. Recently, this area has received significant attention in the research community and several potential architectures have been proposed. In [22], the need for a feedback path during training was removed and a live implementation was demonstrated using hardware radios.

16.5 Cognitive networks The focus of cognitive networks covers many different aspects of networking. This includes the higher-level objective of realizing an autonomous system of systems:

226

Advances in communications satellite systems 2: ICSSC-2019

the autonomous aspects not only understand the interfaces between the systems but also can optimize to fulfill specific objectives. At a lower level, there is the objective of autonomy and automation for the existing network infrastructure.

16.5.1 Delay-tolerant networking NASA has a demonstrated need for networks that are tolerant to delay and disruption, known as delay-tolerant networking (DTN). DTN can refer to a network that is affected by disruption or delay, an architecture [23], a set of tools, a protocol [24,25], or a specific implementation of a protocol [26]. For the sake of clarity, this chapter will disambiguate its use of DTN as it relates to the various ways in which cognition may be added to different elements of the DTN portfolio. Generally, DTN is an architecture. An implementation of DTN is a set of protocols and techniques that realizes a network that is tolerant to delay and disruption. One key aspect of DTN is the capability to customize an implementation to match its environment: if an environment is not anticipated to carry links with long delays, for example, protocols such as LTP [27] may not be necessary.

16.5.2 Intelligence in the DTN architecture One aspect of DTN study is the identification, maintenance, and assignment of endpoints and their mapping to minimum reception groups (MRGs). The DTN architecture allows an entire constellation of spacecraft (for instance) to act as a single, shared DTN endpoint in a larger network. To date, however, there generally has been a 1:1 correspondence between physical machines with specific endpoints. In an intelligent system, this does not need to be the case. An intelligent DTN application can provide flexibility: rather than each node in a constellation acting as an individual endpoint, the entire constellation may serve as a single endpoint (with a shared logical bundle protocol agent). The constellation can then send and receive DTN data in a coherent, locally coordinated fashion, reducing the number of times that bundles must be encoded and decoded. This reduces unnecessary overhead, allowing implementation of DTN in nontraditional locations, such as between individual nodes in a constellation that may not individually suffer from large degrees of delay or disruption. An intelligent approach to realizing a delay-tolerant network is to monitor a constellation for failures and/or new available assets and to adjust the MRG accordingly. The system might also predict and move data to specific elements of an MRG that would be most likely to need that data for future events. Another potential advantage is the freedom to manage data movement between nodes in an MRG; this capability (Figure 16.4) allows selection of a prime data downlink based on expected availability, and dynamic reassembly of data fragments within the context of a distributed MRG. This area of study is referred to as drop data anywhere (DDA). An additional area relates to the means by which data might be prepared and processed: this applies not only at the data’s origin but also, through the application of virtualization, at each incremental hop in data’s movement toward its

Cognitive communications for NASA space systems

227

Single DTN element – ipn:1

Cubesat 1

Cubesat 3 Internal data management over stable cross-links

Cubesat 2

Disrupted link between ground and space

Ground station 1

Ground station 2

Internal data management over internet backhaul

Single logical network – 1 ... N DTN nodes

Figure 16.4 Managing data across distributed endpoints destination. A cognitive routing scheme could ensure that the data passes through incremental nodes that have the correct processing ability. Routing between multiple delay-tolerant networks is a common challenge. While a number of approaches have been explored to date, there is still more work needed on the integration of physical and link scheduling with network routing decisions. For example, there is active work in the area of applying spiking neural networks (SNNs) to the area of information transfer through delayed and/or disrupted networks [28]. The referenced author has considered using software-defined networking (SDN) protocols to propagate routing information through a network.

16.5.3 Cognition in the DTN protocols Cognition in the DTN protocol suite would blend global (architectural) goals with local ones. For instance, specific protocols might be necessary to support the maintenance and establishment of an MRG. While these protocols would not be cognitive in and of themselves, the protocols would serve as a means to realize cognition in the larger network. One such protocol example is the ability to dynamically discover, enumerate, and control radio capabilities, allowing the system to understand its environment and communicate its decisions to others. Work in this area can build on existing protocols that are well known [29]. It can also build on newer entries into the protocol arena such as [30], which is an approach to the management and monitoring of a delayed and/or disrupted network: while not suited for a well-connected network, this protocol is useful for situations where

228

Advances in communications satellite systems 2: ICSSC-2019

coherent networks need to exchange information or commands through hops that are somehow constrained. The other approach to protocol-level cognition is through optimization of protocol parameters themselves. If, for example, a system knows what the expected link round trip time and link rate will be, it can optimize elements of its own protocols to use that information without having to rediscover that information for itself. Work here has explored modeling existing protocols [31] to predict the impact of a specific protocol-level decision.

16.5.4 Legacy, infrastructure, and bootstrapping intelligence One major challenge associated with DTN involves its infusion into existing infrastructure. Traditionally, NASA’s space network infrastructure has focused on information transfer from an asset in space to the ground. Expanding such a network requires increasing the number of available dedicated antennas or number of unique spectrum bands available to support missions. The network will expand until the supply of either frequency or antennas is exhausted. Recent decreases in launch costs, an evolution in mission design to move toward distributed (e.g., multispacecraft) missions, and a general increase in demand for space-faring missions have all begun to place a severe strain on the fixed pool of resources available. This has led to applied research at the intersection of resource allocation and DTN. Automated scheduling techniques to reserve network resources (e.g., antennas and bandwidth) are one mitigation. To implement automated scheduling, DTN can find an optimal network routing solution, and a machine-to-machine interface can schedule time on the corresponding nodes. This creates a strong sense of vertical integration between different elements. Additionally, MRGs can fuse multiple short physical events (i.e., antenna time) into one larger logical event. This allows missions to create a long logical event without necessitating the implementation of point solutions to support such a usecase. Given that smaller scheduling blocks allow for improved flexibility when scheduling, this offers benefits to both the mission and the service provider: missions can get more aggregate service time, and providers can increase the duty cycle of the resources they have available to them.

16.5.5 Virtualization in future cognitive networks Virtualization is an attractive mission-level solution to support flexible on-board processing and routing capabilities in a cognitive network. When implemented correctly, the overhead associated with virtualization can be minimal, while offering benefits to reliability, load balancing, and the ability to scale-up/scaledown. One candidate for virtualization in a space environment involves the core flight executive (cFE) framework [32]. The common development libraries, framework, and environment have allowed for code re-use across a diverse set of missions. Work is ongoing in the area of building cFE applications as real-time executive for

Cognitive communications for NASA space systems

229

multiprocessor systems (RTEMS) virtual machines. This would allow an intelligent system to migrate elements of its own execution to any compatible virtualization environment in real-time. It also offers the network a capability to treat members less like an individual, and more like a distributed piece of a coherent whole, realizing a cloud-compatible approach to service management in space environments. Notably, both intelligent systems (e.g., inference and online learning capabilities) as well as network functions themselves (e.g., DTN) can be realized as discrete sets of coherent functions, each of which can execute on a participating member of a cloud-centric service architecture in a space setting.

16.6 Cognitive systems The Cognitive Systems focus area aims to optimize performance across entire space communication systems, improving the interaction between mission spacecraft and service provider infrastructure. With increased system autonomy, the mission spacecraft can negotiate access to communication services based on its current data transfer needs. This architecture of automated resource allocation from spacecraft-initiated requests, called user-initiated service (UIS), aims to provide more responsive access to high-performance space communications [33–35].

16.6.1 User-initiated service In current practice, network operators manage access to the highest performance services based on requests from each mission’s operations staff. Mission operators must anticipate spacecraft commanding and data transfer needs potentially weeks in advance. However, an increasing number of spacecraft has event-driven service requirements with demand that is difficult to predict. Current practice limits the network’s ability to negotiate schedules rapidly. The ideal network infrastructure incorporates automation and cognitive techniques to allocate resources exactly meeting each user’s immediate demand. In the UIS concept, the mission spacecraft itself originates a request for communications or navigation service based on current needs. For example, a spacecraft recording scientific data from a transient astronomical event can automatically send a request for data downlink service when its onboard storage is nearly full. In this new paradigm, access to high-performance service can be requested using machine-to-machine communications over low-rate, high-availability control channels. The use of control channels to request resources from a wireless network is seen during the base station association process of terrestrial mobile networks or the Proximity-1 hailing channel used by Mars spacecraft [35,36]. Examples of existing control channels include the multiple-access capability of NASA’s TDRSS and various commercial services that provide low-rate, on-demand messaging capabilities. Figure 16.5 shows an overview of the process. A request for service is received over the control channel by a central Event Manager that is aware of both spacecraft and communications network capability. The Event Manager contracts on behalf of the requesting spacecraft for communication service with any network capable of establishing a link with the mission spacecraft. The Event Manager then

230

Advances in communications satellite systems 2: ICSSC-2019

High-rate data channel

Low-rate control channel

Ground station/ relay satellite Event manager Spacecraft mission operations center

System scheduling interface

Asset automation and control software

Figure 16.5 Overview of the user-initiated service process sends the necessary waveform configuration, ground station or space relay location, and contact time back to the requesting spacecraft via the same control channel. At the start of the negotiated access window, the spacecraft exchanges data with its mission operations center over the contracted high-rate data channel. An on-orbit experiment using NASA’s SCaN Testbed demonstrated the concept with TDRSS [37]. The low-rate S-band multiple access service was used as a control channel to send requests to schedule high-rate Ka-band single access service for data transfer. Though both services were provided by the TDRSS, the high-performance Ka-band link is capable of supporting data rates 5,000 times greater than those of the S-band multiple access system [38]. For missions that do not have space-to-space communications capability, the feasibility of establishing a control channel between an Earth-orbiting spacecraft and an omnidirectional antenna colocated at a NASA ground station site was evaluated in [39]. Future work will demonstrate the use of a low-rate commercial relay service as an additional control channel.

16.6.2 System-wide intelligence There has been a growth of commercial ground station and relay satellite networks in recent years, offering more potential links to a spacecraft than ever before. By adding scheduling interfaces for commercial networks, the Event Manager can schedule service with one of several networks on behalf of the spacecraft. This capability expands the capacity of communications service available to the spacecraft. Such a multiprovider framework offers a heterogeneous mix of links representing a trade-off between different characteristics. The Event Manager must choose the optimal link. System optimization could take place across many factors: availability, cost, latency, data volume, contact time, etc. Machine learning techniques have recently shown great promise in handling these types of multiobjective optimization problems including in the space communication environment [17]. Using these and

Cognitive communications for NASA space systems

231

similar techniques, the Event Manager could select the optimal link to meet the UIS request while balancing network load across multiple assets. An added benefit is that machine-learning techniques can detect service provider anomalies by examining mission performance across multiple providers or provider performance across multiple missions. Time-varying characteristics such as optimization for weather conditions affecting a given ground station could be considered. Furthermore, a cognitive engine learns from past allocations to improve mission communications performance over time.

16.7 Enabling technology 16.7.1 Reconfigurable hardware The hardware necessary to implement cognitive communications capabilities onboard spacecraft typically mimics the hardware that enables artificial intelligence and machine learning on the ground. Radiation-hardened space processors tend to be about two generations behind terrestrial processors. Examples of enabling cognitive processing capabilities for space include: ● ● ●

General purpose graphics processing units (GPUs) and multicore processors; Neuromorphic processors; and Field-programmable gate arrays (FPGAs).

In each case, these technologies must have low size, weight, and power (SWaP) for integration into the spacecraft communication system, with tight coupling to the functions of the spacecraft software-defined radio. Additionally, processor radiation tolerance is necessary for long-term survivability in the space environment, although software techniques (e.g., triple mode redundancy and regular system resets) can mask radiation effects to some extent. Neuromorphic processors implement a non-von Neumann computing architecture that utilizes analog and digital electronic circuits to mimic the neurobiological architectures present in the nervous system. Neuromorphic systems exhibit increased energy efficiency, execution speed, and robustness over traditional computing architectures, and can provide pattern recognition capabilities for SWaP-constrained applications [40].

16.7.2 Cognitive processing challenges Enhanced onboard processing of science data products reduces the amount of data transferred to a mission operations center, therefore reducing cost and demand on the network. High fidelity science instruments (e.g., synthetic aperture radar) are capable of generating volumes of data that far surpass a spacecraft communication system’s capability, requiring compression or preprocessing [41]. The machine learning algorithms discussed throughout this chapter may also require significant computation during the training process.

232

Advances in communications satellite systems 2: ICSSC-2019

16.8 Conclusion The objectives of cognitive links, networks, systems, and enabling hardware discussed in this chapter aim to provide increased autonomy and reliability for NASA’s communication architecture. This will require increased onboard processing in the space environment, eventually resembling a terrestrial cloud computing architecture. Instead of simply providing point-to-point links, the future architecture for space communication must include communication, processing, and storage. In such a service-oriented architecture with distributed cognition, all of the optimizations and concepts discussed in this chapter become possible. Machine learning and related automation technologies are a new thrust in space communication. Implemented correctly, these technologies have the potential to make communications networks more efficient and resilient in the challenging space environment as they have done on the ground. As the envisioned NASA space communications network evolves into a cognitive system-of-systems, this will enable improved science return, resource efficiency, and reliability for both missions and the communication network providers.

References [1] Schoolcraft, J., Klesh, A., and Werne, T. “MarCO: Interplanetary mission development on a CubeSat scale.” SpaceOps 2016 Conference. Daejeon, Republic of Korea. 2016. [2] Schwartz, J. and Feinberg, E. “STDN in the TDRSS and shuttle era.” IEEE Transactions on Communications. 1978; 26(11): 1506–1513. [3] Stegeman, J., Pham, N., and Bhasin, K. “How public private partnerships enable NASA future space communication needs.” 24th Ka and Broadband Communications Conference. Niagara Falls, Canada. 2018. [4] SpaceX’s 60-satellite launch is just the beginning for starlink megaconstellation project [Online]. Available from https://www.space.com/spacex-starlinksatellites-launch-just-beginning.html [Accessed September 12, 2019]. [5] Acosta, R. “NASA Ka-band RF propagation studies.” COST ICO802: Management Committee Meeting (MCM7). Portsmouth, London, United Kingdom. 2012. [6] Houts, J., Nessel, J., and Zemba, M. “Design of a Ka-band propagation terminal for atmospheric measurements in polar regions.” EuCAP Information Sharing at Kaiserslautern University, Kaiserslautem, Germany. 2016. [7] Hylton, A., Raible, D., Juergens, J., and Iannicca, D. “On applications of disruption tolerant networking to optical networking in space.” 30th AIAA International Communications Satellite System Conference (ICSSC). Ottawa, Canada. 2012. [8] Merriam-Webster. “Cognitive” [Online]. Available from https://www.merriam-webster.com/dictionary/cognitive [Accessed August 23, 2019].

Cognitive communications for NASA space systems

233

[9] Mitola, J. and Maguire, G. “Cognitive radio: Making software radios more personal.” IEEE Personal Communication Magazine. 1999; 6(4): 13–18. [10] Haykin, S. “Cognitive radio: Brain-empowered wireless communication.” IEEE Journal on Selected Areas in Communications. 2005; 23(2): 201–220. [11] Mitola, J. Cognitive Radio Architecture: The Engineering Foundations of Radio XML. Hoboken, NJ: Wiley; 2006. [12] Devroye, N., Vu, M., and Tarokh. V. “Cognitive radio networks.” IEEE Signal Processing Magazine. 2008; 25(6): 12–23. [13] Ivancic, W., Paulsen, P., Vaden, K., and Ponchak, D. “Cognitive networking with regards to NASA’s space communication and navigation program.” NASA Technical Report. Cleveland, Ohio. 2013. [14] Zimmermann, H. “OSI reference model–The ISO model of architecture for open systems interconnection.” IEEE Transactions on Communications. 1980; 28(4): 425–432. [15] Koch, M. and Downey, J. “Interference mitigation using cyclic autocorrelation and multi-objective optimization.” NASA Technical Memorandum. Cleveland, Ohio. 2019. [16] Downey, J., Mortensen, D., Evans, M., et al. “Adaptive coding and modulation experiment with NASA’s space communication and navigation testbed.” 34th AIAA International Communications Satellite Systems Conference. Cleveland, Ohio. 2016. [17] Ferreira, P., Paffenroth, R., Wyglinski, A., et al. “Multiobjective reinforcement learning for cognitive satellite communications using deep neural network ensembles.” IEEE Journal on Selected Areas in Communications. 2018; 36(5): 1030–1042. [18] Smith, A., Evans, M., and Downey, J. “Modulation classification of satellite communication signals using cumulants and neural networks.” IEEE Cognitive Communications for Aerospace Applications Workshop (CCAA). Cleveland, Ohio. 2017. [19] O’Shea, T., Roy, T., and West, N. “Approximating the void: Learning stochastic channel models from observation with variational generative adversarial networks.” ICNC 2019. Honolulu, HI, USA. 2019. pp. 681–686. [20] O’Shea, T. and Hoydis, J. “An introduction to machine learning communications systems.” Available from https://www.arXiv/1702.00832 [Accessed September 19, 2019]. [21] Smith, A. and Downey, J. “A communication channel density estimating generative adversarial network.” NASA Technical Memorandum. Cleveland, Ohio. 2019. [22] Anderson, A., Young, S., Reed, F.K., and Vann, J. “Deep modulation (Deepmod): A self-taught PHY layer for resilient digital communications.” Available from https://www.arxiv/1902.11218 [Accessed September 19, 2019]. [23] Cerf, V., Burleigh, S., Hooke, A., et al. “Delay-tolerant networking architecture.” IETF RFC 4838. 2007. [24] Scott, K. and Burleigh, S. “Bundle protocol specification.” IETF RFC 5050. November 2007.

234 [25] [26]

[27] [28] [29] [30] [31] [32]

[33]

[34]

[35]

[36]

[37]

[38] [39]

[40]

[41]

Advances in communications satellite systems 2: ICSSC-2019 CCSDS. “Protocol specification – IETF RFC 5050.” CCSDS Blue Book (Recommended Standard) 734.2-B-1. September 2015. Burleigh, S. “Interplanetary overlay network: An implementation of the DTN bundle protocol.” 2007 4th IEEE Consumer Communications and Networking Conference. Las Vegas, Nevada. 2007. Ramadas, M., Burleigh, S., and Farrell, S. “Licklider Transmission Protocol – Specification.” IETF RFC 5326. September 2008. Lent, R. “A cognitive network controller based on spiking neurons.” IEEE International Conference on Communications. Kansas City, Missouri. 2018. IEEE. Station and Media Access Control Connectivity Discovery. IEEE Standard 802.1AB-REV. June 2009. Birraine, E. Asynchronous Management Architecture. IETF Internet Draft. June 2018. Lent, R. “Analysis of bundle throughput over LTP.” IEEE 43rd Conference on Local Computer Networks. Chicago, Illinois. 2018. Wildermann, C. “cFE/CFS (core flight executive/core flight system).” November 13, 2008 [Online]. Available from https://ntrs.nasa.gov/search. jsp?R¼20090005965 [Accessed September 19, 2019]. Israel, D., Roberts, C., Morgenstern, R., et al. “Space mobile network concepts for missions beyond low earth orbit.” AIAA 2018 SpaceOps Conference. Marseille, France. May 2018. Reinhart, R., Schier, J., Israel, D., et al., “Enabling future science and human exploration with NASA’s next generation near earth and deep space communications and navigation architecture.” 68th International Astronautical Congress (IAC) 2017. Adelaide, Australia. 2017. Israel, D., Heckler, G., and Menrad, R. “Space mobile network: A near earth communication and navigation architecture.” 37th IEEE Aerospace Conference. Big Sky, Montana. 2016. Kazz, G. and Greenberg, E. “Mars relay operations: Application of the CCSDS Proximity-1 space data link protocol at Mars.” AIAA SpaceOps 2002 Conference. Houston, Texas. 2002. Mortensen, D., Roberts, C., and Reinhart, R. “Automated spacecraft communications service demonstration using NASA’s SCaN testbed.” SpaceOps Conference. 2018. Space Network User’s Guide, 450-SNUG, Rev. 10. NASA Goddard Space Flight Center, Greenbelt, Maryland. 2012. Gannon, A., Mortensen, D., Piasecki, M., and Downey, J. “On-orbit validation of a framework for spacecraft-initiated communication service requests with NASA’s SCaN testbed.” NASA Technical Report. Cleveland, Ohio. 2019. Schuman, C., Potok, T., Patton, R., et al. “A survey of neuromorphic computing and neural networks in hardware.” Available from https://www.arXiv Preprint arXiv:1705.069361 [Accessed September 19, 2019]. Bergeron, A., Doucet, M., Harnisch, B., et al. “Satellite on-board real-time SAR processor prototype.” International Conference on Space Optics – ICSO 2010. Rhodes Island, Greece. 2010.

Chapter 17

Supporting NASA Artemis 1 mission with JAXA Uchinoura station Timothy Pham1, Hiroshi Takeuchi2 and Atsushi Tomiki2

This chapter presents an ongoing effort in preparing JAXA Uchinoura station support to the Artemis 1 mission, scheduled for launch in late 2020. The system involves three key participants: JAXA ground station at Uchinoura, the Deep Space Network (DSN) components at the Jet Propulsion Laboratory, California, and the Artemis 1 mission navigation at the NASA Johnson Space Center, Texas. Demonstration of Uchinoura station support to the future Artemis signal relies on the use of a low-cost, highly-portable software-defined radio (SDR) test equipment as well as the tracking of the Lunar Reconnaissance Orbiter (LRO) spacecraft. Using the SDR equipment, we validated the compatibility of signal format between the Artemis flight radio and the Uchinoura ground station without having to send the flight equipment to the station. By tracking an ongoing operational spacecraft such as LRO, we were able to calibrate the performance of the system in real operational conditions. The measured Doppler noise of 0.03 Hz (1-sigma), or 0.002 m/s range rate at S-band, for Uchinoura station is deemed suitable to the Artemis 1 mission navigation needs. This chapter also discusses the test equipment capability and its performance. In addition to being low cost, the equipment offers many advantages compared to the traditional full-scaled test signal simulator. Chief among them is portability making system easy to set up and transport, and the fidelity of the test signal that it captures from spacecraft flight equipment. Some of the lessons learned, such as internal frequency stability of the test signal, are also reflected. Key Words: EM-1; Uchinoura; DSN; three-way; Doppler

1 2

Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA JAXA Institute of Space and Astronautical Science, Yoshinodai, Chuo-ku, Sagamihara, Japan

236

Advances in communications satellite systems 2: ICSSC-2019

17.1 Introduction In late 2020, the National Aeronautics and Space Administration (NASA) plans to launch the Artemis 1 spacecraft, previously known as Exploration Mission-1, on a 3-week trajectory to the Moon. The spacecraft will stay in a distant retrograde orbit around the moon for 6–10 days and then return to Earth [1]. Artemis 1 is a robotic precursor to the Artermis 2 mission which will be carrying astronauts for a flyby the Moon, as part of the NASA’s plan for human return to the Moon. In future crewed missions, precise navigation, especially for the Earth-bound return flight segment, would be critically important to ensure successful return of astronauts to Earth. Tracking Doppler data from the Japan Aerospace Exploration Agency (JAXA) antenna at Uchinoura will supplement data received from the NASA Deep Space Network (DSN) by providing additional baseline configurations among all tracking antennas to increase the accuracy of orbit determination.

17.2 Operation concept Figure 17.1 shows the configuration for both DSN and Uchinoura support to Artemis 1. The DSN 34-m antennas at the three tracking communications complexes at Goldstone, Canberra, and Madrid will provide two-way Doppler data, as well as ranging data, to the Artemis navigation team at the Johnson Space Center (JSC). In addition, the JAXA 34-m Uchinoura antenna will provide three-way Doppler data (no ranging) to Artemis navigation. Three-way Doppler refers to a configuration where the uplink signal to the spacecraft comes from a station that is different from the downlink. Uchinoura station also has a smaller 20-m antenna that can serve as a backup to the 34-m in the event of failure. Both antennas were included in the tests. In terms of the sequence of activities, a typical Artemis tracking pass would involve the following steps. All steps would need to be tested prior to declaring the system operational. First, the Artemis navigation team submits the spacecraft ephemerides to a service-preparation data server at JPL that is responsible for

3-way Doppler with JAXA Uchinoura

2-way Doppler with DSN

Mission NAV

Figure 17.1 Configuration for Artemis support

Supporting NASA Artemis 1 mission with JAXA Uchinoura station

237

preparing necessary data products to enable the DSN antennas to track spacecraft, for example, predicted antenna pointing, predicted signal conditions, and configuration to the ground equipment to best match the spacecraft signal. The ephemeris data are then retrieved by the Uchinoura operation team and translated into predicted pointing angles for Uchinoura antenna. Once the Uchinoura antenna points and locks to Artemis signal, the received carrier frequency data are delivered to a tracking data server at JPL (which also receives tracking data from the DSN antennas). The Artemis mission navigation at JSC then retrieves data from all antennas involved in the track. From using the uplink frequency generated by the DSN antenna and the downlink frequency received at Uchinoura antenna, threeway Doppler can be computed. Similarly, two-way Doppler data can be derived from the uplink and downlink frequency of the DSN antenna.

17.3 Recording/playback test equipment Current commercial-off-the-shelf SDRs have sufficient capabilities to serve as a cost-effective solution for mission development. The benefits come from faster development and lower implementation cost compared to what is typically required for a custom-design equipment. Another key benefit that we found useful is portability. The whole RPA weighs slightly more than a laptop computer, making it very easy to hand carry to the ground station under test which is typically at a remote location to minimize radio frequency interference. This reduces the logistic burden associated with the preparation of equipment shipping and installation. Using the COTS products, we assembled a system that can record an RF/IF signal and playback the recorded data at a selectable frequency within the 10 MHz–6 GHz range. The three key components of the recorder/playback assembly (RPA) are shown in Figure 17.2: a laptop computer with fast input/output throughput for high-rate data transfer between the SDR transceiver and the data storage; a high capacity, high throughput data storage (multi-TB as dictated by the application need); and a commercial SDR transceiver.

External clock input

RF input/output

RF transceiver

Data storage drive

Figure 17.2 Key components of the SDR recorder/playback

238

Advances in communications satellite systems 2: ICSSC-2019

The computer controls the operation of the transceiver. It provides configuration setting for signal recording as well as signal playback. Typical control parameters are done via the graphical user interface and involve: (1) the gain setting of signal amplifiers in various stages of processing at baseband, intermediate frequency (IF), and the radio frequency (RF); (2) sampling rate of the signal to be recorded/played back; (3) local oscillator frequency for down-conversion from signal input to baseband as in the case of recording or for up-conversion from baseband to the output frequency as in the case of playback; and (4) the data file for recording or playback. A fast Fourier transform (FFT) spectrum of the recorded or playback signal is available during runtime to validate the signal presence. The transceiver-control software is leveraged on modules available in the public-domain GNU library. This speeds up the development process and minimizes the implementation effort—another big benefit for using the commercial SDR. The disk storage, externally connected to the laptop, archives the data samples. Both the computer and disk storage use a high-speed USB interface for data transfer. At the highest sampling rate of 20 MHz, the data throughput can be as much as 160 MB/s. The most crucial element of the RPA is RF transceiver [2]. On the recording, it is equivalent to a heterodyne receiver where the RF signal is first down-converted to baseband frequency and then digitized. On the playback, the functions are reversed. The particular commercial unit we employed in the RPA can digitize an RF signal input up to a rate of 20 Msps, I/Q sampling. The use of GNU library necessitates the need to keep the digitized samples with standard 32-bit (4 bytes) format. This prompts a high I/O rate (up to 160 MB/s at the highest sampling rate) and a large data storage capacity. Our particular chosen SDR supports an input/ output RF frequency range of 1 MHz–6 GHz. This allows the signal recording, or playback, directly at S-band frequency for space missions (2.2–2.3 GHz) that the Uchinoura ground station expects to be receiving from the Artemis 1 spacecraft. The SDR transceiver can operate with a clock from an internal oscillator or an external 10 MHz reference. Since the SDR expects a low voltage transistor– transistor logic (LVTTL) input level for the frequency reference input, a custom-built adapter that converts a sinusoidal waveform at (þ13 dBm) to that of LVTTL is necessary. This adapter essentially is a resistor-capacitor-diode network to level shift and it clamps the signal level to the 0-V minimum and 3.3-V maximum, as required for LVTTL compatibility. The use of an external frequency reference from a highly stable clock, such as those produced by a Hydrogen maser at the DSN and Uchinoura ground stations, produces a significantly more stable signal. Signal stability performance with and without the external reference is shown later in the chapter. We used the RPA to capture signal from Artemis flight communications system during the DSN RF compatibility testing with Artemis. To conduct compatibility testing at Uchinoura, we played back the RPA recorded signal at RF (2.3 GHz) and injected it into Uchinoura station via a test coupler, as if it came from the spacecraft (Figure 17.3).

Supporting NASA Artemis 1 mission with JAXA Uchinoura station

239

RF compatibility testing Record RPA Playback

Figure 17.3 RPA recording and playback for Artemis EM-1/UCS-34m on DOY335/2017 400

Costa loop tracking via symbols

Residual carrier phase lock loop tracking

300 Carrier frequency 2,216,497,500, Hz

TLM-2-1a, 72 ksps 200

TLM-2-1a, 72 ksps (30 Hz PLL BW)

RNG-2-3a, 1Msps

TLM-2-3a, 1 Msps

TLM-2-1a, 72 ksps (300 Hz PLL BW)

TLM-2-3a, 1 Msps

100 0 –100 –200

TLM-2-9, 2 Msps (30 Hz PLL BW)

–300 TLM-2-9, 2 Msps

TLM-2-9, 2 Msps (300 Hz PLL BW)

–400 TLM-2-11, 4 Msps (RCV tracked at 2 Msps)

–500 –600 3.50

4.00

4.50

5.00

5.50

6.00

Time, (h)

Figure 17.4 Detected frequency of Artemis signal, December 2017

17.4 Result of Artemis compatibility test Under nominal configuration, Artemis signal consists of a residual carrier signal and telemetry data that are direct carrier modulated. The recorded data, with four representative telemetry data rates ranging from 72 ksps to 2 Msps, were played back at the spacecraft expected downlink frequency at S-band and injected into the front of the low noise amplifier of the Uchinoura ground station. Artemis signal has another high rate configuration at 4 Msps and Offset QPSK modulated on the carrier. This data rate is beyond the capability of the receiver in Uchinoura; thus, it is not supported. The impact is negligible because this configuration is not suitable to DSN ranging measurement and thus would not be present during Uchinoura three-way support. The first series of tests with Artemis recorded signal was done in December 2017. The detected carrier frequency, which represents Doppler information in the normal spacecraft tracking passes, is shown in Figure 17.4. The result demonstrated

240

Advances in communications satellite systems 2: ICSSC-2019 Frequency vs. time, EM-1 signal, Jan. 23, 2019 2.0418E+09

Costa loop on telemetry data

Phase lock loop on carrier

2.0416E+09

Frequency, (Hz)

2.0414E+09 2.0412E+09

TLM-2-2, 384 ksps

RNG-2-3, 1 Msps

TLM-2-2, 384 ksps

TLM-3-1, TLM-2-4, TLM-2-1a, RNG-2-3, 1 Msps 2 Msps 72 ksps 1 Msps

2.0410E+09 2.0408E+09

TLM-2-1, 72 ksps

2.0406E+09 2.0404E+09 2.0402E+09 12

12.5

13

13.5

14

14.5

15

UTC time, (h)

Figure 17.5 Detected frequency of Artemis signal, January 2019

that Uchinoura system can acquire and lock to the Artermis spacecraft signal carrier, either directly on the residual carrier with the phase-locked loop or using telemetry data with the Costa loop. The cyclic appearance within a test run of a single data rate was caused by using the data repeat option during the playback. Within a data segment (without data repeat) of a single run of the same data rate, we observed a significant frequency drift. This was different from signal’s constant frequency used in the original DSN RF Compatibility test. We suspected that the drift was caused by an inherent instability of the RF transceiver’s internal oscillator since the recording was originally done without using external frequency reference, rather than caused by the Uchinoura ground equipment. Further testing with and without the external reference demonstrated that the frequency drift indeed came from the transceiver internal oscillator. We observed that without the external reference, the drift was as much as 60 Hz over 14 min. With the external reference, the drift was reduced to a minimal level of 70 mHz over 2.5 h. As seen in Figure 17.5, a repeat of the recording and playback with external reference shows that the drift was eliminated. Segments of receiver in-lock are shown as a straight line. Segments of high-frequency variation correspond to periods where the receiver being out of lock (as expected) because of transition among different data rates, thus, are not a cause of concern.

17.5 Result of LRO spacecraft tracking We were able to track the LRO spacecraft in two occasions. The Doppler data were then processed by the LRO navigation team to remove errors caused by spacecraft

Supporting NASA Artemis 1 mission with JAXA Uchinoura station

241

DS38 LRO AVG = + 1. 4332 DEV = 0.0240 STD = 0.0297 RSS1σ = 1.4336 PTS = 792

RR2 Doppler (Hz) residual

1.55

1.5

1.45

1.4

1.35

1.3 19:05

19:10 Unedited

19:15 23-Jan-2019 (UTC) Edited

Elevation

19:20

19:25

Invalid

Figure 17.6 LRO Doppler residuals, January 2019

dynamic and trajectory. Figure 17.6 shows a post-fit Doppler residual with a standard deviation of 29.7 mHz at 5-s integration. This corresponds to a range rate (velocity) uncertainty of 2 mm/s. Such a performance is deemed useful to the orbit determination by the mission navigation team. It should be noted that the Doppler data have a bias of around 1.4 Hz. The cause is still under investigation.

17.6 Conclusion In summary, this chapter presents an ongoing development effort to enhance NASA Artemis 1 mission support with the use of the JAXA Uchinoura 34-m antenna. Using the SDR recorded and playback EM-1 signal, we demonstrated that the system will be able to track Artemis 1 signal and the system Doppler performance is sufficient for the navigation need.

Acknowledgments The work described in this chapter was carried out by the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. Support for testing was provided by the Japan Aerospace Exploration Agency. This recording/playback equipment was originally developed by Leslie White at the Jet Propulsion Laboratory. Doppler

242

Advances in communications satellite systems 2: ICSSC-2019

residuals were processed by Juan Crenshaw at the NASA Goddard Space Flight Center.

References [1] Hack, R.F. Available from https://greatscottgadgets.com/hackrf/ [Accessed September 15, 2019]. [2] Artemis 1 Map. Available from https://www.nasa.gov/image-feature/artemis-i-map [Accessed September 15, 2019].

Section 5

High-speed optical communications and feeder links 1

This page intentionally left blank

Chapter 18

Implementation of the method for estimating propagation direction of laser beam transmitted from ground to satellite Hiroki Yamashita1 and Yoshihisa Takayama1

An implementation of the method to estimate the direction of the laser beam emitted from ground to satellite is proposed. The implementation employs two boresight cameras to observe the backscattered light of the emitted beam. To carry out the proposed method, it is necessary to estimate the center line of the image of the scattered light from the laser beam taken by the boresight camera. A simple method to find the center line of the image of the backscattered light is also proposed. Since the accuracy of the estimation of the beam propagation direction should be known to design a free-space optical terminal, an experiment is performed to measure the angular error of the estimation by the proposed method. Key Words: free-space optical communication; communication laser beam pointing; acquisition and tracking; boresight camera; Rayleigh scattering

18.1 Introduction Recently demands for broadband data communications in space missions have increased. Free-space optical communications (FSOCs) are one of the promising methods to satisfy the demands by employing the sharp laser beam for communications to increase gain in the link budget [1,2]. Thus, the diameter of the beam illuminating the communication partner is small and precise control of beam direction is necessary for FSOC systems. Especially, in the initial acquisition sequence, the narrow beam should accurately irradiate the counter communication terminal. To make the beam transmission direction highly accurate, both the well-aligned optical setup for transmission and the real-time monitoring of the emitted beam direction are necessary. In the case of establishing the ground to satellite FSOC link, the backscattered light of the emitted laser beam is observed by a boresight camera mounted on a 1

Graduate School of Science and Technology, Tokai University, Tokyo, Japan

246

Advances in communications satellite systems 2: ICSSC-2019

telescope of the ground terminal to confirm the propagation direction [3,4]. In this method, the focus of the boresight camera is set to infinity by looking at the direction parallel to the emitted laser beam, where the backscattered light is observed as a wedged shape. The center line of the wedged shape is a projection of the beam propagation path onto a two-dimensional plane. However, it is impossible to determine the propagating direction of the emitted laser beam with the method that uses a single boresight camera. Because we cannot observe the tip of the emitted laser beam due to the atmospheric scattering that is not occurred in the space. A method to resolve the problem has been proposed in 2019 [5]. In this method, the backscattering of the emitted laser beam is observed from two different viewpoints by using dual boresight cameras to estimate the propagation direction of the emitted beam in the space. This chapter gives the mathematical explanation to show the validity of this method but no experimental demonstrations are conducted to evaluate the measurement accuracy of the beam propagation direction. In this chapter, we experimentally estimate the propagating direction of a transmitted light where two images are taken from different viewpoints. We show an implementation of the optical set to estimate the propagation path of the emitted laser beam in the space by using both images obtained from different viewpoints. To apply the proposed method to operational FSOC systems, the real-time image processing is necessary for feedback control of the telescope direction. Therefore, we also propose a simple data processing algorithm based on the isotropic characteristic of the Rayleigh scattering. Then, we estimate the accuracy of the emitted beam direction that is measured with the proposed optical set and the proposed data processing algorithm.

18.2 Estimating beam propagation direction using dual boresight camera In the proposed method, the Rayleigh scattering of the emitted laser beam is observed by using two boresight cameras to estimate the laser beam direction. Figure 18.1 depicts a setup to observe the backscattering of the emitted laser beam. Where z is the direction of the laser beam emitted, CAM2 is placed on h axis, and CAM1 is placed on x axis. These cameras look at the direction parallel to the laser beam emitted direction and the focus is set to infinity. Images taken by CAM1 and CAM2 show the scattered light of laser beam projected to focal planes of these cameras, respectively. Figure 18.2 depicts the relation of these focal planes and scattered light from the emitted laser beam. In this figure, the lenses of CAM1 and CAM2 are assumed ideally thin for simplicity. A1 and A2 are respectively the principal points of the lenses of CAM1 and CAM2. Point Pt is the target of the laser transmission. The area surrounded by dotted lines Si (i ¼ 1,2) is the focal plane of CAMi. Solid lines depicted in the area Si are projected images of scattered light of the laser beam. Points pti are the projection image of the target point Pt on the area Si. According to the geometrical optics, the scattered light at point T1 is projected to the point ti1 and the scattered light at point T2 is projected to the point ti2.

Estimating direction of laser beam emitted from ground to satellite

CAM1

247

ζ

ξ

Laser CAM2

η

Figure 18.1 The alignment of the laser transmitter and boresight cameras to observe the scattered light z ζ

Pt

T1

T2 A2 t22 A1 t11 t12

t21 S1

S2

η pt2

pt1 ξ

Figure 18.2 The coordinate definition to determine the direction of the laser beam using the image taken by boresight cameras

Generally, an arbitrary straight line in three-dimensional space can be defined as the line of intersection of two planes. In Figure 18.2, the line representing the direction z is included in both the plane A1-t11-t12 and the plane A2-t21-t22. Therefore, the line regarded as trajectory of the laser beam in three-dimensional space is defined as the line of intersection of the planes defined with the center line of projected images of scattered laser beam and principal point of camera lenses. The proof with mathematical expression is given in [6]. In order to have the laser beam propagate to the target point Pt, the center line of the projected image of scattered light should be adjusted to direct to the pti in both focal planes S1 and S2 simultaneously.

248

Advances in communications satellite systems 2: ICSSC-2019

18.3 Experimental method to measure the angular error Accuracy of the estimation of the emitted laser beam direction is important to secure reliability of the optical ground terminal when acquiring and tracking the satellite as the counterpart of communications. Assuming the standard atmosphere, the accuracy has been estimated in [5] by taking the dependence of Rayleigh scattering on altitude and the angular error caused by pixel pitch of the image sensors of the boresight cameras into considerations. In this section, the angular error of estimating the laser beam direction using dual boresight camera is confirmed experimentally.

18.3.1 Experimental method To measure the angular error of the proposed method, we emit a laser beam and adjust the direction to point a target located in far enough to be considered infinite with the dual boresight cameras. Then, we irradiate a screen located several meters away from the laser beam and measure the distance between the pixel corresponding to the infinitely far target and the pixel of the beam centroid on the screen. The relative positions of the two cameras and the laser transmitter should be kept. Figure 18.3 shows the experimental method. Figure 18.3(a) is a schematic of adjusting the laser beam direction to the far target. Where Pt is the position of the target, pti (i ¼ 1,2) is the pixel of the projected image of the target point Pt. The dashed lines in CAM1 and CAM2 images are the center lines of the projected images of the scattered light. Figure 18.3(b) is a schematic of the measurement of Pt

Laser

CAM1 image pt1 ps1

CAM1

ps2

pt2 pt1 CAM1 image (a)

∆2 CAM2 image

Screen

CAM1

∆1

CAM2

Ps

CAM2 pt2

CAM2 image

Laser 50 m

(b)

Figure 18.3 (a) and (b) are schematics of the experiment method of the pointing far target and the irradiating of the screen with the laser beam, respectively

Estimating direction of laser beam emitted from ground to satellite

249

the distance between the pixel of the projected far target and the pixel of the projected actual beam centroid in 50 m away. Where Ps is the point of the beam centroid on the screen, psi is the pixel of the projected image of the target point Ps. Di is the distance between the psi and the dotted line.

18.3.2 Geometrical angular error If there is no error in the adjustment of the laser beam direction to point the target located infinitely far, the points ps1 and ps2 should be on the center line of the image of scattered light from the laser beam exactly. However, the error caused by the use of the digitized image sensor cannot be eliminated from the measurement results. Figure 18.4 depicts a relationship between the lens and image sensor of the camera. In this figure, the camera lens is represented by a single lens for simplicity. Where f is the focal length of the camera lens, dp is the pixel pitch of the image sensor, and qp is the angular width covered by a single pixel. Therefore, the proposed method has angular error at least angular width defined by the qp. When dp is smaller than f enough, the qp is   dp 1 dp (18.1) qp ¼  tan f f

18.3.3 The procedure of the experiment 18.3.3.1 The setup for this experiment We construct an optical setup to steer the direction of the emitted laser beam with dual boresight cameras on a breadboard. Figure 18.5 is the overview of the optical setup for this experiment. Where using the same CCD cameras for boresight observation, the output fiber of the laser source is connected to the fiber collimator to emit a laser beam. The fiber collimator mounted on the two-axis gimbal to steer the direction of the laser beam. Regarding the center of the aperture of the fiber collimator as the origin of the coordination in Figure 18.1, the CAM1 is aligned in Image sensor

dp

Camera lens

θp

f

Figure 18.4 The schematics of the angular width of the received light detected by the same pixel of the image sensor

250

Advances in communications satellite systems 2: ICSSC-2019 CAM2

Laser source

Fiber collimator

CAM1

Figure 18.5 The optical setup to observe the scattered light from the emitted laser beam with dual boresight camera Table 18.1 The specifications of the setup for the experiment Parameters

Value

Wavelength of the laser beam Focal length of camera lenses Pixel pitch of CCD cameras Image size of CCD cameras

785 nm 200 mm 8.4 mm(H)  9.8 mm(V) 720(H)  480(V) pixels

x–z plane and the CAM2 is aligned in h–z plane. Some specifications of this experimental setup are given in Table 18.1.

18.3.3.2

The method to obtain the center line of image of scattered light

To carry out the proposed method, we should identify the center line of the scattered light from the images taken by the boresight cameras. In this section, a simple method to obtain the line in the image data regarded as the direction of the laser beam propagation is described. Figure 18.6 is a schematic of the image data taken by one of the boresight cameras, where the solid line is the center line of the image of the scattered light. L1 and L2 are the slice of the image data of 1-pixel width chosen to intersect the solid line. The ay and by are the centers of gravity of the pixel values of the slice data L1 and L2, respectively. Furthermore, ax and bx are coordinates of each slice data in the image. According to the definitions and the isotropic characteristics of the Rayleigh scattering, both the points of a(ax, ay) and b(bx, by) are on the solid line. Generally, a straight line in a two-dimensional plane can be determined by two points. Therefore, the mathematical expression of the center line of the scattered light in the image data is

Estimating direction of laser beam emitted from ground to satellite

a

ay by

b

L1 ax

251

L2

bx

Figure 18.6 The schematics of the estimating the center line of the image of the scattered light in the image data taken by the boresight camera y ¼ ax þ b (18.2) a¼

ay  by ax  bx

b ¼ ay  aax

(18.3) (18.4)

18.3.3.3 Observation for the experiment In the first step of the experiment, the direction of the laser beam is adjusted to point the infinitely far target. A star is employed for the target. At this time, the focus of the camera lenses is set to infinity. Then, the direction of the laser beam is determined by adjusting the center line of the image of the scattered light to the pixel corresponding to the star in both CCD cameras simultaneously. Figure 18.7 shows image data which are taken by the CCD cameras before adjusting the laser beam direction. There is no star on the center line of the image of the scattered light. Figure 18.8 shows image data which are taken by the CCD camera after adjusting the laser beam direction. There is the target star on the center line of the scattered light. Figures 18.7(a) and 18.8(b) correspond to CAM1 and CAM2, respectively. To confirm the direction of the emitted laser beam, we irradiate a screen by the laser beam and observe the beam centroid on the screen by the boresight CCD cameras where the focus of the camera lenses is set on the screen. Therefore, the pixel on which the beam centroid is projected should move along the center line of the image of the scattered light. Furthermore, we employ a graph paper as a screen irradiated by the laser beam to measure the physical distance par 1 pixel. Figure 18.9 shows the image data taken by the CCD cameras to investigate the position of the emitted laser beam. Figure 18.9(a) and (b) correspond to the images taken by CAM1 and CAM2, respectively.

18.3.4 Results Table 18.2 shows the angular errors of the estimated beam propagation direction by using the dual boresight cameras. The measured D1 and D2 are 14 pixels and 6 pixels, respectively. Therefore, the angular error in this experiment is 58.38 m rad

252

(a)

Advances in communications satellite systems 2: ICSSC-2019

(b)

Figure 18.7 The image taken by the boresight camera before adjusting the emitted beam angle to direct the star

(a)

(b)

Figure 18.8 The image taken by the boresight camera after adjusting the emitted beam angle to direct the star

(a)

(b)

Figure 18.9 The image taken by the boresight camera with the screen irradiated by the emitted laser beam and 23.28 m rad with CAM1 and CAM2, respectively. On the other hand, the angular errors predicted by the geometrical parameters are 42 m rad and 49 m rad along the horizontal axis and the vertical axis of the CCD camera, respectively.

Estimating direction of laser beam emitted from ground to satellite

253

Table 18.2 The results of the measurement of the angular error in the proposed method Parameters

CAM1

CAM2

Angular error Position of the pixel corresponding to the far target Position of the beam centroid on the screen Distance between beam centroid on screen and the center line of the scattered light

58.38 m rad 269,184 pixels

23.28 m rad 406,168 pixels

365,170 pixels 14 pixels

400,168 pixels 6 pixels

Assuming the pixel of the CCD camera is square, the maximum value of the geometrical angular error is tan1(H8.4 þ 9.8/(200  103)) ¼ 64.53 m rad. Therefore, the angular error of the beam propagation direction adjusted by the dual boresight cameras is within the angular error caused by the use of the digitized image sensor. This result suggests the effectiveness of applying the proposed method to the FSOC ground terminal. However, future work will corroborate the accuracy of the estimation in the center line of the image of scattered light from the laser beam in image data. The current method only uses the luminance center of gravity. In this method, the estimation results may fluctuate due to the effects of light from other stars and light pollution.

18.4 Conclusion In this chapter, an example of the implementation of the method which determines the propagation direction of the emitted laser beam by using the dual boresight cameras from the transmitter side is presented. The accuracy of the estimation of the propagation direction of the transmitted beam is evaluated. The angular error has not exceeded the value theoretically predicted. The experimental result also suggests the effectiveness of the proposed method.

References [1] Toyoshima, M. “Trends in satellite communications and the role of optical free-space communications.” Journal of Optical Networking. 2005; 4(6): 300. [2] Dan Williams, W., Collins, M., Boroson, D.M., et al. “RF and optical communications: A comparison of high data rate returns from deep space in the 2020 timeframe.” NASA Technical Reports NASA/TM-2007-214459. 2007. [3] Toker-Nielsen, T., and Oppenhauser, G. “In-orbit test result of an operational optical intersatellite link between ARTEMIS and SPOT4, SILEX.” Proceedings of SPIE. 2002; 4635(2002): 1.

254

Advances in communications satellite systems 2: ICSSC-2019

[4] Smutny, B., Lange, R., Ka¨mpfner, H., et al. “In-orbit verification of optical inter-satellite communication links based on homodyne BPSK.” Proceedings of SPIE. 2008; 6877(2008): 687702-1. [5] Takayama, Y., Takenaka, H., and Kunimori, H. “A simple improved method using dual boresight cameras in verifying propagation direction of laser beam transmitted from ground to satellite.” Physica Scripts. 2019; 94: 065001. [6] Yoshihisa, T. “Confirmation of propagation direction of uplink beam in ground-to-satellite laser transmission.” The Online Proceedings of the 24th Ka and Broadband Communications Conference. 2018, Ka6–3.

Chapter 19

Studies on site diversity to mitigate cloud blockage in satellite-ground optical communications based on long-term ground meteorological observation data Yuki Ueda1, Tatsuya Mukai2 and Yoshihisa Takayama1

Cloud blockage in Japan area is analyzed with the long-term ground-based observation data to find candidate sites for the site diversity scheme. First, the period in which both ground-based observation data and satellite-based observation data are available is confirmed. Next, from the viewpoint of the average cloud amount, and in consideration of the installation conditions of the optical ground station, some appropriate candidate sites are selected. Finally, the cloud amount correlation values among the selected sites are obtained, and the final candidate site combinations are listed. Key Words: optical direct communication; cloud blockage; optical ground station; site selection for effective diversity; network switching controls

19.1 Introduction In the field of high-speed space communication, the free-space optical communications draw attention as a technology to provide higher data rate and smaller equipment size than conventional radio communications. In the free-space optical communications, the cloud causes blockage of the communication links, while the atmospheric impact on the communication links causes degradation of link quality. From this point of view, it is necessary to avoid the cloud blockage to keep communication links without link outage. The site diversity technique that employs multiple optical ground stations is considered as a method to mitigate this issue on cloud blockage and the network switching control technique from station to station according to cloud coverage on a given satellite path over a given optical ground station will be an additional 1 2

Tokai University, Tokyo, Japan Japan Aerospace Exploration Agency, Ibaraki, Japan

256

Advances in communications satellite systems 2: ICSSC-2019

method to avoid the cloud blockage in communication links between an optical terminal mounted on a satellite and an optical ground station [1]. In Japan, there are two types of meteorological observation data which are based on ground observation of cloud amount and satellite observation of cloud images viewed from a geostationary orbit. Therefore, they can be used for this kind of study on cloud blockage for site diversity based on cloud amount over domestic locations in Japan. The study of global candidate locations for space optical direct communications are reported by the Optical Link Study Group (OLSG) formed in IOAG (Interagency Operation’s Advisory Group) [2]. In addition, a study on the necessary number of optical ground stations and their locations to realize the site diversity for space optical direct communications in Japan is also reported based on the analysis of cloud images observed from meteorological satellites called Himawari operated by Japan Meteorological Agency (JMA) [3]. The analysis of the ground observation data of cloud amount for this purpose is somewhat limited and the time length of data analysis based on Himawari satellites is also not sufficiently long compared with the ground-based observation of cloud amount [4]. From this point of view, it may be necessary to analyze satellite observation data again based on longer time length at least more than 1 year and also to compare the result of satellite observation data analysis with that of ground

Meteorological GEO satellite Expected image from space Satellite - based analysis High cloud

Low cloud Expected image from ground

Ground - based analysis Weather station

Weather station Weather station

Figure 19.1 Overview of cloud observation method in Japan

Cloud blockage in satellite-ground optical communications

257

meteorological data analysis over Japan because of difference of cloud layers observed by both of satellites and ground observations. Figure 19.1 shows the overview of the cloud observation method in Japan from ground and space. Figure 19.2 shows the overview of site diversity for space optical direct communication system, where one of three optical ground stations without cloud blockage is communicating with a satellite and two other optical ground stations cannot communicate with it because of cloud blockage by lower or higher cloud. In order to realize higher operation availability of optical direct communications, it is very important to locate optical ground stations at separate locations under small cloud amount as possible as a site diversity method and to switch the ground network connections according to the cloud amount status at each location. In parallel to the study of the network switching control method [1], we think it important that the study of site diversity also needs to be performed again starting from the long-term ground-based observation data of cloud amount and following the satellite-based image analysis to get an overall result of site diversity based on both point of views from ground and space. Both study approaches will be combined later to reinforce our study of optical ground station network system for space optical direct communications to tolerate cloud blockage between space and ground.

Optical terminal on a satellite

Optical ground stations (site diversity)

Figure 19.2 Overview of site diversity for space optical direct communication system

258

Advances in communications satellite systems 2: ICSSC-2019

As a part of the first study of site diversity based on the ground-based observation data of cloud amount, this study shows the result for domestic site diversity in Japan based on long-term ground observation data corresponding to only lower cloud blockage in space optical direct communications. Future cloud analysis at the same locations and during the same period based on satellite observation data corresponding to the higher cloud is necessary to finally understand domestic candidate locations based on both analyses from ground and space. In this chapter, we first investigate the period in which both ground-based observation data and satellite-based observation data exist, the ground-based observation cloud amount data is analyzed, and candidate locations are determined from the viewpoint of the average cloud amount. Following the above process, we select more appropriate candidate locations in consideration of the installation conditions of optical ground stations. Finally, the cloud amount correlation values among the selected candidate locations are obtained, and the combinations of the candidate locations are analyzed.

19.2 Methodology 19.2.1 Investigation of cloud amount data in Japan In Japan, there are two types of cloud observation data: ground-based cloud data and satellite-based cloud data and the period when both ground-based and the satellite-based data exist is constrained. To find out such suitable candidate locations with small cloud amount, it is necessary to confirm the period where both types of cloud data exist as long as possible because cloud blockage occurs during relatively short periods in a given transmission path between space and ground. The ground-based cloud data have ever been created by ground observers of JMA based on their cloud amount measurement and exists from 1961 and later increased up to 90%. This type of cloud amount measurement has been performed every 3 h as a maximum observation cycle, but the number of ground-based observation sites gradually started to decrease from 2001 and it reached the percentage less than 50% of all ground-based observation sites in 2009. Therefore, we concluded that we need to use cloud amount data before 1998. The satellite-based cloud data exists from 1977 when the first Geostationary Meteorological Satellite (GMS) was launched. The full observation operation started in 1978 and the data started to increase from 1981 and more than 90% of data are dominant in Japan compared with the decreased number of the ground-based data. A series of GMS satellites, performed their observations by 1995, have an observation sensor called visible and infrared spin scan radiometer (VISSR) with observation band (IR band: 10.5 mm to 12.5 mm; Visible band: 0.5 mm to 0.70 mm) and furthermore, the observation band of the meteorological satellite called Himawari No. 5, performed its observation from 1995 to 2003, was improved by a new observation band allocation (Visible band: 0.55 mm to 0.90 mm, IR1: 10.5 mm to 11.5 mm, IR2: 11.5 mm to 12.5 mm, IR3: 6.5 mm to 7.0 mm). The currently available satellite-based cloud amount data in IR band is from 1981 and that in Visible band is from 1987.

Cloud blockage in satellite-ground optical communications

259

Figure 19.3 shows the investigation result of both cloud observation data from 1961 to 2019 in Japan and the period when both ground and satellite-based data exist enough is shown with a red-colored dotted square. Each colored bar (green, blue, yellow, and orange) means the percentage of available data in each type of data. Our investigation resulted in that the 10 years data from 1988 to 1997 is the longest data with both types of data. Therefore, we decided to focus on this 10-year period considering also about the future analysis of satellite-based data in this same period. As mentioned before, the ground-based observation data of cloud amount has been ever performed every 3 h as a maximum observation cycle during day and night by JMA ground observers. It means it is divided into day and night observation data. In addition, the satellite-based observation data are also divided into day and night data. For the cloud amount analysis during day time is possible by using the Visible band and that during night time is possible by using the IR bands. In our analysis of ground-based data of cloud amount includes the data during day and night time. From this point of view, we need to perform the future data analysis of satellitebased data not only during night time but also during day time to confirm the result of data analysis of ground-based data.

19.2.2 Domestic candidate locations analyzed by ground-based cloud amount data in Japan For selection of suitable locations of optical ground station for space optical direct communications, we need to focus on cloud amount during the selected 10 years period at each observation location. This is because we need to put optical ground stations at separate locations with small cloud amount as possible in order to avoid cloud blockage between space and ground. For this analysis, the 10-year period from1988 to 1997 is selected because longest data of both ground and satellite exist enough and we decided to analyze the cloud amount at 155 observation sites during

1) Ground based data –JMA observer Longest time period overlapped 2) Satellite based data –GMS/VISSR (IR)

Over 90%

3) Satellite based data –GMS/VISSR (VIS)

Over 50%

Over 80%

Less than 50% 4) Satellite based data –MTSAT/Himawari

Figure 19.3 Comparison of the existence of both cloud observation data from 1961 to 2019 in Japan

260

Advances in communications satellite systems 2: ICSSC-2019

the selected 10 years period and selected such locations where the averaged cloud amount is as small as possible including a large number of locations for forming a site diversity with optical ground stations located separately at long distance to each other. The data of the cloud amount used for this analysis is from the meteorological database of the ground-based observation supervised by Japan Meteorological Business Support Center and it is categorized by 11 levels from 0 to 10 determined by JMA. Level 0 means a blue sky without cloud and level 10 means being cloudy in the whole sky. Based on this classification, the average cloud amount is obtained by the following formula (19.1):  x¼

1 Xn x i¼1 i n

(19.1)

In the equation, n is the number of data at the weather station, xi is the ith observed data, and x is the average cloud cover. Based on formula (19.1), with 11 levels, there are only three locations (Irako, Nobeoka, and Minamitorishima) found under the conditions of cloud amount less than level 6. In fact, it is very difficult to find out many candidate locations under such strict conditions of cloud amount less than level 6. This is because the average cloud amount in Japan is 6.91, which exceeds level 6. From this point of view, it is clear that we need to search for more locations with the level more than level 6, but we also need to consider about the operation difficulty under a high possibility of cloud blockage to a satellite path over an optical ground station when the cloud amount is more than level 6. Figure 19.4 shows the 47 candidate locations with 10 years averaged cloud amount of level 6.5 or less. This value 6.5 or less means 10 years averaged cloud amount and therefore it includes decimal value as a calculation result, while the level 6.5 as the number of cloud amount level classified by JMA does not exist, but level 6. From this analysis, it is also understood that most locations with this cloud amount (level 6.5 as 10 years averaged value) are populated mainly at coast sides of the Pacific Ocean and isolated islands because it is generally known the Pacific Ocean side tends to be sunny. Though the number of candidate locations with the condition of 10 years averaged cloud amount of level 6.5 or less is 47 locations, we also need to confirm the number of candidate locations with the condition of cloud amount level 7 which is more than this calculated averaged level 6.5 to understand if the location number will be same or increase for both comparisons. The 10 years averaged cloud amount level 7 means only 30% of the whole sky is not cloudy and the optical direct communication needs to be done under such cloudy conditions. It is possible to consider cases where the 30% cloud free space in a given whole sky is contiguous or spread over a given satellite arc viewed from a given optical ground station.

Cloud blockage in satellite-ground optical communications

261

Figure 19.4 Forty-seven candidate locations with 10 years averaged cloud amount 6.5 or less over Japan. Process and use a map of Japan, from Geospatial Information Authority of Japan, MLIT, Government of Japan As a result of the analysis, candidate locations with 10 years averaged cloud amount of level 7 or less are found in 94 locations in Japan. These candidate locations with this 10-year averaged cloud amount of level 7 are also mainly at the coast side of the Pacific Ocean and isolated islands and show almost the same result

262

Advances in communications satellite systems 2: ICSSC-2019

of the 10 years averaged cloud amount of level 6.5 or less. From this result, it is figured out that it is necessary to find out more suitable locations from those 47 locations populated along the coast side of the Pacific Ocean under the condition of 10 years averaged cloud amount 6.5 or less because of their locations along the Pacific Ocean and the level 7 meaning operation difficulty compared with level 6.5 or less. Furthermore, it is also important to know how many percentages of incidences during selected 10 years period such a cloud amount of level 6.5 or less will exist at 47 candidate locations. These 47 candidate locations have also a cloud amount value of leveld 0–10 because the cloud amount of level 6.5 or less is the 10 years averaged value. From this point of view, we also investigated the percentages of incidences of cloud levels 0–10 during the selected 10 years period at these 47 candidate locations. As mentioned before, the cloud amount level has 11 levels from 0 to 10, and there is no decimal point like 6.5. Because of this, we analyzed the percentage of incidences with a cloud amount of level 6 or less below the averaged value 6.5 and figured out the percentage of incidences with a cloud amount of level 6 or less at these 47 locations varies from 37% to 45%. Therefore, we selected 35 of 47 candidate locations with 40% or more as the percentages of incidences of level 6 or less during the selected 10 years period.

19.2.3 Domestic candidate locations further analyzed by the installation conditions In Section 19.2.2, the analysis is based on 10 years averaged cloud amount and the percentage of incidences during the selected 10-year period over Japan. It revealed that the number of candidate locations with cloud amount level 6 is only three locations and the number with cloud amount level 6 or less is 47 locations and the percentage of incidences with such cloud amount levels is 37% to 45%. The selected 35 of 47 candidate locations had 40% or more incidence percentages of level 6 or less during the selected 10-year period. To select more suitable candidate locations from these 35 candidate locations, we need to consider not only the cloud amount condition, but also installation conditions, as described in a paper by the National Astronomical Observatory of Japan (NAOJ), which operates the same type of optical ground system. In their paper, many other conditions which are wind speed, visibility, and damage are shown as necessary items to be taken into account for the operation of the optical ground system [5]. Table 19.1 shows the installation conditions for optical ground systems For the space optical direct communications, the installation conditions are almost the same as that for the astronomical observatory because both are optical ground system, but there are some conditions to be relaxed in terms of space optical direct communications. For instance, there is a point of view of data downlink from a given satellite data recorder based on a given operation plan under a required BER quality for the space optical direct communications. It means a short time and high-speed data downlink during a given whole operation time is acceptable if the planned data can be downloaded from a given satellite during a given operation

Cloud blockage in satellite-ground optical communications

263

Table 19.1 Installation condition for optical ground systems Installation condition

No.

Item 1 2 3 4 5 6 7 8 9 10

Reason

Cloud amount Space blockage Wind speed Atmospheric seeing Yellow sand Visibility Volcanic ash Visibility Humidity Damage Salinity Damage Sulfuric acid Damage gas Freezing Damage Condensation Damage Snow amount Accessibility

Astronomical observatory for scientific observation

Optical ground station for satellite communication

Requirement Remarks Requirement Remarks Small Weak

Small Weak

Small Small Low Small Small

Small Small Low Small Small

Small Small Small

Note

Note Note

Small Small Small

Note

Note Note

Note: Possible to be measured by temperature and humidity.

time or those data can be distributed partially at separate optical ground stations and finally, the downlinked data can be transported properly to each user. In astronomical observation, the continuous observation is necessary to acquire the required image of a given target. This is the different point of view in both operations. As further analysis by installation conditions of visibility, we checked the influence of yellow sand and volcanic ash on visibility. It is known that the yellow sands problem occurs from March to June over Japan starting from the south and spreading over the entire west and reaching the north of Japan and volcanic ash occurs at active volcanos monitored by JMA and the ash like the yellow sand will be problematic to the visibility and also will do some damage to optical ground systems. We excluded two from 35 candidate locations, which is within 50 km from the 13 active volcanos classified as lank A [6] and also excluded eight more candidate locations because of average wind speeds exceeding 4 m/s. Finally, 25 candidate locations were selected. Table 19.2 shows the 25 candidate locations. In this analysis, it is also possible to take into account for all installation conditions but we decided to confirm first their correlation of these 25 candidate locations in terms of cloud amount because this number of candidate locations is enough small, and we thought it better to reconsider about the remaining installation conditions after correlation analysis.

19.2.4 Domestic candidate locations further analyzed by their uncorrelation relations To achieve site diversity across multiple optical ground stations, the cloud amounts at the ground stations need to be uncorrelated. If optical ground stations are arranged in a narrow distance, the correlation value from 0 to 1 is close to value 1 in

264

Advances in communications satellite systems 2: ICSSC-2019

Table 19.2 Twenty-five candidate locations Cloud amount No.

Location

6 or less

5 or less

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

Obihiro Onahama Sutra Maebashi Kumagaya Gifu Nagoya lida Kofu Talmo Hamamatsu Shizuoka Owashi Taleyama Kobe Wakayama lzuhara Oita Nobeoka Miyazaki Tadolsu Takamatsu Kochi Tokushima Shimizu

11.04% 40.22% 43.12% 41.81% 40.69% 42.74% 41.59% 43.75% 40.74% 41.33% 44.31% 46.78% 41.11% 46.31% 40.75% 40.76% 40.41% 43.72% 45.26% 43.49% 40.63% 42.07% 45.40% 42.72% 44.84%

38.32% 31.27% 39.54% 39.04% 37.79% 38.67% 38.21% 39.89% 38.25% 37.25% 41.47% 37.62% 37.82% 37.03% 37.63% 36.92% 36.90% 39.93% 42.19% 46.72% 37.61% 38.86% 42.14% 38.63% 42.11%

18.75% 19.10% 16.45% 19.39% 20.86% 15.54% 17.69% 19.62% 22.29% 17.94% 19.59% 18.47% 17.80% 17.53% 19.17% 14.62% 17.81% 17.14% 21.79% 23.18% 16.38% 16.75% 20.37% 16.24% 21.72%

the cloud amount, and it occurs in a case that neighboring optical ground stations are covered with some clouds. This is an important point of view to select uncorrelated candidate locations in terms of cloud amount. Figure 19.5 shows three different groups (A, B, and C) of selected 25 candidate locations. We selected more suitable candidate locations from these groups, in terms of their uncorrelation relation. Use the following formula (19.2) to find the correlation: 1 Xn ðx  x Þðyi  y Þ sxy i¼1 i n qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi r¼ Pn Pn sx sy 2 1 1  ðx  x Þ y Þ2 i i¼1 i¼1 ðyi   n n

(19.2)

In this formula, x and y are the respective candidate locations, sxy is the covariance of x and y, sx is the standard deviation of x, sy is the standard deviation of y, n is the total number of data, xi and yi are the respective cloud amount data, and x and  y are average values. From these 25 candidate locations, we performed their correlation analyses in terms of cloud amount and figured out six uncorrelated candidate locations because

Cloud blockage in satellite-ground optical communications

265

Figure 19.5 A group of 25 candidate locations for uncorrelated cloud cover. Process and use a map of Japan, from Geospatial Information Authority of Japan, MLIT, Government of Japan their correlation values are smaller than those of the other 19 candidate locations. Finally, we created six combinations of these six uncorrelated candidate locations. For the selection of these six uncorrelated candidate locations, we selected Obihiro from area A, Onahama or Tateyama from area B, and Kochi or Nobeoka or

266

Advances in communications satellite systems 2: ICSSC-2019

Miyazaki from area C. In fact, there are other selectable candidate locations (Hamamatsu, Iida, and Suwa) with low cloud amount but we did not select them because of their high correlation values with other candidate locations in area C. Katsuura and Mito in area B are also selectable as candidate locations by relaxing the percentage of incidences of cloud amount level 6 or less from 40% to a bit lower value and Izuhara and Shimiz in area C are good for cloud conditions, but we did not select Izuhara because it is a distant and inconvenient island and Shimizu clearly faces windy seaside. Table 19.3 shows the six combinations of six uncorrelated candidate locations. The reason why three candidate locations are required for the site diversity of the optical ground stations is the percentage of incidences of cloud amount level 6 or less during the selected 10 years at each candidate location is about 45% of 10 years period. Therefore, we selected three candidate locations for each combination to realize a high system availability as a target. From Table 19.3, it was figured out that these selected six candidate locations forming six combinations have low correlation values. Finally, we analyzed how many percentages of incidences of cloud amount 6 or less at these six combinations exist during the selected 10 years period and Table 19.4 shows this result. As a result of the analysis, the combination 6 has the highest value of 74.74% as the percentage of incidences of cloud amount 6 or less during the 10-year period. If the cloud amount level 5 or less which is lower than level 6 or less is required, combination 3 has the highest value of 71.20% as the percentage of incidences of cloud amount 5 or less during the 10-year period. It depends on which cloud Table 19.3 Six combinations of six uncorrelated candidate locations Location Combination 1 Combination 2 Combination 3 Combination 4 Combination 5 Combination 6

Obihiro Onahama Kochi Obihiro Onahama Nobeoka Obihiro Onahama Miyazaki Obihiro Tateyama Kochi Obihiro To:cyanic Nobeoka Obihiro Tateyama Miyazaki

Onahama Kochi Obihiro Onahama Nobeoka Obihiro Onahama Miyazaki Obihiro Taleyama Kochi Obihiro Taleyama Nobeoka Obihiro Tateyama Miyazaki Obihiro

Approx. distance (km)

Correlation

670 760 1,300 670 960 1,520 670 1,000 1,570 920 600 1,300 920 800 1,520 920 800 1,570

0.25 0.35 0.13 0.25 0.29 0.07 0.25 0.26 0.06 0.13 0.39 0.13 0.13 0.34 0.07 0.13 0.32 0.06

Cloud blockage in satellite-ground optical communications

267

Table 19.4 Percentages of incidences of cloud amount 6 or less for each combination Cloud amount

Data: 1988–1997

Combination Combination Combination Combination Combination Combination

1 2 3 4 5 6

6 or less

6 or less

0

73.06% 74.46% 74.31% 73.73% 74.64% 74.74%

69.76% 70.35% 71.20% 70.11% 70.96% 71.17%

42.83% 41.83% 46.47% 41.29% 40.75% 44.81%

amount level is required and what percentage of incidences with selected cloud amount level is required. It means it reaches around 70% to 75% of incidences with selected cloud amount level 6 or less and 5 or less. In another word, around 70% to 75% of 10-year period satisfies the cloud amount level (6 and 5 or less). Simply, system availability only in terms of cloud blockage would be around 70% to 75% with three selected and uncorrelated candidate locations. To realize more than these values, we need to find more distant locations but we also need to relax the cloud amount level from 6 to 7 increasing the operation difficulty.

19.3 Results In this study, we investigated first the type of available cloud data for both ground and space and analyzed 10 years averaged cloud amount of 155 locations over Japan based on the long-term ground meteorological data provided by JMA and showed three different results with the following cloud amount conditions: 1. 2. 3.

Three locations for 10 years averaged cloud amount 6 or less Forty-seven locations for 10 years averaged cloud amount 6.5 or less Ninety-four locations for 10 years averaged cloud amount 7 or less

From the above result, we selected 47 locations for 10 years averaged cloud amount 6 or less and 25 locations additionally filtered by the percentage of incidences with selected cloud amount level 6 or less for 10 years period and installation conditions. Finally, we selected six candidate locations because of small correlation values and created six combinations of six candidate locations and also calculated expected system availability from around 70% to 75% by three separate candidate locations. From this result, it was found difficult to achieve 100% system availability at three separate candidate locations in Japan with averaged cloud amount 6 or less. Since yellow sand will spread over especially south to west of Japan from March to June and the vast cloud of typhoon will come and spread over especially south to west and east of Japan from August to October. From this point of view, it would be

268

Advances in communications satellite systems 2: ICSSC-2019

another option to place one in Hokkaido and another optical ground station in Australia in the Asia Pacific region to form an international optical ground station network with existing optical ground stations of NASA and ESA. The candidate locations shown in this chapter are ground-based meteorological data limited for lower cloud amount. Therefore, we need to analyze the satellitebased data at the same locations and for the same periods and also detailed field surveys referring to installation conditions will be necessary before installing optical ground stations.

19.4 Conclusion In this study, we showed six combinations forming six uncorrelated candidate locations in Japan for the site diversity of optical ground stations to mitigate cloud blockage in space optical direct communications based on long-term ground-based cloud amount data corresponding to only lower cloud amount and also the expected system availability (around 70% to 75%) to be realized by these combinations. This time, we analyzed only the ground-based long-term cloud amount over Japan related to lower cloud and we believe we need to analyze and check the satellite data for higher cloud during the same period for the same locations in future.

Acknowledgments This research has been conducted under the joint research agreement between JAXA and Tokai University.

References [1] Mukai, T., Takayama, Y., and Araki, T. “Research and development approach to realize flexible optical ground network operations for effective data downlink from space to ground.” 36th International Communications Satellite Systems Conference, Niagara Falls, Canada. 2018. pp. 15–18. [2] IOAG.T.OLSG.2012.V1. Interagency Operations Advisory Group. Optical Link Study Group. Final Report. June 5, 2012. [3] Takayama, Y., Toyoshima, M., and Kura, N. “Estimation of accessible probability in a low Earth orbit satellite to ground laser communications.” Radioengineering. 2010; 19(2): 249–253. [4] Ninomiya, H., and Takayama, Y. “Diversity effects in satellite-ground laser communications using satellite images.” 29th AIAA International Communications Satellite Systems Conference. Nara, Japan, November 28–December 1, 2011. [5] Renichi, S. “Climatological selection of an ideal place for astronomical observation.” Journal of Meteorological Research. 1955; 7(1). [6] Hayashi, Y., and Uhira, K. “The classification of active volcanoes in Japan by way of ‘volcanic activity indexes’ based on the eruption histories for the past 10,000 years.” Quarterly Journal of Seismology. 2008; 71(1–4): 59–78.

Chapter 20

Overview of optical ground systems developments for network switching controls to avoid cloud blockage in space optical direct communications Tatsuya Mukai1, Yoshihisa Takayama2 and Tomohiro Araki1

For high-speed space communications, future needs of space optical communication technologies are further increasing to provide operation services of intersatellite links and direct links in the field of remote sensing and space explorations. However, cloud blockage of direct communication links is one of the issues in space optical direct communications, while the atmospheric effect on the communication links is also an important issue on the link quality in a given satellite path connected with optical ground stations. Space optical communication experiments have been performed and results have also been reported internationally, but many of them are mainly focused on communication link evaluations for stable optical direct links under atmosphere, while experiments of optical ground systems for avoidance of cloud blockage with ground network switching controls have not been sufficiently reported. For the network switching controls among optical ground stations based on cloud blockage on a given satellite path over stations, JAXA developed ground network systems. First, the laser ground network planning system is explained. Next, our optical ground stations and the infrared cloud monitoring and discrimination system for lower cloud blockage are explained. Finally, our planned network switching testings are explained. Key Words: optical direct communication; cloud blockage; optical ground station; network switching control

1

Japan Aerospace Exploration Agency, Sengen, Tsukuba, Ibaraki, Japan Department of Communication and Network Engineering, Tokai University, Takanawa, Minato-ku, Tokyo, Japan 2

270

Advances in communications satellite systems 2: ICSSC-2019

20.1 Introduction As a ground-based measurement equipment, some of the cloud monitoring systems are introduced in the Consultative Committee for Space Data Systems (CCSDS) book [1] with other meteorological measurement equipment attached to optical ground stations and the study on the global site diversity due to cloud blockage between space and ground are also reported by Optical Link Study Group (OLSG) in Interagency Operation’s Advisory Group (IOAG) [2]. As one of the methods for selection of one of optical ground stations from cloud coverage analysis using meteorological satellites, a ground analysis system for the cloud coverage analysis over optical ground stations was used for optical direct communication links with a moon explorer called Lunar Atmosphere and Dust Environment Explorer (LADEE) [3]. In the 36th International Communications Satellite Systems Conference (ICSSC) of Advances in Communications Satellite Systems (AIAA), 2018, Niagara Falls, Canada, JAXA reported on research and development approach to realize flexible optical ground network operations for effective data downlink from space to ground and the ground network system with functions of cloud monitoring and discrimination for lower and higher cloud blockage and network switching controls based on cloud discrimination on a satellite path was written [4]. The ground network system is composed of a ground-based infrared cloud monitoring and discrimination system for lower cloud blockage, a satellite-based infrared cloud monitoring and discrimination system for higher cloud blockage and a laser ground network planning system for the network switching controls among optical ground stations. The ground-based infrared cloud monitoring and discrimination system for the lower cloud blockage will be placed at each local site with each optical ground station and creates specific messages of the percentage of cloud amount on a given satellite orbit path viewed from optical ground stations, while the satellite-based infrared cloud monitoring and discrimination system for the higher cloud blockage will be placed near a meteorological data center and also creates specific messages of the percentage of cloud amount on a given satellite orbit path viewed from a meteorological GEO satellite. These messages will be transmitted via ground fiber networks to the laser ground network planning system for the network switching controls among optical ground stations. Based on a joint research between JAXA and Tokai University, the laser ground network planning system for the network switching controls among optical ground stations is placed at Takanawa campus of Tokai University, Tokyo and two ground-based infrared cloud monitoring and discrimination systems for the lower cloud blockage will be placed at two spatially separate locations such as Kumamoto (Kyushu) and Sapporo (Hokkaido) campuses of Tokai University. First, the basic network switching control experiment focused on lower cloud blockage will be performed under the joint research and the whole ground network system experiment with a further complementary function for monitoring and discrimination of higher cloud blockage will be additionally performed for the real network switching operations in future.

Optical ground systems for network switching controls

271

In this chapter, the development overview of optical ground systems, such as the laser ground network planning system, the infrared cloud monitoring and discrimination system, optical ground stations for network switching controls to avoid cloud blockage, and the network switching testings for space optical direct communications are analyzed.

20.2 Methodology 20.2.1 Laser ground network planning system As a ground network system, the optical ground systems for network switching controls to avoid cloud blockage between space and ground are divided into a laser ground network planning system for the network switching controls for optical ground stations, a ground-based infrared cloud monitoring and discrimination system for the lower cloud blockage, and a satellite-based infrared cloud monitoring and discrimination system for the higher cloud blockage. In this section, the laser ground network planning system, optical ground stations, the infrared cloud monitoring and discrimination system for the lower cloud blockage, and the network switching testings are explained with future development scenarios of the satellite-based infrared cloud monitoring and discrimination system for the higher cloud blockage. Figure 20.1 shows how the laser ground network planning system works in the whole optical ground systems with each system. The system development except Indoor

Outdoor C&M and planning file (Note 4) Cloud blockage (%)

HDMI cable Extended monitor (Note l)

Satellite orbit file

Client PC Remote access (Note 2) LAN cable

Planning file (Note 4) C&M and planning file (Note 4) Server PC Cloud blockage (%) (Note 3) Planning file (Note 4)

USB cable

System for higher cloud blockage

C&M and planning file (Note 4) Cloud blockage (%)

Cloud blockage (%) Planning file (Note 4)

Planning file (Note 4)

C&M system to given OGS System for lower cloud blockage C&M system to given OGS System for lower cloud blockage C&M system to given OGS System for lower cloud blockage

For Asia pacific region Note 1: Simulation monitor of cloud blockage in lower and higher cloud layers and C&M systems for optical ground stations (OGSs) Note 2: Remote PC for the server PC (Laser ground network planning system) controlling the OGS network. Note 3: Server PC (Laser ground network planning system) with simulation functions of cloud blockage and C&M system. Note 4: Planning files shared with C&M systems and systems for lower and higher cloud blockage for OGSs.

Figure 20.1 Laser ground network planning system working in the whole optical ground systems

272

Advances in communications satellite systems 2: ICSSC-2019

for parts shown with dotted lines is almost complete and a step-by-step network test is planned for the whole system development. Basically, the number of the cloud monitoring and discrimination system for the lower cloud blockage will be attached up to three optical ground stations and the cloud monitoring and discrimination system for the higher cloud blockage using imageries from a meteorological GEO satellite over Japan will be also attached in the future. Currently, one of the simulators built in the server PC for the laser ground network planning system works as the cloud monitoring and discrimination system for the higher cloud blockage. Cloud blockage simulation is realized in such a manner that the satellite orbit path is divided into three sections (A, B, and C) and the increase and decrease of cloud amount according to a ruled data format and control and monitoring (C&M) systems connected to each optical ground station is also simulated. In this simulation, the data acquired through real cloud observation can be also used as the simulation data by a manual process. Figure 20.2 shows the laser ground network planning system installed at Takanawa campus of Tokai University, Tokyo in cooperation with JAXA under our joint research contract. This system works for controlling and monitoring of the network switching among optical ground stations. First, the satellite orbit file is taken into this system and the planning file for the satellite is created and set it in not only this system but also in other ground systems for the lower and higher cloud blockage and for the C&M (Control and Monitor) of optical ground stations. Figure 20.2 shows the laser ground network planning system installed at Takanawa campus of Tokai University, Tokyo in cooperation with JAXA under our joint research contract.

Figure 20.2 Laser ground network planning system

Optical ground systems for network switching controls

273

The upper part of the monitor screen is for the management of operation time and planning. The planning management is composed of visible and assigned plans. Furthermore, there are four more managements of pre-pass, post-pass, blocked pass, and required pass. The pre-pass is for ground system pre-setting and the post-pass is required for post process of optical ground stations. The blocked pass means the assigned pass blocked by cloud overlaid on a given satellite orbit path when the percentage of cloud blockage reaches an unacceptable level set previously in the system. The middle part of the monitor screen (half of the left side) is for the management of cloud blockage at optical ground stations. This part is divided into two types of managements which are for past and current cloud amount shown by line graphs to encompass the change of tendency of cloud amount at locations and for the monitoring the cloud mount in three different sections (A, B, and C) on a given satellite orbit path. This system takes in the cloud amount in percentage from the ground-based infrared cloud monitoring and discrimination systems for the lower cloud blockage and the satellite-based infrared cloud monitoring and discrimination system for the higher cloud blockage in a given satellite path between space and ground. In this laser ground network planning system, the information of the percentage of cloud blockage in the lower and higher cloud layers is taken in and used for control and monitoring to switch optical ground stations. The middle part of the monitor screen (half of the right side) is for the management of operation performances of optical ground stations and the bottom part works for logging network status. This laser ground network planning system can work with groundbased infrared and satellite-based cloud monitoring and discrimination systems for the lower and the higher cloud blockage and C&M systems for optical ground stations and also with whole simulators built-in itself. Figure 20.3 shows the extended system attached to the laser ground network planning system for control and monitor in case of using simulators. This extended system is not used normally, once whole real ground systems are completely set at given locations and the system starts its operations. Currently, this extended system is utilized partially for simulations because the satellite-based cloud monitoring and discrimination system for the higher cloud blockage and the interface to the C&M systems for optical ground stations need to be developed. However, the basic parts of C&M system of optical ground stations work already for two optical ground stations whose diameters are 60 cm (fixed OGS) and 30 cm (transportable OGS) at two locations (Nagano and Hokkaido) in Japan.

20.2.2 Optical ground stations Figures 20.4 and 20.5 show two optical ground stations which are a fixed 60 cm OGS in Nagano and a transportable 30 cm OGS in Hokkaido, Japan. These developments were reported before in Japan [5]. Each site has already an outdoor space to install the ground-based infrared cloud monitoring and discrimination systems for the lower cloud blockage and also their C&M systems will be connected with the laser ground network planning system, once the technical interface will be built in.

274

Advances in communications satellite systems 2: ICSSC-2019 Indoor

Outdoor C&M and planning file (Note4) Cloud blockage (%)

Satellite orbit file

HDMI cable

Extended monitor (Note l)

Client PC (Note 2)

Remote access LAN cable

Server PC (Note 3)

Planning file (Note 4) C&M and planning file (Note 4) Cloud blockage (%) Planning file (Note 4) C&M and planning file (Note 4)

USB cable

Cloud blockage (%) System for higher cloud blockage

Cloud blockage (%) Planning file (Note 4)

Planning file (Note 4)

C&M system to given OGS System for lower cloud blockage C&M system to given OGS System for lower cloud blockage C&M system to given OGS System for lower cloud blockage

For Asia pacific region Note 1: Simulation monitor of cloud blockage in lower and higher cloud layers and C&M systems for OGSs. Note 2: Remote PC for the server PC (laser ground network planning system) controlling the OGS network. Note 3: Server PC (laser ground network planning system) with simulation functions of cloud blockage and C&M system. Note 4: Planning files shared with C&M systems and systems for lower and higher cloud blockage for OGSs.

Figure 20.3 Extended system of the laser ground network planning system when the simulator function is used

Figure 20.4 Fixed 60 cm OGS in Nagano

These two optical ground stations basically have their open-loop tracking functions and last year we developed the closed-loop function based on visible and infrared tracking cameras. First, we tested it with the transportable OGS (TOGS) in Hokkaido, Japan and then it will be also installed in the other OGS in Nagano, Japan. These optical ground stations can deal with visible laser (e.g., 532 nm) and infrared laser (e.g., 1550 nm) in uplink and downlink. Tables 20.1 and 20.2 show each OGS characteristics.

Optical ground systems for network switching controls

275

Figure 20.5 Transportable 30 cm OGS in Hokkaido Table 20.1 Characteristics of fixed 60 cm OGS No.

Items

Characteristics

1 2

Diameter Optical interface

60 cm Cassegrain (Note 1); Direct focus (Note 2); Nasmyth; Coude (Note 3)

3

Pointing and tracking

55 dBW, >7 dB/K Regenerative communication mode Uplink: 1.5, 6, 24, 51, 155 Mbps Downlink: 155 Mbps Bent-pipe communication mode up to 3.2 Gbps

494

Advances in communications satellite systems 2: ICSSC-2019 MBA APAA

Cover area of MBA

Cover area of APAA

Figure 38.1 Schematic of Wideband InterNetworking engineering test and Demonstration Satellite (WINDS) coverage areas

Figure 38.2 Wideband InterNetworking engineering test and Demonstration Satellite (WINDS) vehicle earth station a three-axis gimbal system, a modulator, and a demodulator. The earth station was mounted on a van, as presented in Figure 38.2. The antenna system mounted on the vehicle earth station was capable of automatically acquiring and tracking WINDS, which provided the exact position of the vehicle earth station (determined via GPS signal) and the beacon signal level (global beam from the satellite). These features allowed the vehicle earth station to continuously link with the satellite even while moving. Here, we define the beacon signal as residual carriers of the network monitoring information. Other features,

Seasonal vegetation changes and Ka-band satellite communication

495

Table 38.2. Wideband InterNetworking engineering test and Demonstration Satellite (WINDS) vehicle earth station specifications Item

Value

Tx frequency Rx frequency

27.5–28.6 GHz 17.7–18.8 GHz 18.9 GHz (for receiving beacon signal) Linearly polarized wave (V/H) Cassegrain antenna diameter: 65 cm El: 20
–90
Az: endless rotation X-El: 15
Less than 0.2
Ethernet (1,000 base-T) ≧2.8 kVA

Polarization Antenna Antenna driving range Tracking accuracy User interface Power generation capacity

such as high-definition cameras and wireless local area network (WLAN) access points, were also installed in the vehicle earth station for more effective emergency/ disaster communications.

38.3 Measurement experiments Measurements of the data transmission characteristics were performed at the same time as propagation measurements. Signal measurements were obtained in an environment wherein trees were located between the satellite and the vehicle earth station, and radio waves were shielded. This allowed us to investigate the impact of tree shadowing on the received signal.

38.3.1 Experimental site The measurement experiments were conducted in Sendai, Japan. Because the direction of the satellite was south, the trees on the south side of the street were the objects of measurement. The dimensions of each tree were measured monthly for approximately 1 year to observe how each tree changed with the seasons. The measurement duration ranged from early April, when there were no leaves on the deciduous trees, to February, when the deciduous trees were leafless again. In addition, measurements were performed every 2 weeks in April and November when there were noticeable changes in leaf cover on the deciduous trees. Two deciduous trees (cherry, Trident maple) and an evergreen tree (pine) were studied to determine the differences in tree type and the seasonal changes between deciduous and evergreen trees.

38.3.2 Measurement system and experimental method The measurement system in this experiment is presented in Figure 38.3. The experiment to measure the transmission characteristics was conducted using a

496

Advances in communications satellite systems 2: ICSSC-2019

Earth station #1 (move)

Earth station #2 (fix)

Figure 38.3 Network configuration of data transmission experiment Table 38.3. Measurement specifications Item

Value

Tx frequency Rx frequency Bandwidth Data rate modulation Time resolution Vehicle speed

28.2 GHz (vehicle earth station #1) 18.4 GHz (vehicle earth station #2) 20 MHz 18.8 Mbps QPSK (R ¼ 1/2) 10 ms 10 km/h

DVB-S2 modem in the WINDS bent-pipe communication mode. We performed evaluations using UDP communication with Iperf [8] between small in-vehicle station #1 (tree side: Miyagi Prefecture, MBA Tohoku Beam) and small in-vehicle station #2 (NICT Kashima Space Technology Center: Kashima City, MBA Kanto Beam) via satellite. Vehicle earth station #1 near the tree side transmitted a signal with a center frequency of 28.2 GHz and bandwidth of 20 MHz while moving at 10 km/h. Vehicle earth station #2 then received the signal in a stationary state. Data were measured at a fixed distance based on the travel speed using vehicle speed pulses, with a minimum distance resolution of 2 cm at a speed of 10 km/h. The measurement specifications are listed in Table 38.3. An example of shadowing scenario is presented in Figure 38.4, which shows the positional relationship between a tree and the vehicle earth station at the measurement location. The trees were located between the satellite and the vehicle earth station antenna, causing the received radio waves to be shadowed.

38.4 Results and discussion To perform a comparison with previous propagation measurement results, we provide the measurement results of the cherry tree (deciduous tree) and pine tree (deciduous tree) reported in a previous study [7]. Photographs of the seasonal

Seasonal vegetation changes and Ka-band satellite communication

497

Direction of satellite

Figure 38.4 Example of shadowing scenario

Branch spread Tree crown Tree height

Tree base April

Mid-April

May

September

October

November

Figure 38.5 Seasonal changes in the cherry tree

changes in pine and cherry leaf cover are presented in Figures 38.5 and 38.6, whereas the branch spread of the pine and cherry trees during each measurement day is listed in Table 38.4. The results of the propagation of the cherry and pine trees from April to February are provided in Figures 38.7 and 38.8. Shadowing intervals, which are defined as tree intervals in which the received signal is

498

Advances in communications satellite systems 2: ICSSC-2019

Figure 38.6 Seasonal changes in the pine tree

Table 38.4. Branch spread of cherry and pine trees Branch spread (m) Month

Cherry tree

Pine tree

Apr Mid-Apr May Jun Jul Aug Sep Oct Nov Mid-Nov Dec Jan Feb

4.6 4.6 5.0 5.1 5.1 5.1 5.0 5.0 4.3 4.3 4.2 4.2 3.6

3.9 3.5 4.0 3.9 3.9 3.8 3.9 4.0 4.0 3.9 3.9 4.0 3.9

attenuated by 3 and 10 dB, were determined to detect any attenuation fluctuations due to seasonal variations in deciduous and evergreen trees. The results are listed in Table 38.5, and the relationship between the month and the shadowing interval is presented in Figure 38.9. The cherry tree, a deciduous tree, has a shadowing interval with 10 dB attenuation between mid-April and October, with an approximately constant shadowing interval from May to September when the leaf conditions are almost the same. The cherry tree has no shadowing interval with 10 dB attenuation in April and in November to February, which corresponds to the months in which the tree does not have full leaf cover. However, the pine tree, an evergreen tree, has an approximately constant shadowing interval across the entire study period [7].

Seasonal vegetation changes and Ka-band satellite communication

499

Received between level (dB)

0

–5 Apr Mid-Apr May Jun Jul Aug Sep Oct Nov Mid-Nov Dec Jan Feb

–10

–15

–20

–25 0

1

2

3

4

5

6

7

8

Distance (m)

Figure 38.7 Normalized beacon received power for the deciduous cherry tree

Received beacon level (dB)

0

–5 Apr Mid-Apr May Jun Jul Aug Sep Oct Nov Mid-Nov Dec Jan Feb

–10

–15

–20

–25 0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Distance (m)

Figure 38.8 Normalized beacon received power for the evergreen pine tree The results in the measurement of the data transmission characteristics were based on the amount of packet loss that occurred during the measurement period recorded at the receiver. Because we used UDP communication, if communication was interrupted due to the shadowing of a tree, packet loss was observed on the receiving side, and no packet loss occurred when the propagation path was cleared. Figure 38.10 and Table 38.6 present the seasonal changes of packet loss normalized by the maximum value. The pine tree data for August were excluded because they could not be measured due to communication equipment failure.

500

Advances in communications satellite systems 2: ICSSC-2019

Table 38.5. Shadowing intervals for cherry and pine trees Tree

Cherry tree (m) 3 dB 3.42 3.86 4.36 4.28 4.80 4.44 5.06 4.10 3.12 2.86 3.32 3.52 2.42

Attenuation Apr Mid-Apr May Jun Jul Aug Sep Oct Nov Mid-Nov Dec Jan Feb

Cherry 3 dB

Pine tree (m)

10 dB 0.00 2.94 3.54 3.60 4.00 3.62 4.04 3.20 0.00 0.00 0.00 0.00 0.00

Cherry 10 dB

3 dB 1.66 1.66 1.94 2.30 2.30 2.92 2.76 2.66 2.16 2.34 2.54 2.66 2.58

Pine 3 dB

10 dB 1.26 1.26 1.48 1.78 1.76 2.30 2.16 2.06 1.66 1.78 1.96 1.90 1.96

Pine 10 dB

Shadowing interval (m)

6

5

4

3

2

1

0 APR

MIDAPR

MAY

JUN

JUL

AUG

SEP

OCT

NOV

MIDNOV

DEC

JAN

FEB

Measurement month

Figure 38.9 Seasonal changes in shadowing interval Figure 38.10 suggests that deciduous cherry tree shadowing peaked in August and that more than 67% of the packet loss occurred from May to October. In addition, the occurrence of packet loss in April and after November when the leaves fell was less than 36% of the peak packet loss. For the evergreen pine trees, packet loss increased from April to July and was constant from September onward. This variation exhibited the same trend as the seasonal change in the shadowing interval presented in Figure 38.9 and Table 38.5. For the pine trees, the observed differences in the size of the shadowing section may have been due to the stop north–south control of WINDS. Fluctuations in the elevation angle at each

Seasonal vegetation changes and Ka-band satellite communication

501

Pine

Cherry 1.00 Normalized packet loss

0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10

Month

Fe b

Ja n

-N M

id

D ec

ov

ov N

ct O

Se p

ug A

l Ju

n Ju

M ay

A

pr

0.00

Figure 38.10 Seasonal changes in normalized packet loss Table 38.6. Normalized packet loss for cherry and pine trees Normalized packet loss Month

Cherry tree

Pine tree

Apr May Jun Jul Aug Sep Oct Nov Mid-Nov Dec Jan Feb

0.32 0.74 0.80 0.69 1.00 0.98 0.67 0.36 0.32 0.33 0.35 0.19

0.39 0.51 0.63 0.59 – 0.85 0.82 0.90 0.84 0.81 0.86 1.00

measurement time may have induced fluctuations in the shadowing width. These results reveal the correlation between the attenuation interval due to shadowing and the occurrence of packet loss as the transmission characteristic for both evergreen and deciduous trees.

38.5 Conclusion In this study, we measured the Ka-band fluctuations in data transmission at the vehicle earth station from April to February to investigate the influence of seasonal

502

Advances in communications satellite systems 2: ICSSC-2019

variations in tree cover on mobile satellite communications. Our results revealed a correlation between the attenuation interval due to shadowing and the occurrence of packet loss for both evergreen and deciduous trees. We observed that data transmission was affected by the increase and decrease of leaf cover due to seasonal variations.

References [1] Special Issue on Wideband InterNetworking engineering test and Demonstration Satellite (WINDS). Journal of the National Institute of Information and Communications Technology. 2007; 54. [2] Special Issue on Terrestrial Communication Technology and Ultra-HighSpeed Satellite Communications Technology. Journal of NICT. 2017; 64: 159–165. [3] Akaishi, A., Takahashi, T., Okawa, M., et al. “Ka-band mobile earth station for WINDS.” 29th ISTS 2013. June 2013. [4] Kan, T., Takahashi, T., Kawasaki, K., et al. “A measurement of propagation in high mobility environments for Ka-band satellite communication.” IEICE Technical Report SAT. 2015; 115: 79–84 (in Japanese). [5] Kan, T., Takahashi, T., Kawasaki, K., et al. ‘Propagation measurement for mobile satellite communications using WINDS at Kyushu area.” IEICE General Conference. B-3-16.2016 (in Japanese). [6] Kan, T., Takahashi, T., Kawasaki, K., et al. “Propagation measurement for mobile satellite communications using WINDS at the Coast of Japan Sea of Western Japan.” IEICE Technical Report SAT. 2017; 116: 81–86 (in Japanese). [7] Kan, T., Jeong, B., Susukita, H., et al. “Experimental results of seasonal variation of shadowing by Ka-band mobile satellite communication.” 32nd ISTS 2019. 2019. [8] iPerf - the ultimate speed test tool for TCP, UDP and SCTP. iPerf.fr. 2019. Available from https://iperf.fr/ [Accessed September 19, 2019].

Chapter 39

Experimental study of external interference for LEO-based sensing (AIS) Daichi Hirahara1, Toshiyuki Nishibori1, Toshiyoshi Kimura1, Shuji Shimizu1, Junichiro Ishizawa1, Shinichi Sobue1, Satoko Miura1 and Shinichi Suzuki1

In 2014, the Japan Aerospace Exploration Agency (JAXA) launched a second space-based automatic identification system (AIS), SPace-based AIS Experiment 2nd (SPAISE2), mounted on DAICHI-2. Then in May 2015, SPAISE2 entered an extended operations phase and is now continuously observing the seas around Japan to evaluate external interference and signal collision. This study reports on the separation of AIS messages and the external interference of large spike noise. And this chapter cites the heatmap of external interference and AIS messages on a low Earth orbit. Key Words: space-based AIS; satellite AIS; radio frequency interference; RFI

39.1 Introduction In satellite communications and satellite remote sensing, external interference is a problem that degrades performance. This problem is particularly prominent in the vicinity where ground systems and human activities are concentrated along intricate coasts. For security, safety, and other global near-real-time service, the radio wave environment, such as external interference and signal collision, must be understood for efficient system operation. The identification of interference sources, their locations, and signal strength provide a delimitation of areas and are important for improving telecommunications and systems design. Automatic identification system (AIS) data are used for ship identification, location information, and collision avoidance and are increasingly being used in other novel applications. AIS data identify vessels of interest, send alerts regarding potential threats, and assist in search and rescue operations, as well as ocean 1

Japan Aerospace Exploration Agency, Tsukuba, Japan

504

Advances in communications satellite systems 2: ICSSC-2019

environmental monitoring [1]. The coverage of the terrestrial AIS is limited to coastal areas. Satellites can expand the service of AIS to a global range, although the communication between ships and a low Earth orbit (LEO) satellites becomes asynchronous. For detecting and tracking ships, global near-real-time service [2,3] needs other Earth observation such as the geolocation of emitters across a broad range of RF signals [4,5], high-performance observation [6,7], and advanced satellite communication systems.

39.1.1 SPace-based AIS experiment (SPAISE) In 2014, the Japan Aerospace Exploration Agency (JAXA) launched SPace-based AIS Experiment (SPAISE2) mounted on DAICHI-2, the Advanced Land Observing Satellite-2 (ALOS-2) [8]. Figure 39.1 shows the SPAISE2 antenna on DAICHI-2. In May 2015, SPAISE2 entered an extended operations phase and is now continuously observing the seas around Japan to evaluate external interference and signal collision [9]. This study reports on the separation of AIS messages and the external interference of spike noise for LEO-based sensing.

39.2 External interference The collision of AIS messages from thousands of ground cells results in a loss of all collided messages [10,11]. Moreover, satellites detect the interference sources that may impact the capture of VHF AIS signals [12,13]. There are both legal and illegal terrestrial sources that can create interference with a space-borne AIS. Legal interference sources operate in accordance with the international framework

(a)

DAICHI-2

(b) Position of antennas (c)

AIS antenna

Figure 39.1 SPAISE2 antenna on DAICHI-2

Experimental study of external interference for LEO-based sensing

505

created by the International Telecommunication Union (ITU). Figure 39.2 shows an example of the observation results of the satellite AIS. The green line depicts the experimental trajectory and magenta plots denote the detected AIS positions. The sampling data of these observations are shown with spike noise and collision AIS messages. Figure 39.2 shows an example of observation results on the descending path. The roll angle is tilted by 30 for the SAR observation of DAICHI-2 and the pattern of the AIS cross-dipole antenna is tilted. In this observation, the AIS signal cannot be detected due to the reception of strong external noise at high latitudes. And on the eastern side of Japan, detection is not possible even with reduced external noise due to signal collision. The timing for most signal collisions can be read from the Doppler shift of the spectrum (when the reddest area of the contour is centered).

Start

AIS antenna

30°

Experimental orbit End

Detected AIS positions

Wave form of received signals

Interference

Spectrum

Collision

Count of detected AIS messages Start

End Sampling point

Figure 39.2 Example of observation results: experimental trajectory and detected AIS positions are shown with the elapsed time in reception status

506

Advances in communications satellite systems 2: ICSSC-2019

39.2.1 Separating AIS messages and external interference The external interference of spike noise corresponds to broadband interference. This spike noise shows interference produced by radar systems. Spike noise refers to a few sampling points on the detecting signal. Therefore, frequency domain filters are not effective in such cases of interference due to the broadband frequency domain characteristics of the spike noise. Figure 39.3 is indicative of a diagram that separates AIS messages and external interference. Spike noise NRFI and AIS signal PAIS are separated from observing signal Pall with a single dimension median filtering and noise threshold TH. The classical approach to the removal of impulsive noise is the median filter [14]. The threshold is set much higher than the complex power of AIS collision. For this reason, weak noise cannot be detected. It is enough to capture strong external interference. The separated AIS signal and spike noise are evaluated for intensity distribution at each observation point in the satellite trajectory. This does not specify the source. For the separated AIS signals, sensitivity deviation is corrected by satellite attitude and AIS antenna gain. Assuming that the ship’s antenna is a dipole antenna, the reception power of AIS signals at the orbital observation point is allocated within the field of view (FoV) by the link budget. By adding the power allocated from all observation points and dividing by the number of observations, the surface distribution is calculated from the on-trajectory distribution. Due to limited observation resources, the area of the target is observed only once in 3 months to experiment with the sampling data. These results are the observation for 1 year (i.e., four times). It should be noted that continuing this research to satisfy operational reductions in the future will be difficult.

39.3 Experimental results Figure 39.4 shows the power distribution of external interference on the ascending (ASE) paths by SPAISE2. Figure 39.5 shows observation of the descending (DSE) paths. A large difference was confirmed between the ASE and DSE paths. This is because several paths are tracked by powerful ground radar in the DSE paths. The main external interference areas of large spike noise are around the Kamchatka Peninsula, Mongolia, Russia, the northern part of the Korean Peninsula, and Kazakhstan. Figure 39.6 shows the main external interference sources by synthesizing all the results, roughly organized into the orbital positions of more than 10 cases of powerful external noise. In fact, there are also cases of weak and frequent external interference. Thus, detecting weak spike noise for all the data of 1 year remains a challenge for future research. Figure 39.7 shows the power distribution of receiving AIS signals on the ASE paths by SPAISE2. Figure 39.8 shows observation of the descending DSE paths. The main collision sources of AIS messages are the skies above Okinawa, Kobe, Kitakyushu, Shanghai, Taiwan, Hong Kong, Brunei, the Spratly Islands, and Malacca Strait. As a seasonal variation, we confirmed the time when the South China Sea and the area southwest of Okinawa were strengthened. The position of

Experimental study of external interference for LEO-based sensing

507

Received signal Pall

1-D median filtering PAIS’

PRFI’ = Pall - PAIS’ PRFI’ PRFI’ >= TH PAIS = PAIS’ PRFI = PRFI’

PRFI’ < TH PAIS = Pall PRFI = zero

PAIS (Figs. 7, 8) PRFI (Figs. 4, 5, 6) Antenna pattern of SAT-AIS & Dipole antenna of ships Observation trajectory T(long, lat)

RFoV(x,y) = f{Gsat(atan(∆lat,∆long)), Gship} RFoV(x,y)

PAIS(x,y,z)

Pn(long, lat) =

RFoV(x,y,z)

T(long, lat)

P(long, lat) =

∑ n

Pn(long, lat) counts(S)

P(long, lat) (Figs. 9, 10, 11)

Figure 39.3 Block diagram for separating AIS messages and external interference

508

Advances in communications satellite systems 2: ICSSC-2019

(a) Jun. 4 to Aug. 26, 2018

(b) Aug. 27 to Nov. 18, 2018 (c) Nov. 19, 2018 to Feb. 10, 2019 (d)

Feb. 11 to May 5, 2019

Figure 39.4 The power distribution of external interference on ascending paths by SPAISE2

(a)

Jun. 4 to Aug. 26, 2018

(b)

Aug. 27 to Nov. 18, 2018

(c)

Nov. 19, 2018 to Feb. 10, 2019 (d) Feb. 11 to May 5, 2019

Figure 39.5 The power distribution of external interference on descending paths by SPAISE2

Figure 39.6 The orbital positions of more than10 cases of powerful external noise

Experimental study of external interference for LEO-based sensing

(a)

Jun. 4 to Aug. 26, 2018

(b)

Aug. 27 to Nov. 18, 2018

(c) Nov. 19, 2018 to Feb. 10, 2019 (d)

509

Feb. 11 to May 5, 2019

Figure 39.7 The power distribution of receiving AIS signals on ascending paths by SPAISE2

(a)

Jun. 4 to Aug. 26, 2018

(b)

Aug. 27 to Nov. 18, 2018

(c) Nov. 19, 2018 to Feb. 10, 2019 (d)

Feb. 11 to May 5, 2019

Figure 39.8 The power distribution of receiving AIS signals on descending paths by SPAISE2

(a)

Jun. 4 to Aug. 26, 2018

(b)

Aug. 27 to Nov. 18, 2018

(c)

Nov. 19, 2018 to Feb. 10, 2019

(d)

Feb. 11 to May 5, 2019

Figure 39.9 The modeling heatmap of AIS signals calibrated by footprints on ascending paths power peaks is different for ASE and DES. The FoV is shifted to the right according to the mounting angle of the cross-dipole antenna and attitude of the ALOS-2 satellite. Figures 39.9 and 39.10 show the heatmaps calculated using an antenna beam and satellite orbits. Figure 39.11 shows the merged heatmaps of all ASE and DSE paths for each quarter. The difficulty of observing the East China Sea, South China Sea, and Sea of Japan can thus be confirmed. In particular, the distribution is particularly strong around the route from Hong Kong to Taiwan and around Okinawa. A slight difference from the general distribution of ships may be due to the reporting intervals

510

(a)

Advances in communications satellite systems 2: ICSSC-2019

Jun. 4 to Aug. 26, 2018

(b)

Aug. 27 to Nov. 18, 2018

(c)

Nov. 19, 2018 to Feb. 10, 2019

(d)

Feb. 11 to May 5, 2019

Figure 39.10 The modeling heatmap of AIS signals calibrated by footprints on descending paths

Figure 39.11 The calibrated heatmap of merged AIS signals of AIS depending on ship speed. Moreover, there may be errors regarding the lack of experimental data, geometric correction inaccuracies, and changes in distribution over the period. These results roughly correspond to the AIS plots collected by space-based AIS as shown in Figure 39.12.

39.4 Conclusion In conclusion, the present study has confirmed seasonal fluctuations of RSSI and RFI in the seas around Asia from the SAT-AIS. Several external interference sources and high signal density ranges were confirmed. Such data reveal the

Experimental study of external interference for LEO-based sensing

511

Figure 39.12 AIS plots collected by space-based AIS number of required reception antennas and Digital Beam Forming (DBF) conditions and also provide the need and effective range of alternatives, such as a ship’s radio profiling and satellite communications. However, certain challenges remain for future research on model optimization and DBF optimization by future missions.

References [1] ORBCOMM. 2018. “Satellite AIS.” Available from https://www.orbc omm. com/en/networks/satellite-ais/ [Accessed September 12, 2018]. [2] exactEarth. 2018. “Satellite AIS overview.” Available from https://www. exactearth.com/technology/satellite-ais/ [Accessed September 12, 2018]. [3] Spire. 2018. “Satellite and terrestrial ais data services products.” Available from https://spire.com/-data/maritime/ [Accessed September 12, 2018]. [4] HawkEye 360. 2019. “RFGeoTM.” Available from https://www.he360.com/ products/rfgeo/ [Accessed September 13, 2019]. [5] Kleos Space S.A. 2019. “Enhancing your geospatial intelligence.” Available from https://kleos.space/ [Accessed September 13, 2019]. [6] JAXA. 2003. “About advanced land observing satellite-4 (ALOS-4).” Available from https://global.jaxa.jp/projects/sat/alos4/ [Accessed September 13, 2019].

512

Advances in communications satellite systems 2: ICSSC-2019

[7] Hirahara, D. “Ultra-high performance satellite AIS utilizing ground-based DBF for marine traffic density area in the sea around Japan.” IEICE. SAT2016-19. 2016; 116(144): 45–50. [8] JAXA. 2003. “About advanced land observing satellite-2 “DAICHI-2” (ALOS2).” Available from https://global.jaxa.jp/projects/sat /alos4/ [Accessed September 13, 2019]. [9] Hirahara, D., Shimizu, S., Ishizawa, J., Miura, S., and Suzuki, S., “Received signal analysis of SPace based AIS experiment 2nd (SPAISE2) in the sea around Japan (part 4).” IEICE Technical Report. SAT2018-52. 2018; 118 (237): 51–56. [10] NASA. “Updated global survey of RFI levels observed by the aquarius and SMAP scatterometers at 1260 MHz and radiometers at 1413 MHz.” SFCG35, SF35-22/D. July–August 2015. [11] ESA. “Revision of SFCG report 34-2 on the global RFI survey of EESS L-band sensors: SMOS RFI update by July 2015.” SFCG-35, SF35-27/I. July–August 2015. [12] Zhou, M., Veen, A., and Leuken, R. “Multi-user LEO-satellite receiver for robust space detection of AIS messages.” IEEE ICASSP 2012. 2012. pp. 2529–2532. [13] Luxspace. “Technical note TN 4.2 PASTA MARE interference mapping issue 2 September 2010.” Preparatory Action for Assessment of the Capacity of Spaceborne Automatic Identification System Receivers to Support EU Maritime Policy. August 2012. [14] Vaseghi, S.V. Advanced digital signal processing and noise reduction. New York, NY: John Wiley & Sons; 2008.

Section 10

Future technologies for 5G and beyond

This page intentionally left blank

Chapter 40

Advanced demonstration plans of high-speed laser communication “HICALI” mission onboard the engineering test satellite 9 Yasushi Munemasa1, Yoshihiko Saito1, Alberto Carrasco-Casado1, Dimitar R. Kolev1, Phuc V. Trinh1, Hideki Takenaka1, Kenji Suzuki1, Toshihiro Kubo-oka1, Tetsuharu Fuse1, Hiroo Kunimori1, Koichi Shiratama1, Yasushiro Takahashi1 and Morio Toyoshima1

The National Institute of Information and Communications Technology (NICT) in Japan has over 20 years of experience in R&D of space–ground laser communications with missions. We are currently developing a laser communication terminal named “HICALI” (HIgh-speed Communication with Advanced Laser Instrument), aiming to achieve 10 Gbps-class space communications with a 1.5 mm-band laser beam between optical ground stations (OGSs) and the next-generation high-throughput satellite called ETS-9 with a hybrid onboard communication system using radio and optical frequencies, which will be launched into the geostationary orbit in 2021. Moreover, we have studied laser communication terminals for terrestrial networks, as an alternative wireless system to radio frequency (RF) band. The development of a test and breadboard model for HICALI has been conducted for several years and we are now carrying out an engineering model as well as designing the OGSs segment. In this chapter, we describe the current development status and advanced demonstration plans of the “HICALI” mission with ETS-9. Key Words: laser communications; satellite communications; HICALI

1 Space Communications Laboratory, Wireless Networks Research Center, National Institute of Information and Communications Technology (NICT), Koganei, Tokyo, Japan

516

Advances in communications satellite systems 2: ICSSC-2019

40.1 Introduction Recently, the capabilities of remote-sensing satellites are becoming much more sophisticated, with an increasing number of sensors with more and more resolution. Therefore, the data gathered by these satellites is becoming larger and larger. To deal with the increasing bandwidth requirements, many satellite-communication operators around the world are developing broadband satellite-communication services based on the Ka-band. In December 2010, Eutelsat launched KA-SAT, an example of a high-throughput satellite (HTS). The KA-SAT satellite has 82 spot beams and a capacity in excess of 70 Gbps [1]. In October 2011, the ViaSat-1, known as HTS, was launched, with a capacity of 140 Gbps. Since January 2012, a satellite-based broadband Internet Protocol (IP) service has been deployed over North America. Sometime after 2019, ViaSat-3 should be providing global coverage with a capacity of 1 Tbps [2]. The Inmarsat-5 satellite network is currently providing the Global Xpress service, which is a Ka-band broadband satellitecommunication service with 5/50 Mbps for uplink/downlink. Since August 2015, three operational Inmarsat-5 satellites have been launched to the Geostationary Earth Orbit (GEO), giving global coverage with 89 spot beams per satellite [3]. In general, compared to laser communication systems, the RF-based satellite communication systems can simplify the mechanism of acquisition and tracking for communication beams, and also, the service area per beam is wider than that of laser communication systems. The RF systems are easy to use for portable or mobile communication systems on the ground. However, the above-mentioned HTS systems are spread out by the worldwide mega constellation. On the ground segment, the 5th generation (5G) mobile communication system service will be provided in the beginning of 2020s, featuring the higher bandwidth, which is in the L-band, S-band, and C-band, in order to meet the increasing demand of the high throughput wireless communications. In the near future, the available Ka-band for RF satellite communications may be tight or depleted. Furthermore, downlink and uplink communication capacities between the satellites and ground stations using RF band are usually very limited due to the compromise of resources (power consumption, size capacity, mass, placement, etc.) within the satellite and frequency regulation issues. On the other hand, laser communication technologies are attracting a great deal of attention because of their large capacity in data transmission. In addition, the size and mass of optical terminals are smaller than their RF counterparts. Since the restriction of satellite resources is tighter for smaller satellites, the combination of optical communications and small satellites is regarded as suitable for applications generating large amounts of data. The beam of laser communications is extremely narrow, the usages of laser communication system are limited, such as single link systems for intersatellite relay communication in the space. However, the systems offer many advantages for wireless communications, which are high capacity data link and highly secure communication such as a cryptographic key distribution system.

Advanced demonstration plans of high-speed laser communication After 10–20 years later, Intersatellite links will be all optical communication instead of RF but feeder links will need both of them. Broadband satellite comm.

517

Small sat. laser comm.

Hand over

Space/ocean broadband satellite communications network technology

Global optical communications network technology

Exclusive economical zone (EEZ) Ocean platforms Underwater autonomous vechicles (UAVs)

Laser comm. is need for Gbps class comm. Optical ground stations and site diversity technology and Ka band comm. will produce the boradband communication means and contribute to the future space big-data.

Figure 40.1 Overview of the R&D activities of the NICT’s Space Communications Laboratory Then, we are embarking on feasibility study of a scalable laser communication system and terminal for field application in the space and terrestrial communication scenarios. Figure 40.1 shows an overview of the R&D activities of the NICT’s Space Communications Laboratory. Over the next 10–20 years, the trend in space communications is to replace RF systems with ones based on laser. For this, it is necessary to utilize site diversity technique to mitigate the unavailability of individual laser links due to local weather conditions, which requires international collaboration. The target of the transmission speed should be around 100 Mbps for mobile users at Ka-band in the next-generation HTS communication systems [4]. Then, we are currently developing a laser-communication terminal called “HICALI” (HIgh-speed Communication with Advanced Laser Instrument), with the goal to achieve 10 Gbps-class space communications in the 1.5-mm band between OGSs. In this chapter, we describe the current development status and advanced demonstration plans of the “HICALI” mission with ETS-9.

40.2 Overview of HICLI project Regarding the R&D of the next generation of communication satellites, NICT has established a user consortium to identify the future needs of communicationsatellite users, studied satellite-communication system concepts covering those needs, and settled on technical issues for increasing communication speeds. Then, we have initiated HICALI project to facilitate the next-generation space laser-communication research. The aim of the project is to achieve 10 Gbps-class

518

Advances in communications satellite systems 2: ICSSC-2019

space communications with a 1.5-mm-band laser beam from a geostationary satellite to ground. The terminal is going to be launched on GEO in 2021, as a communication mission payload of ETS-9. The main objectives of the HICALI project are as follows: 1. 2. 3. 4.

In-orbit verifications of the first 10-Gbps-class laser communication from GEO to ground, In-orbit verifications of novel optical modulation/ demodulation methods, In-orbit verifications of novel high-speed optical devices, and Acquisition of laser-beam propagation data and in-orbit experiences. New approaches will be explored to find new usages such as:

● ●

Acquisition of development knowledge in conjunction with manufacturers and Search for new users who have the potential to use laser communications.

A feasibility study for the HICALI project was conducted in 2014, and to date, a number of critical parts for the project have been identified. A breadboard model (BBM) was developed in 2015 and the evaluation was done in 2016. Key devices such as optical delay line interferometers, tunable lasers, optical modulator, optical detector, clock data recovery circuit, and high-speed digital processing devices are based on commercial off-the-shelf (COTS) parts. In addition to the objectives, it is expected that NICT and Japanese manufactures will be able to acquire knowledge of the development of space lasercommunication components and explore new users who have the potential to use laser communications through the HICALI project. The detailed design of the terminal was done in 2018, the main specification is shown in Table 40.1. Figure 40.2 shows the functional block diagram of the HICALI system. To realize ultra-high-speed optical communications, the HICALI space segment consists of mainly five components: OHA (Optical Head Assembly), OAMP (Optical AMPlifier), OTRX (Optical Transmitter and Receiver), and HDU (HICALI Data Unit).

Table 40.1. HICALI main specification Item

Characteristics

Remark

Data rate Modulation

Down/up link: 10 Gbps Down link: NRZ-DPSK Up Link: RZ-DPSK 2.5 W >40 nW 150 mm Approx. 80 kg Approx. 340 W

Nominal

TX power RX power Antenna diameter Mass Power consumption

HPA output LNA input System total, nominal System total, nominal

Advanced demonstration plans of high-speed laser communication

519

Engineering test satellite (ETS-9) Ka-RF communication mission RF Signal processor

Common communication system (RF beacon & HICALI TLM TX)

RF TRX

HICALI space segment

OTRX (optical TRX)

HDU (HICALI data unit)

OAMP (optical AMP)

High res. HICALI TLM

ETS-9 BUS sys.

OHA (optical head assembly)

CMD/TLM

1.5-µm band CMD /TLM

Ka-band

RF earth station

Satellite Operation Center ALI TLM High res. HIC

OGSs

HICALI ground segment

Figure 40.2 The functional block diagram of HICALI system OHA is in the optical section, which is mainly composed of optics and beamsteering mechanics for acquisition and tracking the OGS. The HICALI communication transmitting/receiving beams are collimated using the optics installed in the OHA and the beams (transmitting and receiving optical signals) travel to OTRX via OAMP. OTRX consists of an MZ modulator and DBF-LDs for transmitting, a delay-line interferometer, a balanced receiver, and a clock data recovery circuit for receiving, and high-speed digital processing devices for high-speed code or decode processing. HDU has functions of control and storage of the telemetry for OHA, OTRX, and OAMP and receiving command and sending real-time telemetry for HICALI system via the satellite BUS system. Since the telemetry capacity is limited in HICALI, HDU can send detailed mission telemetry via Ka-band RF Communication Mission subsystem, which are high-resolution monitoring data of HICALI components, receiving power at OHA and OAMP for analyzing the influence of atmospheric turbulence during uplink communication and measuring the BER (Bit Error Rate) at OTRX for system analysis.

40.3 Demonstration plans of HICLI project Figure 40.3 shows a conceptual diagram of the optical communication experiment between the next technical test satellite using HICALI and the ground. To realize the world’s highest optical communication speed of up to 10 Gbps and to ensure RF bandwidth resource availability is becoming tight at the maximum, the HICALI

520

Advances in communications satellite systems 2: ICSSC-2019 Portable station RF

Overseas station

Use

r lin

k

ETS-IX

Kashima Koganei

HICALI space segment Okinawa

HICALI feeder link CM

Detailed telemetry

Fe ed er l

in

k

D Satellite control

RF

OGS2 OGS1

RF-gateway station

HK telemetry

Ground station (telemetry and command)

CMD HK telemetry

Figure 40.3 A conceptual diagram of the optical communication experiment between the next technical test satellite using HICALI and the ground stationary orbit to the ground optical communication experiment is being conducted. The following items are planned: 1. 2. 3.

4. 5. 6. 7. 8.

Confirm basic function operation of large-capacity optical communication device on-orbit; Confirm ultra-high-speed optical communication function (10 Gbps); Propagate data of laser light, check the channel quality, coding/interleaver function/performance so that various kinds of communication methods for reducing degradation of communication quality due to atmospheric fluctuation can be demonstrated; Confirm the communication function of conversion of light wave to radio waves and radio wave to light wave; Perform the optical communication experiment at night (experiment in daytime as extra target); Site diversity experiment according to weather conditions; Test a new technology on optical ground station (OGS), such as adaptive optics (AO) system, multiaperture transmission/reception; and Conduct directional/tracking test using a portable (vehicle-mounted type) OGS.

Currently, the development of HICALI is at the final stage of detailed design for optical components and optics.

40.4 Conclusion In this chapter, we described the current development status and advanced demonstration plans of the optical feeder link aiming for demonstration with

Advanced demonstration plans of high-speed laser communication

521

HICALI installed in ETS-9. The demonstration of HICALI will replace the radio feeder link, since less RF bandwidth availability will result as communication data rates increase in the future. Currently, CCSDS (the Consultative Committee for Space Data Systems), which is a standardization activity in the field of space data systems, setting standards for communication and exchange of spatial data in satellites, between satellites and ground stations and deep space to ground stations. Discussion on adopting the standard is underway. HICALI plans to adopt standard techniques as much as possible with the aim of spreading adoption of the developed system widely.

Acknowledgments The authors would like to express their appreciation to the members of HICALI project members in Astro Terrace Inc., BridgeComm Inc., and NEC Corp., the members of ETS-9 satellite bus system project team in JAXA (Japan Aerospace Exploration Agency), and Melco (Mitsubishi Electric Corporation), who support the progressing project.

References [1] Fenech, H., Lance, E., Tomatis, A., and Kalama, M. “Next generation high rate broadband satellites.” 16th Ka-band Conference. 2010. [2] Viasat Inc. High-capasity satellite system. Available from https://www.viasat.com/products/high-capacity-satellites [Accessed January 1, 2018]. [3] Inmarsat plc. Newsroom: Inmarsat global Xpress full global coverage completion. December 3, 2015. Available from https://www.inmarsat.com/news/ inmarsat-global-xpress-fullglobal-coverage-completion/ [Accessed January 1, 2018]. [4] Toyoshima, M., Fuse, T., Carrasco-Casado, A., et al. “Research and development on a hybrid high throughput satellite with an optical feeder link.” Proceedings of IEEE International Conference on Space Optical Systems and Applications 2017. Naha, Japan. 2017, S9.3.

This page intentionally left blank

Chapter 41

Optical communication experiment with microsatellite body-pointing using VSOTA on RISESAT Hideki Takenaka1, Hiroo Kunimori1, Toshinori Kuwahara2, Yuji Sakamoto2, Shinya Fujita2, Homio Tomio2, Morokot Sakal2, Junichi Kurihara3, Toshihiro Kubo-oka1, Tetsu Fuse1 and Morio Toyoshima1

The optical communication device VSOTA (a very small optical transmitter) developed by National Institute of Information and Communications Technology (NICT) was installed on the Rapid International Scientific Experiment Satellite (RISESAT) and launched in 2019. We are starting an initial orbit check using VSOTA. In this chapter, initial checkout experiments are performed and confirmed that VSOTA functions are operating normally. Key Words: free-space optical communications; satellite communications; microsatellite

41.1 Introduction In recent years, the performance of observation equipment mounted on satellites has improved such that it can obtain an increased amount of data from only a single observation. Radio waves are used as a method for transmitting large volumes of data acquired by satellites to the ground. However, the radio frequencies used make it difficult to improve the communication speed, owing to interference problems and the carrier frequency. Space optical communication is expected to be a solution to this problem. National Institute of Information and Communications Technology (NICT) developed small optical transponders (SOTA) for very small satellites, launched in 1

National Institute of Information and Communications Technology, Koganei, Japan Tohoku University, Sendai, Japan 3 Hokkaido University, Sapporo, Hokkaido, Japan 2

524

Advances in communications satellite systems 2: ICSSC-2019

May 2014, within the framework of the Space Optical Communication Research Advanced Technology Satellite (SOCRATES) project [1,2]. SOTA is a gimbaled optical communication device capable of 10 Mbps communication. A very small optical transmitter (VSOTA) has been developed with the technology developed at SOTA. VSOTA is an optical communication device that uses some of SOTA’s functions and is not equipped with gimbals [3]. Its mass is less than 1 kg, and optical communication is performed by satellite body-pointing. The optical communication speed is performed by body-pointing and pointing accuracy is lower than gimbal-mounted type; however, it can be installed easily on satellites. A comparison of VSOTA and SOTA is shown in Table 41.1. VSOTA has two LDs and can output two different wavelengths. Also, LD is output from each collimator at different divergence angles. In this chapter, we will describe the progress of the initial check experiment regarding the VSOTA experiment conducted after the launch. After launch, it was confirmed that VSOTA was operating without problems on the orbit.

41.2 Component of VSOTA VSOTA was mounted on a small satellite (RISESAT: Rapid International Scientific Experiment Satellite) developed by Tohoku University and launched in January 2019 [4]. Table 41.1 shows a comparison of VSOTA and SOTA. The VSOTA collimator (VSOTA-COL) and VSOA electrical components (VSOTA-E) are shown in Figure 41.1. VSTOA does not have a gimbal, so it weighs less than 1 kg and consumes up to 5 W. In addition, it has LDs of 980 nm (TX1) and 1550 nm (TX4) for communication but does not have a received detector. The communication speed is 1–100 kbps. The satellite performs satellite-to-ground optical communication while tracking the ground station using body-pointing as shown in Figure 41.2 because VSTOA does not have a gimbal. Table 41.1 Specifications of VSOTA and SOTA

Mass Power consumption Wavelength Divergence angle Wavelength Data rate

TX1 TX4 TX2, TX3 TX1 TX4 RX

VSOTA

SOTA

100 Gbps dual polarization coherent transceivers). For such missions, a combination of notch filters, suppressing the reflected transmit signal with band-pass filters may not be enough. In the developed bench, an additional space filter has been considered. All reflecting surfaces have been examined and the components with highest impact are slightly misaligned. In combination with a few masks, the reflected signal is removed from the receiving path. The main drawback is that such solution is limiting the field of view of the receiver and may be a problem for the tracking and acquisition system. An example of the performed analysis is shown in Figure 42.8.

Figure 42.7 A picture of the developed optical bench prior to being mounted on the telescope Nasmyth

Surf 1

Surf 2

Surf 3

Surf 4

Surf 5

Surf 6

FPM mask

QD sensor

Figure 42.8 An example of the transmitted light reflections analysis and special filtering

Optical ground station supporting GEO- and LEO-to-ground links

539

42.4 Conclusion In this chapter, the newly developed optical bench in NICT has been introduced and the necessary updates for different types of LEO and GEO missions have been discussed. Furthermore, uplink wavefront aberration problem was defined and a novel solution for uplink wavefront precompensation was proposed.

References [1] Koyama, Y., Toyoshima, M., Takayama, Y., et al. “SOTA: Small optical transponder for micro-satellite.” 2011 International Conference on Space Optical Systems and Applications (ICSOS). pp. 97–101. [2] Edwards, B.L., Daddato, R., Schulz, K.-J., et al. “An update on the CCSDS optical communications working group.” 2017 IEEE International Conference on Space Optical Systems and Applications (ICSOS). Naha. 2017. pp. 1–9. [3] Fuse, T., Akioka, M., Kolev, D., et al. “Development of a breadboard model of space laser communication terminal for optical feeder links from GEO.” ICSOS. Biarritz, France. 2016 [4] Petit, C., Vedrenne, N., Michau, V., et al. “Adaptive optics results with SOTA.” 2015 IEEE International Conference on Space Optical Systems and Applications (ICSOS). New Orleans, LA. 2015. pp. 1–7. [5] Wright, M.W., Morris, J.F., Kovalik, J.M., Andrews, K.S., Abrahamson, M. J., and Biswas, A. “Adaptive optics correction into single mode fiber for a low Earth orbiting space to ground optical communication link using the OPALS downlink.” Optics Express. 2015; 23: 33705–33712. [6] Schmidt, C. and Fuchs, C. “The OSIRIS program at DLR.” Proceedings of SPIE 10524, Free-Space Laser Communication and Atmospheric Propagation XXX. 2018. [7] Andrews, L.C. and Phillips, R.L. Laser beam propagation through random media. SPIE Press, Bellingham, Washington, USA; 2005. [8] Pennington, D.M. “Laser guided adaptive optics for high-resolution astronomy.” Summaries of Papers Presented at the Lasers and Electro-Optics. CLEO ‘02. Vol.1. Technical Diges, Long Beach, CA, USA. 2002. pp. 68–69. [9] Pique, J.P. and Farinotti, S. “Efficient modeless laser for a mesospheric sodium laser guide star.” Journal of the Optical Society of America B. 2003, vol. 20, no. 10, pp. 2093–2101.

This page intentionally left blank

Chapter 43

Optical observations of nonoperational satellites in graveyard orbits Michelle K. Turberfield1,2, Tetsuharu Fuse2,3 and Toshihiro Kubo-oka1

National Institute of Information and Communications Technology (NICT) plans to launch Engineering Test Satellite-9 (ETS-9) in the geosynchronous Earth orbit (GEO) in 2021. When the operation of a GEO satellite is terminated, it must be removed from the GEO protected region and maneuvered to a graveyard orbit. For example, ETS-8, former generation of ETS-9, ended its operation as planned in 2017 and moved to a graveyard orbit. Wideband Inter Networking engineering test and Demonstration Satellite (WINDS) suddenly ended its operation due to a communication error in 2019 and it could not be maneuvered to a graveyard orbit. From the above, we should be aware of the rotation of the objects not only in graveyard orbits but also of objects in GEO to ensure the safety of the operation of ETS-9 and also telecom, earth observation, and navigation satellites. In this study, we carried out optical observations for some of nonoperational satellites including ETS-8 and WINDS using a CCD camera attached to a 1-m telescope. We will report how to select the nonoperational satellites, the results of the optical observations, and the light curves of the satellites using photometry. As a result, we estimate the rotational status of the nonoperational satellites from the light curves. Key Words: optical observation; geostationary satellite; graveyard orbit

43.1 Introduction NICT is conducting research in the field of satellite laser communications, aiming for higher capacity links [1]. 1

National Institute of Information and Communications Technology, Koganei, Tokyo, Japan Graduate School of Information Systems, University of Electro-Communications, Chofu, Tokyo, Japan 3 National Institute of Information and Communications Technology, Ibaraki, Japan 2

542

Advances in communications satellite systems 2 : ICSSC-2019

In 2006, we launched ETS-8 and conducted communication experiments for 11 years (Figure 43.1) [2]. In 2008, we launched WINDS and conducted experiments on high-speed and large-capacity communications between satellites and ground stations (Figure 43.2) [3]. The main goal of the development of HIgh-speed Communication with Advanced Laser Instrument (HICALI) is to establish basic technology for optical feeder links. HICALI will be installed in ETS-9 and will be launched to GEO in 2021 (Figure 43.3) [4]. When the operation of a GEO satellite such as "ETS-9" is terminated, it must be removed from the GEO protected region and maneuvered to a graveyard orbit. For example, ETS-8, former generation of ETS-9, was terminated its operation as planned in 2017 and moved to the graveyard orbit and ceased to its function. Due to a communication error, JAXA determined that WINDS could not be operated from the ground in February 2019 and terminated WINDS operation, so it could not be maneuvered to a graveyard orbit. From the above, the number of nonoperational satellites in graveyard orbits and GEO will continue to increase. In order to prevent the generation of new space debris, it is necessary for satellites to have structure with a strength that does not decompose by the rotation after its operation. The rotation of the satellites in graveyard orbits or GEO can be grasped by optical observation. Kurosaki et al. (2019) are conducting research to estimate the rotational component by optical observation and Fourier transform to grasp the rotational status of geostationary satellites [5]. The study by Okumura et al. (2019) reported the light curve of ETS-8 and roughly reproduced with a simple model consisting of the main body, large antenna, and solar cell paddle [6]. To clarify the rotation of the objects in graveyard orbits, we carried out optical observations of some nonoperational satellites, both in GEO and in a graveyard orbit, including ETS-8 and WINDS using a CCD camera attached to a 1-m telescope, and estimated the rotational motion of the satellites from light curves obtained by the observations using the 1-m telescope.

Figure 43.1 ETS-8

Optical observations of nonoperational satellites in graveyard orbits

543

Figure 43.2 WINDS

Figure 43.3 ETS-9

43.2 Definition of graveyard orbit The graveyard orbit is the orbit in which a geostationary satellite moves when its operation ends. The Inter-Agency Space Debris Coordination Committee (IADC) determined the recommended perigee for maneuverer to a graveyard orbit can be calculated from DH ¼235 þ (1,000 CR A/m) [km] [7], where DH is the minimum increase in perigee altitude from GEO, CR is the solar radiation pressure coefficient, and A/m is the aspect area to dry mass ratio (m2 kg1). The distance 235 km is the sum of the upper altitude of the GEO protected region (200 km) and the maximum descent of a reorbited spacecraft due to lunisolar and geopotential perturbations (35 km) [7]. In addition, the eccentricity is also recommended to be less than or equal to 0.003. All satellites launched after 2002 must comply with the above equation and must be maneuverer to graveyard orbits according to the definition at the end of the operation, but there are also satellites that fail to transition, such as WINDS.

43.3 Methodology 43.3.1 Optical system We carried out optical observations of some nonoperational satellites, both in GEO and in a graveyard orbit, including ETS-8 and WINDS using a cooled CCD camera

544

Advances in communications satellite systems 2 : ICSSC-2019

Figure 43.4 1-m telescope at Kashima Space Technology Center

Table 43.1 Specification of 1-m telescope Name

Value

Optics type Focal length Diameter Field of view

Classical Cassegrain reflector F ¼ 12 1000 mm 7.9  5.3 arcs

attached to a 1-m telescope located in the Kashima Space Technology Center (Figure 43.4). Table 43.1 is the specifications of the telescope optical system. Figure 43.5 shows the cooled CCD camera used for observation attached to the 1-m telescope and Table 43.2 is the specifications of the cooled CCD camera.

43.3.2 Selecting satellites The satellites in graveyard orbits have longer orbital radius and longer periods than the GEO satellites. Therefore, it is necessary to select satellites to be observed because there is no satellite in a graveyard orbit that stays still at one point. We use two line elements (TLE) from Space Track [8], which shows the orbital elements of the objects in orbits provided by the North American Aerospace Defence Command (NORAD) to make observation plans. However, TLE does not show whether the satellite is in operation or has been terminated. Then we use the satellite catalog (SATCAT) from CelesTrak [9] to select the satellites that had been terminated using the codes indicating the operational status of each satellite. We confirmed the satellite that is large enough to observe using radar cross section (RCS) data of SATCAT. When we observe the same satellite after a while, we

Optical observations of nonoperational satellites in graveyard orbits

545

Figure 43.5 1-m telescope with cooled CCD camera

Table 43.2 Specification of a cooled CCD camera Name

Value

CCD camera Pixels Pixel size

SBIG STX-6303 3072  2048 0.2  0.2 arcs

calculated the period using the following equation 1/p ¼ |1/p1 – 1/p2|, where p1 is a satellite in graveyard orbit and p2 is a GEO satellite.

43.4 Image processing 43.4.1 Image processing with IRAF We performed image reduction for captured images. First of all, the dark frame was taken with the telescope lid closed and the light blocked in the container. The exposure time was adjusted to when each satellite image was taken. For the flat frame, we took the images of the twilight flat using the twilight sky. For image reduction, we use IRAF which is an astronomical image processing software developed by the National Astronomical Observatory of America. Using IRAF, the dark frame is composed of 10 images and taken median to synthesized to form a master dark frame. The flat frame is synthesized after subtracting the master dark frame to make a master flat frame. Finally, the master dark frame is subtracted from the images and standardization is performed using the master flat frame.

43.5 Observation 43.5.1 Observation Table 43.3 shows the observation date and satellites observed. In this chapter, we report the results taken on July 27, 2019. In this experiment, we observed about

546

Advances in communications satellite systems 2 : ICSSC-2019 Table 43.3 The list of observed satellites Observation date

Satellites name

January 28 and 29, 2019 February 25, 2019 May 16, 2019 July 27, 2019

ECHOSTAR-3, EUTE 12, JCSAT3, NSTAR-A, SUPERBIRD-C, WINDS ETS-8, WINDS ETS-8, WINDS ETS-8

Figure 43.6 ETS-8 in July 27, 2019. The point source is ETS-8 and line sources are stars 40 min and took 300 images of ETS-8. The total amount of cloud at the time of observation was about 33% and the temperature was about 26  C. Figure 43.6 is one of the images of ETS-8 taken in July 27, 2019.

43.6 Photometry 43.6.1 Photometry with IRAF We identified the stars in the image using the star catalog USNO-2.0. Thereafter, the relative magnitude of the satellite can be calculated from the magnitude of the star and the count values of the star and the satellite. After the image reduction was described, photometry was performed using the phot task of the apphot package of IRAF. Figure 43.7 shows the photometric results of the ETS-8 observations. The vertical axis is the count value before calculating the satellite magnitude, and the horizontal axis is the observation time. As a method of photometry, the imexam task was executed on the point image to measure the full width at half maximum (FWHM), and photometry was performed using the phot package with a photometric range of three times the FWHM. From Figure 43.7, it can be seen that ETS-8 rotates once in about 8 min because the count value repeatedly increases and

Optical observations of nonoperational satellites in graveyard orbits

547

16,00,000 1 s exposure Observing time = 0:39:45 0.0 = 21:15:13 (JST) Read out time = 8.0 s

14,00,000

Count

12,00,000 10,00,000 8,00,000 6,00,000 4,00,000 2,00,000 0 0

5

10

15 20 25 Elapse time (min)

30

35

40

Figure 43.7 Light Curve of ETS-8 in July 27, 2019

Echostar lightcurve

16,00,000

5 s exposure Observing time = 0:11:36 0.0 = 19:05:54 (JST) Read out time = 12.0 s

14,00,000 12,00,000

Count

10,00,000 8,00,000 6,00,000 4,00,000 2,00,000 0 0

2

4

6 8 Elapse time (min)

10

12

Figure 43.8 Light Curve of ECHOSTAR 3 in January 28, 2019 decreases. Figures 43.8–43.12 are the light curves of ECHOSTAR 3, EUTE12, JCSAT 3, NSTAR-A, and Superbird-C taken in January 28, 2019. From the light curves, we found that each nonoperational satellite has different rotational elements.

43.7 Conclusion In this study, we performed optical observations of the nonoperational satellites using the 1-m telescope at the Kashima Space Technology Center. As a result, we

548

Advances in communications satellite systems 2 : ICSSC-2019 Eute12 lightcurve

16,00,000 14,00,000 12,00,000

Count

10,00,000 8,00,000 6,00,000 4,00,000

5 s exposure Observing time = 0:11:26 0.0 = 19:19:12 (JST) Read out time = 12.0 s

2,00,000 0 0

2

4

6 8 Elapse time (min)

10

12

Figure 43.9 Light Curve of EUTE 12 in January 28, 2019

JCSAT 3 lightcurve 16,00,000 5 s exposure Observing time = 0:11:28 0.0 =19:32:36 (JST) Read out time = 12.0 s

14,00,000 12,00,000

Count

10,00,000 8,00,000 6,00,000 4,00,000 2,00,000 0

0

2

4

6 8 Elapse time (min)

10

12

Figure 43.10 Light curve of JCSAT 3 in January 28, 2019

got the image of the light curve of ETS-8, ECHOSTAR 3, EUTE12, JCSAT 3, NSTAR-A, and Superbird-C. As future work, we plan to compare the results with the previous research by Okumura et al (2019), to estimate the rotation period by Fourier transform from the light curves, and to investigate the relationship with the phase angle.

Optical observations of nonoperational satellites in graveyard orbits

549

NSTAR-A lightcurve

16,00,000 14,00,000 12,00,000

Count

10,00,000 8,00,000 6,00,000 4,00,000

5 s exposure Observing time = 0:11:35 0.0 = 19:51:09 (JST) Read out time = 12.0 s

2,00,000 0

0

2

4

6 8 Elapse time (min)

10

12

Figure 43.11 Light curve of NSTAR-A in January 28, 2019

Superbird-c mag-curve

11

Magnitude

10

9

8

5 s exposure Observing time = 0:11:35 0.0 = 20:04:10 (JST) Read out time = 12.0 s

7

6 0

2

4

6 Time (min)

8

10

12

Figure 43.12 Light curve of superbird-C in January 28, 2019

Acknowledgments The authors are grateful to C. Alberto and D. Kolev for carefully proofreading the manuscript.

550

Advances in communications satellite systems 2 : ICSSC-2019

References [1] Space Communications Laboratory. Wireless Networks Research Center. Available from https://www2.nict.go.jp/spacelab/en/ [Accessed October 02, 2019]. [2] JAXA. Engineering Test Satellite VIII "KIKU No.8" (ETS-VIII). https:// global.jaxa.jp/projects/sat/ets8/ [Accessed October 02, 2019]. [3] JAXA. Overview of the "KIZUNA" (WINDS). https://global.jaxa.jp/countdown/f14/overview/kizuna_e.html [Accessed October 02, 2019]. [4] JAXA. Engineering Test Satellite VIIII "ETS-9." http://www.satnavi.jaxa.jp/ project/ETS-9/ [Accessed October 02, 2019]. [5] Okumura, S., Nishiyama, K., Fujiwara, T., et al. "Observed lightcurve and its shape model of the decommissioned satellite ETS-8." Proceedings of the 6th Space Debris Workshop. 2019. pp. 577–589. [6] Kurosaki, H., Yanagisawa, T., and Nakajima, A. "Observation of light curves of space objects." Advanced Maui Optical and Space Surveillance Technologies Conference. E80. 2009. [7] IADC. Report of the IADC activities on space debris mitigation measures. 41th Session of the Scientific and Technical Subcommittee United Nations Committee on the Peaceful Uses of Outer Space. 2007. [8] Space-Track.Org. Available from https://www.space-track.org/ [Accessed October 02, 2019]. [9] CelesTrak: Search Satellite Catalog. Available from http://celestrak.com [Accessed October 02, 2019].

Section 11

Flexible HTS systems and advanced digital payloads

This page intentionally left blank

Chapter 44

Development of Ka-band digital beam forming antenna payload for the engineering test satellite-9 Eiichi Sakai1, Yoshio Inasawa1, Masaaki Kusano1, Hitomi Ono1, Arimasa Kanasash1, Nobuyoshi Horie1, Terumi Sunaga1 and Toshiyasu Tsunoda1

The R&D project entitled “Research and Development of Ka-band Wideband Digital Beam Forming for Efficient Frequency Use” has been started since July 2017. The goal of this R&D is the realization of beam location/shape flexibility for Ka-band high throughput satellite (HTS) and the improvement of frequency use efficiency compared with conventional HTS. This DBF payload is developed and planned to be tested in orbit by the engineering test satellite-9 (launch planned in FY 2021). This chapter describes the overview of R&D program and the current outcome of system development. Key Words: digital beam forming antenna; communications satellite; digital payload; engineering test satellite-9

44.1 Introduction Technical development of the satellite transponder is recognized as a worthwhile activity along with that of satellite bus, satellite constellation system, etc. For example, high throughput satellite (HTS) technique by the frequency reuse, frequency flexibility technique, beam location/shape flexibility technique, and beam radiation power flexibility technique are important technologies for the current and future satellite transponder. Especially, the necessity of flexibility technologies is increasing more than ever, corresponding to the recent necessity of diversification and advancement for the satellite communication service/application (including broadband mobile communications such as ESIM, usage in the system of 5G cellular phone, Internet of things (IoTs), etc.). 1

Mitsubishi Electric Corporation, Chiyoda-ku, Tokyo, Japan

554

Advances in communications satellite systems 2: ICSSC-2019

In these flexibility technologies, beam location/shape flexibility technique, which forms the broadband multiple beams to meet the requirement of the satellite communication operations, is very useful for the satellite operators’ business. This technique can arrange the communications service area corresponding to the positions/configurations of user terminals. Although the most common conventional method is to drive the antenna reflector mechanically, which does not cover the beam shape flexibility function, broadband electronically scanning methods such as active phased array antenna (APAA) and also digital beam forming (DBF) are highly awaited. Kizuna, renamed from WINDS (Wideband InterNetworking engineering test and Demonstration Satellite), is a typical example of the communication satellite which adopt the broadband electronically scanning antenna. It has the Ka-band analog direct radiation APAA. Also, another satellite with the broadband electronically scanning antenna by analog APAA is currently announced to be launched in 2020. In the case of DBF, it is already adopted to the radars and narrow band communication satellites (e.g., L band/S-band satellite). However, in the commercial broadband communication satellites area, DBF is not yet suitable for the practical use. One of the technical challenges of DBF for the broadband communication is the compensation of deviation. Since frequency is one of the dominant parameter of the excitation coefficient, the DBF processing for the wide band frequency carrier signal requires to generate the adequate excitation coefficient of the array with the compensation corresponding to the end to end frequency range of the carrier signal. Furthermore, for the practical use of DBF on commercial satellites, it is necessary to consider the resource management of the satellite system such as power consumptions and mass of the payloads. To resolve these challenges and adopt the DBF to the future commercial HTS, it is desired to study the applicable antenna system and develop it.

44.2 R&D activities Based on the above recognition, the R&D project entitled “Research and Development of Ka-band Wideband Digital Beam Forming for Efficient Frequency Use” has been started since July 2017 [1,2]. Figure 44.1 shows the structure of this project. The following shows the typical examples of the activities and the results, which are the outcome of system engineering. They are of Subject-A and Subject-C as described in Figure 44.1.

44.2.1 Subject-A: system design and comprehensive evaluation System design had started from the requirements analysis for the future satellite communications systems, based on the current market trends of SatCom business. And then, with the result of this requirements analysis, requirement specification for the transponder which has the flexible beam forming subsystem was defined. Considering the realization and verification of the wideband DBF technology,

Development of Ka-band digital beam forming antenna payload

555

GOAL: • Realization of beam location/shape flexibility by the wideband DBF technology and the improvement of frequency use efficiency compared with current HTS. • Development of DBF payload for ETS-IX Subject-A: System design and comprehensive evaluation 1) Study of satellite communications system using DBF technology 2) Overall evaluation of satellite payload system using DBF technology 3) Study of ground system (structure, requirement, etc.) Subject-B: Development of DBF processor 1) Study of excitation coefficient calculation 2) Development of onboard DBF processor Subject-C: Development of antenna/RF for DBF 1) Development and evaluation of onboard transponder system 2) Development of evaluation of ANT/RF for DBF technology

Figure 44.1 Structure of the R&D project

Table 44.1 Required specification for transponder Item

Specification

Note

Frequency Range

Uplink: 29.50–29.75 (GHz) Downlink: 19.70–19.95 (GHz) IPFD Max: 170.6 (dBW/m2/Hz) Min: 193.9 (dBW/m2/Hz) Narrow beam: 10.7 (dB/K) Typical cases (defined 19 cases for Broad beam A: 4.8 (dB/K) G/T narrow and 20 for broad A/B) Broad beam B: 3.7 (dB/K) 56.24 (dBW) EIRP Total number of beams Tx (downlink): more than 2 Required the easy scalability Rx (uplink): more than 2

which are the main target of this R&D, this requirement specification defines the following as the main requirements. ●

●

Requires the beam forming capability of 125 MHz bandwidth per beam, and of forming two beams for downlink and also uplink. Requires the easy scalability for the total number of beams and for the bandwidth of each beam.

After the consideration of additional requirement which is of operational beam coverage, the main specification for transponder had been finalized as shown in Table 44.1.

556

Advances in communications satellite systems 2: ICSSC-2019 50

North longitude (°)

45

40

35 P10

P13 P14 P11 P12

P1

30

25 120

125

130

P18

P15 P16

P4 P5 P2 P3

P19

P6 P7

135

P8

P9

140

145

150

155

East longitude (°)

Figure 44.2 Allocation of reference points for uplink beam

Uplink RHCP LHCP

(a)

Beam 1

Beam 2 [GHz]

29.500

(29.625)

29.750

Downlink RHCP LHCP

(b)

Beam 1 19.700

Beam 2

(19.825)

[GHz]

19.950

Figure 44.3 (a) Spectrum assignment plan for uplink beams, and (b) spectrum assignment plan for downlink beams For the G/T in Table 44.1, the values are at the geographical reference points for uplink beam defined as P1–P19 (Figure 44.2) corresponding to the cases of beam pattern. Figure 44.3(a) and (b) shows the frequency assignment plan of the uplink beams and the downlink beams. Currently, the frequency coordination is undergoing by the regulatory section.

Development of Ka-band digital beam forming antenna payload DBF antenna RXRF-block

557

DBF processor

RXRF-block

RX TX DBFU DBFU

RXRF-block Fixed beam transponder Common ANT for Fixed beam ANT DBF ANT

Channelizer for fixedbeam DMX

MUX

ADC

DAC

RXRF-block From TX side

RXRF-block

Switch block (channelizer)

TXRF-block TXRF-block

RXRF-block To RX side

Figure 44.4 Block diagram of the transponder

Figure 44.5 Image of the transponder on the engineering test satellite-9

44.2.2 Subject-C: development of antenna/RF for DBF Based on the requirement specification, onboard transponder design has been conducted. In this design, flexible beam forming subsystem is implemented by incorporating the DBF function module into the antenna/transponder payload (including wideband digital channelizer), which is developing at another R&D project. This R&D project entitled “Research and Development of Bandwidth-onDemand High Throughput Satellite Communications System” had already been

558

Advances in communications satellite systems 2: ICSSC-2019

started from 2016 [3]. Figure 44.4 shows the block diagram of the transponder design. Figure 44.5 shows the image of the transponder on the target satellite.

44.3 Conclusion As described in Section 44.2, the R&D “Ka-band Wideband Digital Beam Forming for Efficient Frequency Use” is moving ahead according to the planned schedule. Current project phase is at the critical design and also in production of modules. And it is planned to deliver this payload subsystem to the satellite system in the second quarter of 2020, targeting the launch of the engineering test satellite-9 planned in the fiscal year of 2021.

Acknowledgments This study is conducted under the commissioned research of the “Research and Development of Ka-band Wideband Digital Beam Forming for Efficient Frequency Use” by the Ministry of Internal Affairs and Communications.

References [1] Sakai, E., Inasawa, Y., Kusano, M., et al. “Research and development of Ka-band wideband digital beam forming for efficient frequency use – Research Tasks and Plan.” Institute of Electronics, Information and Communication Engineers (IEICE) General Conference 2018. Tokyo, Japan. 2018. p. 226. [2] Sakai, E., Inasawa, Y., Kusano, M., et al. “Research tasks and plan for the R&D project ‘Ka-band wideband digital beam forming for efficient frequency use’.” 24th Ka and Broadband Communications Conference. Niagara Falls, Canada. 2018. [3] Miura, A., Morikawa, E., Yoshimura, N., et al. “On preliminary design of bandwidth-on-demand high throughput satellite communications system technology.” 24th Ka and Broadband Communications Conference. Niagara Falls, Canada. 2018.

Chapter 45

The initial study of calibrating receiving digital beam forming in engineering test satellite-9 Hitomi Ono1, Eiichi Sakai1, Yoshio Inasawa1, Masaaki Kusano1, Arimasa Kanasashi1, Nobuyoshi Horie1, Terumi Sunaga1 and Toshiyasu Tsunoda1

In realizing geostationary high throughput satellite (HTS), building flexible beam adapting to change of communication traffic is needed. Digital beam forming (DBF) has advantages of flexibility of beam steering and high integration. The R&D project focused on developing receiving DBF with Ka-band is ongoing. In DBF system, it is important to calibrate the gain, phase, and delay for each antenna element. In this chapter, we propose a method of calibration using digital processing on the satellite and ground station. The crosscorrelation vector between elements calculated in digital processor on the satellite is useful for detecting error of the gain, phase, and delay for each antenna element. These errors will be corrected with setting the correction coefficient to the satellite, which is calculated at the ground station with a cross-correlation vector transmitted from the satellite. Key Words: digital beam forming; calibration; engineering test satellite-9

45.1 Introduction Recently, a satellite communication system with high reliability for social infrastructure and high frequency use efficiency is needed. For adapting cover area to change of communication traffic, there are several ways of beam forming. Digital beam forming (DBF) is suitable for advanced phase control and dealing with multi-beams more than ten beams due to integration of circuit in comparison with active phased array antenna (APAA) using beam forming network. In addition to 1

Mitsubishi Electric Corporation, Chiyoda-ku, Tokyo, Japan

560

Advances in communications satellite systems 2: ICSSC-2019

this, for realizing higher throughput, high throughput satellite (HTS) with Ka-band has been developed [1]. Based on these, the R&D project, which is entitled “Research and Development of Ka-band Wideband Digital Beam Forming for Efficient Frequency Use,” has been started since July 2017. In this R&D, we focus on developing receiving DBF with Ka-band and planned to be tested in orbit by the engineering test satellite-9 (launch planned in the fiscal year 2021) [2,3]

45.1.1 System configuration of DBF Our DBF system is a mixed configuration of analog and digital components. The receiving DBF is composed of a feed array antenna with eight antenna elements and 1.5 m reflector. The analog components are such as low noise amplifier (LNA), down-converter (DCON), up-converter (UCON), and traveling-wave-tube amplifier (TWTA). The digital components are such as frequency-division de-multiplexer or multiplexer (DEMUX, MUX) and digital beam forming processor (DBFP). At first, the receiving signal reflected by reflector is input to the primary radiator of each element. The receiving signal input from the primary radiator is amplified with LNA and converted to intermediate frequency (IF) with DCON. After that, the receiving signal is converted to digital signal with 1.5 Gsps ADC, quadrature detected and demultiplexed into subchannel (ch) with the digital processor. Each subchannel has bandwidth of 2.5 MHz and DBFP can multiply different excitation coefficient on every subchannel and element. At last, the signals multiplied excitation coefficient of all elements are added and beam-formed. The transmitter is based on reverse procedure of the receiver and amplified with TWTA. In this configuration, we will realize two beams in uplink and downlink. The arrangement of the secondary radiation pattern of eight elements and reference points for uplink is shown in Figure 45.1(a). We defined some beams covering one or multiple reference points (P1–P19) and designed element arrangement (#1–#8) which can satisfy a requirement of antenna gain from satellite link design for each beam. Figure 45.1(b) shows an example of reconfigurable beam aimed at P5,6,7. The target of maximum throughput is 100 Mbps so that the processing bandwidth is 125 MHz/beam.

45.1.2 Difficulty of DBF In beam forming system, calibrating the gain, phase, and delay between elements is important. The gain and phase error causes coverage change (includes antenna gain degradation). In addition to this, the delay between elements causes intersymbol interference (ISI), which degrades the C/(NþI) ratio. Remarkably, realizing wideband DBF requires calibrating the gain, phase, and delay over a wide frequency band. For example, considering the sum of 100 Msps signal, one symbol has a period of 10 ns, so that if the time of interference requires less than 10% of the period, the difference of delay requires less than 1 ns. In addition to this, Ka-band of carrier wavelength makes phase calibration difficult. In our system, the RF path of each element before beam forming at DBFP is much longer than wavelength of Ka-band (sub-cm), so that it is difficult to manage the

Calibrating digital beam forming in engineering test satellite-9

561

7.0 6.5

Elevation (°)

6.0 5.5 5.0

–3 dB (Relative gain to peak)

4.5

4.0 –3.0 (a)

–2.5

–2.0

–1.5

–1.0 –0.5 Azimuth (°)

0.0

0.5

1.0

7.0 6.5

Elevation (°)

6.0 5.5 (Relative gain to peak) –3 dB –2 dB –1 dB

5.0 4.5 4.0 –3.0

–2.5

–2.0

–1.5

(b)

–1.0 –0.5 Azimuth (°)

0.0

0.5

1.0

Figure 45.1 (a) Arrangement of secondary radiation pattern of eight elements (#1–#8) and reference points (P1–P19) for uplink and (b) an example of reconfigurable beam to peak aimed at P5,6,7 phase and the delay of each element. To solve these problems, we discuss how to calibrate gain, phase, and delay between elements in this chapter.

45.2 Method of calibration The calibration concepts in DBF are classified into internal or external calibration [3]. The internal process is a closed system that executes all of the calibration

562

Advances in communications satellite systems 2: ICSSC-2019

processes on the satellite such as making calibration signal, detecting error, and compensating error. On the other hand, external calibration uses some external facilities. The former method takes advantage of less operational restrictions but the load of processing required to payload is large. In the engineering test satellite9, we chose the external calibration method because of digital processing capability. Our digital processor also has the function of switching among subchannels, which is a function of wide-band digital channelizer developed on the other program. We will assign a part of the function of detecting error to the satellite and other function to the terrestrial system.

45.2.1 Detection of gain/phase error between elements First, we explain about the process of detecting gain, phase, and delay error between elements. One of the calibrating methods of antenna elements is the rotating element electric field vector (REV) method [4]. However, this method has a disadvantage of taking a lot of time for rotating phase of each element and finding appropriate phase. Thus, we propose a method with cross-correlation vector and this method can detect phase and amplitude error quickly. We are planning calibration among antenna elements based on detecting error with the cross-correlation vector and updating the excitation coefficient added correction coefficient. To explain this method, we point to the function of digital processing and explain how to detect gain, phase, and delay error between elements [5]. Now we consider the signal (Xn) of subchannel (ch) after DEMUX. [ch] is a complex vector with real data and imaginary data because the digitally converted signal is quadrature detected in DEMUX. In other words, Xn has the amplitude (An) and phase data (fn). X n ½ch ¼ In ½ch þ jQn ½ch ¼ An e jfn

(45.1)

In (45.1), the subscript n means element number. For the following discussions, to simplify the problem, we assume the vectors of all elements after calibration has the same amplitude and the same phase. Here, we consider the cross-correlation vector between element m and n. Rmn ½ch ¼ X m ½chX n ½ch ¼ Am An e jðfm fn Þ

(45.2)

¼ Amn e jfmn As you can see in (45.2), this vector has the information of the complexconjugate product of amplitude and the difference of the phase between element m and n. In the case of considering the difference of group delay and phase, the phase of cross-correlation vector is written by below (45.3). fmn ¼ 2pDt  kf0 þ Df

(45.3)

Cross-correlation vector phase ϕmn

Calibrating digital beam forming in engineering test satellite-9

ϕmn = 2Δτkf0 + Δϕ

563

Delay correction

2Δτf0

Phase correction

Δϕ f0

k [ch] kf0 [Hz]

Figure 45.2 Assumed cross-correlation vector phase versus subchannel

Dt is the difference of group delay between element n and m, which corresponds to the phase rotation over 2p radian, and Df is the difference of phase angle between element n and m, which is 0–2p radian. The k means subchannel number and the f 0 means the bandwidth of subchannel. As you can see in (45.3) and Figure 45.2, the phase slope to the subchannel (equal to frequency step by 2.5 MHz/ch) in cross-correlation vector represents the difference of group delay and offset is the common phase error between elements. We can set the correction coefficient to each subchannel so that if the phase of the correction coefficient is set to the sum inverted in sign of slope and offset in cross-correlation vector phase, the difference of group delay and phase is corrected. Next, in order to consider the gain error, we calculate the ratio of the magnitude of the cross-correlation vector and autocorrelation vector as below. jRmn ½chj X m ½chX n ½ch ¼ jRnn ½chj X n ½chX n ½ch (45.4) Am ½ch ¼ An ½ch As you can see in (45.4), we can get the ratio of the amplitude of element m and n, so that if the amplitude of excitation coefficient is set to the inverse ratio of this, the gain error will be calibrated.

45.2.2 Calibration method with ground station As mentioned previously, in on-orbit verification with the engineering test satellite-9, we will assign the function of calibration to both satellite and terrestrial systems because of the limitation of implementation on a digital processor. Figure 45.3 shows the block diagram of the repeater and conceptual diagram of the calibration method with the ground station. At first, the calibration signal is transmitted from the grand station and the repeater receives this signal with some antenna elements. The ground station transmitting calibration signal has to be set to the point where antenna

564

Advances in communications satellite systems 2: ICSSC-2019 Analog components

Analog components

Digital components

#1

#4 #5

LNA D/C

ADC DEMUX

#6 #7 #8

2. Calculating cross-correlation among elements

∑∑

Multiplied by excitation coeff.

#3

Multiplied by excitation coeff. & calculating cross-correlation

#2

励 振 係 数

TWTA U/C DAC MUX

5. Command

3. Telemetry D/C: Down converter U/C: Up converter

1. Transmitting calibration signal 4. Calculating corrected excitation coeff.

Figure 45.3 Block diagram of a repeater and conceptual diagram of the calibration method with the ground station

elements under calibration have antenna gain enough to get power flux density to communicate. In addition to this, a wideband signal is better for calibration signals in order to get the error of all subchannels at once. The received calibration signal passes through the paths of each element and divides into subchannel with DEMUX after analog digital convert. The calibration signals of all the elements are transmitted from DEMUX to DBFP. DBFP can calculate cross-correlation vector on each subchannel between two elements on the basis of a specific element. This crosscorrelation vector is transmitted to the satellite control center with satellite telemetry system. In the satellite control center, the correction coefficient for amplitude and phase errors among elements is calculated based on the cross-correlation vector received from the satellite. In addition to this, the satellite control center plays a role of determining the coverage of the receiving beam and calculating the excitation coefficient of each receiving antenna element. Therefore, new excitation coefficients can be calculated by adding correction coefficients and excitation coefficients in the satellite control center. At last, these coefficients are set to DBFP of satellite repeater by a command transmitted from the satellite control center and the intended receiving beam is shaped. As you can see from Figure 45.1, the signal for calibration transmitted from one specific location can be input not more than three or four elements. Therefore, the transmission station for calibration signal will be needed

Calibrating digital beam forming in engineering test satellite-9

565

more than three and we are planning to join the result of each station in order to calibrate over the eight elements.

45.3 Conclusion In DBF system, the calibration of each element is important and we propose a method of calibration using digital processing on satellite and ground stations. The calibration signal is transmitted with the ground station and the error of gain, phase, and group delay can be determined with cross-correlation vector calculated in the digital processor on the satellite. In the ground station, the correction coefficient will be calculated based on the cross-correlation vector and the error is canceled by setting the excitation coefficient, which is added correction coefficient, to the satellite. We are now verifying the detecting accuracy of this method with simulation. In the future, we will test this calibration method on the repeater system with the ground test. After launch, we will test coverage change with DBF and total communication performance with the terrestrial system.

Acknowledgments This study is conducted based on the R&D project entitled “Research and Development of Ka-band Wideband Digital Beam Forming for Efficient Frequency Use” from the Ministry of Internal Affairs and Communications. Special thanks to all concerned.

References [1] Miura, A., Morikawa, E., Yoshimura N., et al. “On preliminary design of bandwidth-on-demand high throughput satellite communications system technology.” 24th Ka and Broadband Communications Conference. Niagara Falls, Canada. 2018 [2] Sakai, E., Ono, H., Sunaga, T., et al. “Research tasks and plan for the R&D project.” 24th Ka and Broadband Communications Conference. Niagara Falls, Canada. October 2018. [3] Sakai, E., Inasawa, Y., Kusano, M., et al. “Current design of Ka-band digital beam forming technology for the high throughput satellite communications system.” 25th Ka and Broadband Communications Conference. Sorrento, Italy. 2019. [4] Fulton, C., Yeary, M., Thompson, D., et al. “Digital phased arrays: Challenges and opportunities.” Proceedings of IEEE. 2016; 104(3): 487–503. [5] Mano, S. and Katagi, T. “A method for measuring amplitude and phase of each radiating element of a phased array antenna.” IEICE Transactions B. 1982; J65-B(5): 555–560.

This page intentionally left blank

Chapter 46

Beam pattern optimization based on up/downlink information for multibeam satellite communication systems Kazuma Kaneko1, Shigenori Tani1, Shigeru Uchida1 and Hiroshi Aruga1

To reduce the cost of satellite communications, it is important to minimize the number of beams by arranging the beams so as to suit the traffic load. Although a nonuniform beam pattern optimization method controlled by digital beam forming (DBF) has been proposed, its metric reflects only a single link’s information. This conventional method using a single link’s information cannot optimize the other links, thus it has limited ability to reduce the number of beams. Therefore, we propose a beam pattern optimization method which matches the traffic load in both the up and downlinks to reduce the required number of beams. The effectiveness of the proposed method is verified by simulation. Key Words: beam pattern; HTS; digital beam forming

46.1 Introduction Since satellite communication systems are excellent for supporting disaster recovery, they are attracting attention as important means of communication at the time of a natural or other disaster. In addition, they are widely used because of their wide coverage and their ability to provide communications even at sea and in the air, where terrestrial networks cannot be used. In recent years, Internet traffic has been on the rise worldwide, and rich content is required at sea and in the air, for which reason high capacity satellite communications is desired. From such demand, high throughput satellite (HTS) that uses the Ka-band and has a large number of beams to realize high capacity satellite communications has appeared. HTS is also attracting attention for backhaul for terrestrial networks and is expected to play an important role in realizing 5G in the near future [1]. 1

Information Technology R&D Centre, Mitsubishi Electric Corporation, Kamakura, Japan

568

Advances in communications satellite systems 2: ICSSC-2019

In the 5G era, it is expected that the traffic in the terrestrial network will increase rapidly, so HTS also needs to provide increasingly high capacity. HTS provides a communication environment with multiple beams, using different frequencies for each beam. However, it also reuses the same frequencies on those multiple beams whose spatial separation is sufficiently large for mutual interference to be negligible. This yields high capacity by making the most effective use of the limited spectrum. By reducing the beam diameter and increasing the number of beams, it is possible to increase the frequency reuse rate and further increase the capacity. The number of HTS beams is increasing year by year, and satellites in the 200–300 beams class are becoming popular. In addition, very high throughput satellite (VHTS), which enables mass communication exceeding the capabilities of HTS, has also appeared [2]. However, as the number of beams increases, the cost also increases, which causes a trade-off between capacity and cost. Therefore, increasing the capacity while holding down the number of beams is an important issue for HTS, and the use of a digital channelizer for flexible utilization of the spectrum together with digital beam forming (DBF) to adjust the beam positions and shapes is being considered [3]. Moreover, beam pattern optimization is also effective for cost reduction. In the case of 5G networks, it is not clear where the traffic will increase, and the distribution of traffic generation may vary. If there is a geographical variation in the traffic, the load on each beam will not be even and the traffic on each beam can become sparse or dense. Therefore, it is inefficient to increase the number of beams uniformly over the entire service area. By increasing the number of beams only in areas where the traffic is dense, it is possible to increase the capacity while holding down the cost. In this research, in order to reduce costs, we propose a method of generating a beam pattern that satisfies the requirement with a small number of beams by placing the beams according to the traffic load, and confirm its effectiveness by simulation. The remainder of this chapter is organized as follows. In Section 46.2, the previous research on resource management in HTS is introduced, and the beam pattern generation technique is described. The problem with the conventional beam pattern generation method is also described in Section 46.2. The proposed method for solving this problem is described in Section 46.3. The results of simulations of the proposed method, the conventional beam pattern generation method, and the uniform beam pattern generation method are shown in Section 46.4. Finally, this chapter is brought to a conclusion in Section 46.5.

46.2 Related research on HTS resource management In order to reduce the cost of HTS, it is important to use the communication resources efficiently. There are four communication resources in HTS: frequency, power, time, and space. First, for the management of frequency resources, a channel assignment method using a digital channelizer that varies the frequency assignment for each beam is considered [4]. In conventional

Beam pattern optimization based on up/downlink information

569

satellite communication systems, frequency resources with the same bandwidth have been allocated to each beam, and adjacent beams do not use the same frequency resource to prevent interference. However, since the traffic can differ greatly depending on the beam, this method has the problem that the frequency resource allocated is excessive for a beam with low traffic, while the frequency resource allocated is insufficient for a beam with high traffic. In the case of frequency channel assignment using a digital channelizer, the frequency resources are finely divided, and numerous channels are assigned to beams with a large amount of traffic so that the load is equalized across the beams. This is intended to realize a large capacity without increasing the number of beams. For power resource management, a method of varying the power in each beam using a multiport amplifier is considered. For time resource management, a technique called beam hopping is being studied [5]. Beam hopping is a method of selecting the beams to be used for every time slot instead of using all the beams simultaneously. Since it is only necessary for the satellite to have enough amplifiers for the number of beams to be used at any one time, the cost of the on-board equipment can be reduced. Finally, the beam pattern generation method previously mentioned in Section 46.1 is being considered for spatial resource management. In [6], a beam pattern method using beams with two beam diameters is discussed, which enables the effective use of the beams by using different beam diameters over land and sea. However, with regard to the frequency allocation, it is not possible to break the periodicity of the beam pattern because conventional four-color repetition is used, and the throughput improvement is limited because there is a considerable restriction on the beam pattern. In [7], the greedy method is used to optimize the beam pattern, and in [8], the beam arrangement is optimized using mixed integer linear programming. However, in any of these methods, the beam pattern optimization is based only on the information from either the downlink or the uplink. In the case of optimization in which only one link is considered, there is a possibility that the beam pattern may not be optimum for the other link. In the Kaband used for HTS, the frequencies used in the uplink and downlink are separated by about 10 GHz. Therefore, the antenna radiation patterns are different between the two links, whence the beam diameter also differs. Since the satellite-mounted antenna assumed in this research is shared by transmission and reception, the center of the beam is common to the two links. When the beam pattern optimization is based on uplink information, although the beam pattern is optimal for the uplink, the beams are too close to each other for the downlink, so the throughput is reduced due to interference. On the other hand, when the beam pattern optimization is based on downlink information, the beams are unnecessarily far apart for the uplink. In the uplink, the flux density weakens rapidly as the distance from the beam center increases, and the throughput decreases due to gain reduction if the distance between the beams is too large. Therefore, a method to determine the beam pattern considering both links instead of one or the other is required. In this research, we propose a method to optimize the beam pattern using information for both the uplink and the downlink.

570

Advances in communications satellite systems 2: ICSSC-2019

46.3 Beam pattern optimization 46.3.1 Genetic algorithm In this study, the center of the beam is determined using a genetic algorithm (GA). Figure 46.1 presents an overview of GAs. In a GA, the initial individuals generated are scored against the objective function (fitness), and only individuals with a high score are left as elite. After that, either crossover, which mixes the gene patterns of the elites, or mutation, which randomly changes the genes of the elites, is executed. Elite selection is performed once again, including the original elite individuals and individuals generated by crossover and mutation. This is an algorithm that produces individuals with high fitness by repeatedly executing this process. In this study, the average of the uplink and downlink throughputs is regarded as the fitness function of the GA, and the individual (beam pattern) that maximizes the fitness function is determined. Figure 46.2 shows the flow chart for the GA. In this method, first, beam center positions for all the beams are selected randomly, and the throughputs of both the uplink and downlink are calculated. The details of the throughput calculation are described in Section 46.3.2. After the throughputs of the two links are obtained, the average value is taken as the fitness function of the individuals. The selection of beam centers and the throughput calculations are repeated for the number of individuals, and those individuals with large fitness functions are selected as elite from among all the individuals. The number of elites can be calculated by multiplying the total number of individuals by the elite rate. To improve the fitness function, new generations of individuals are created by crossover or mutation from the selected elites. In this study, we focus on the amount of traffic able to be carried on a beam to improve the fitness function. If the traffic able to be accommodated is small, this means that the beam has been located in an area where the traffic 90 points Elites

Individuals

80 points 60 points 30 points 10 points Number of beams mutation

Crossover

Exchange a gene randomly

Generate new individuals from elites

Figure 46.1 Overview of genetic algorithms

Beam pattern optimization based on up/downlink information

571

Start mi = 0, g = 0, m = 0 mi + = 1 Select NB beam centers randomly Calculate throughput of downlink Calculate throughput of uplink Calculate average of above two values Yes mi > > > > 29  25log10 y; 1:5  y  7 > > < (48.1) 8; 7  y  9:2 GðyÞ ¼ > > >   > 32  25log10 y; 9:2  y  48 > > > : 10; 48 y  180 where Gmax is the peak gain of the antenna. The gain as a function of spherical angle y can also be expressed in terms of azimuth, f, and elevation, q as provided in [6]: cos y ¼ cos ðqÞ cos ðfÞ Substituting (48.2) in (48.1)   ðq; fÞ ¼ Gmax  cos1 ðcos ðqÞcos ðfÞÞ

(48.2)

(48.3)

48.2.1.2 Terrestrial fixed services The terrestrial services use phased array antennas that can have a variable radiation pattern based on the excitation level of each of the individual elements. The standard models used in this chapter are derived from 3GPP specifications [7,8]. For simplicity, the analysis here assumes uniform weighting factor for all the elements of the antenna array [9] instead of a beamforming design with optimized weight factor. Each antenna element is composed of horizontal (azimuth) and vertical (elevation) gain patterns as indicated below:    q  90 ; SLAv (48.4) AE;V ðqÞ ¼  min q3dB ●

q [ [0,180], where 90 represents the direction perpendicular to the array antenna aperture,

600 ● ●

Advances in communications satellite systems 2: ICSSC-2019 SLAv is the side-lobe attenuation limit, typically set to 30 dB, and q3dB is the vertical 3dB beam-width that is assumed to be 65 . ( AE;H ðfÞ ¼ min

● ● ●

f f3dB

)

2 ; Am

(48.5)

f [ [180,180], Am is the front-to-back ratio, typically set to 30 dB, and f3dB is the horizontal 3 dB beam-width that is assumed to be 65 . Combining (48.4) and (48.5), the element radiation pattern is given by    ðq; fÞ ¼ GE  min  AE ðqÞ þ AE;H ðfÞ ; Am (48.6)

where GE is the maximum directional gain of the element (in dB), which is assumed to be 8 dBi. Using the element gain pattern in (48.6), the composite array radiation pattern for multiple columns is given by [8]: Gðq; fÞ ¼ GE ðq; fÞ þ AFðq; fÞ

(48.7)

AF is the array factor that can be defined in terms of the steering matrix and weighting factor as   X 

2

NH XNv

w :v  1 (48.8) AFðq; fÞ ¼ 10log10 1 þ r m¼1

m¼1 m;n m;n where ●

●

●

r ([ [0,1]) is the correlation coefficient between the different antenna elements, which for simplicity is assumed to equal 1, vm;n ¼ expði:2pððn  1Þ dlv cos ðqÞ þ ðm  1Þ dlH sin ðqÞsin ðfÞÞÞ are the steering matrix components, 1 wm;n ¼ pffiffiffiffiffiffiffiffiffi expði:2pððn  1Þ dlv sin ðqetilt Þ  ðm  1Þ dlH cos ðqetilt Þsin ðfescan ÞÞÞ N N H

V

are the weighting matrix components, ●

●

● ●

dv and dH correspond to the spacing between antenna elements in the vertical and horizontal direction, qetilt is the electrical down-tilt steering and fescan is the electrical horizontal steering, m ¼ 1,2, . . . NH is the index of array element in the horizontal direction, and n ¼ 1,2, . . . NV is the index of array element in the vertical direction.

48.2.1.3

Field pattern

In order to integrate the array pattern into the channel model, the polarimetric antenna response at the transmitter and the receiver needs to be calculated. The details that include the polarization terms are derived and discussed in detail in [3].

Impact of antenna and propagation models on coexistence

601

The simplified relationship between the field pattern and the three-dimensional (3D) antenna gain pattern provided in (48.3) and (48.7) with a purely vertically polarized antenna is given below [9]:  Fq ðq; fð¼ Gðq; fÞ (48.9) Fðq; fÞ ¼ Ff ðq; fÞ ¼ 0

48.2.2 Propagation models The path loss used to calculate the desired and interference signal power values is modeled using 3GPP and ITU recommendations [7,10]. It ranges from simplistic free space path loss to more intricate rural and urban deployment channel models that include both line of sight (LOS) and nonline of sight (NLOS) components.

48.2.2.1 Free-space path loss The basic free space path loss is given by the relationship in (48.10). Additional location-specific attenuation incurred due to multipath environment, diffraction, gaseous absorption, ducting/layer-reflection, refraction, troposcatter, and clutter loss can be added to (48.10) as described in [10] PL ðd; f Þ ¼ 32:45 þ 20log10 d þ 20log10 f

(48.10)

where d is the distance in meters and f is the frequency in GHz.

48.2.2.2 3GPP rural and urban LOS/NLOS models The path loss models for rural macro deployment scenarios is summarized in Tables 7.4.1-1 and 7.4.2-1 in [7], based on representative measurement campaigns in the frequency range 0.5 GHz to 100 GHz. The values from these tables can be used instead of the fundamental distance-based calculation of PL in (48.10). It is to be noted that the NLOS path loss for distances between 10 m and 5 km is given by PNLOS ¼ maxðPLOS ; P0NLOS Þ

(48.11)

where PLOS is the LOS path loss and P0NLOS is the NLOS path loss.

48.2.3 Interference calculation Using (48.9) and (48.10), the complex-valued static channel coefficients between a transmit and receive antenna can be approximated as pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2p (48.12) gr;t ðq; f; d; f Þ ¼ PL ðd; f ÞF r ðq; fÞT Mðq; fÞFt ðq; fÞej l d where M(q, f) is the polarization coupling matrix, and the subscripts t and r refer to transmitter and receiver, respectively. As elaborated in [3], the values q and f in (48.12) above need to be transformed from the local coordinate system

602

Advances in communications satellite systems 2: ICSSC-2019

of the antenna frame to arrival and departure angles in global Cartesian coordinates. The aggregate received power can then be calculated as XNT P jg ðq; f; d; f Þj2 (48.13) Pr ðq; f; d; f Þ ¼ t¼1 t r;t where Pt is the transmit power in mW and gr,t(q, f, d, f ) is the individual path channel coefficients from (48.12) in linear units.

48.2.3.1

FSS downlink interference

The desired downlink signal at the ES can experience interference from both BSs and UEs operating in time division duplex (TDD) mode. Using the relationship in (48.13), the desired signal power at the ES receiver from the satellite can be expressed as Ces ðq; f; d; f Þ ¼ Pt jges;t ðq; f; d; f Þj2 and the aggregate interference power can be expressed as ( PNbs 2 t¼1 Pt jges;t ðq; f; d; f Þj ; 5G downlink Ies ðq; f; d; f Þ ¼ PNue 2 t¼1 Pt jges;t ðq; f; d; f Þj ; 5G uplink

(48.14)

(48.15)

where Nbs and Nue are the numbers of base stations and user terminals in the given deployment scenario.

48.2.3.2

FSS uplink interference

In the opposite direction, when the ES transmits signals toward the satellite, the desired signal and interference power at each of the 5G nodes can be expressed as XNbs P jg ðq; f; d; f Þj2 (48.16) Cue ðq; f; d; f Þ ¼ t¼1 t ue;t XNue Cbs ðq; f; d; f Þ ¼ P jg ðq; f; d; f Þj2 (48.17) t¼1 t bs;t Iue ðq; f; d; f Þ ¼ Pt¼es jgue;t¼es ðq; f; d; f Þj2

(48.18)

Ibs ðq; f; d; f Þ ¼ Pt¼es jgbs;t¼es ðq; f; d; f Þj2

(48.19)

To summarize, the fundamental parameters that influence the aggregate interference at any receiver are: (1) frequency, (2) distance between transmit and receive antennas, (3) angles of arrival and angles of departure in both elevation and azimuth directions, and (4) transmit power. The relationship in (48.14)–(48.19) captures the spatial dependencies and not the temporal variations induced due to mobility of satellite or power control/beamforming algorithms at the transmitters. Accordingly, the simulation setup and results discussed in this chapter captures the static variations with fixed nodes in a given deployment scenario.

48.2.3.3

Interference threshold

There are different types of interference thresholds as discussed in [11]. These include ratios involving interference, noise, and carrier signal powers such as

Impact of antenna and propagation models on coexistence

603

I/N , C/I, C/(N þ I), field strength such as power flux density (PFD) and equivalent power flux density (EPFD), or network performance criteria such as availability, reduction in coverage, range, capacity, channel sharing ratios, and derived probabilities of blocking. While some of these thresholds can serve as a hard limit, other values can be used as gating criteria for more detailed analysis for coordination and sharing. Furthermore, various national regulatory bodies and the ITU can set threshold limits to ensure a certain level of compatibility prior to deployment. For purposes of analysis in this chapter, the aggregate interference power expressed as a power spectral density (PSD) value in dBm (or dBW)/MHz has been used to evaluate compatibility for cochannel operations. As an example, according to Title 47 CFR 96.17 [5]: “The aggregate passband radiofrequency (RF) power spectral density at the output of a reference RF filter and antenna at the location of an FSS earth station operating in the 3600-3700 MHz band, produced by emissions from all co-channel CBSDs (within 150 km) operating in the Citizens Band Radio Service shall not exceed a median root mean square (RMS) value of 129 dBm/MHz.” A more detailed explanation for calculating the permissible interference power is given in Appendix 7 of ITU Radio Regulations (ITU-RR) [12]. Accordingly, the threshold for a receiving ES in Ka-band can be deduced from values provided in Table 8d in Appendix 7 of ITU-RR. These values are based on certain assumptions of the receiver noise temperature, link performance margin, and number of short-term entries, as applicable for short-term and long-term threshold calculations.

48.3 Simulation setup The main purpose of the simulation in this study was to demonstrate the impact of various system parameters on the downlink interference experienced at an FSS ES antenna due to transmissions from 5G BSs and UEs. As derived in (48.15), the aggregate interference power is a multivariate function that is influenced by the Table 48.1 General simulation parameters Parameter

C-band

Ka-band

Frequency (GHz) Channel bandwidth (MHz) BS antenna height (m) BS transmit power (dBm) BS electronic downtilt (deg) UE antenna height (m) UE transmit power (dBm) FSS ES antenna height (m) BS antenna pattern FSS ES antenna pattern

3.4 20 35 49 1–6 1.5 23–26 5 3GPP TR 38.901 Title 47 CFR 25.209(a)(1)

26 20 35 55 1–6 1.5 40–55 5 3GPP TR 38.901 Title 47 CFR 25.209(a)(1)

604

Advances in communications satellite systems 2: ICSSC-2019

deployment geometry. The interference at BSs and UEs due to uplink transmissions from FSS ES will be considered in future work. The 3D channel modeling framework developed in [3] is used to generate power maps associated with desired and interfering signals. Some of the basic simulation parameters are listed in Table 48.1. The antenna patterns corresponding to (48.1) and (48.8) are depicted in Figures 48.2 and 48.3, respectively. The impact of the distance between the transmitter and the receiver on the path loss associated with the different 3GPP propagation models is shown in Figure 48.4. As expected from the basic relationship in (48.10), the path loss is

20

Power

0 –20 –40 –60 –100 0 Elevation

100 200

100

–100

0

–200

Azimuth

Figure 48.2 Composite BS antenna pattern with ten vertical elements and element gain of 8 dBi

60 Power

40 20 0 –20 –100 –50 0 Elevation

0

50 100 200

100

–100

Azimuth

Figure 48.3 Composite FSS ES antenna pattern with 10 elevation and peak gain of 52.6 dBi. Adapted from [13]

Impact of antenna and propagation models on coexistence

605

240 220

Path loss in dB

200 180 160 140 3GPP-38.901-RMa-LOS 3GPP-38.901-RMa-NLOS 3GPP-38.901-UMa-LOS 3GPP-38.901-UMa-NLOS 3GPP-38.901-UMi-LOS 3GPP-38.901-UMi-NLOS Freespace

120 100 80 0

10

20

(a)

30 40 Distance in km

50

60

70

240 220 200

Path loss in dB

180 160 140 120 3GPP-38.901-UMa-NLOS-3.4 GHz Freespace-3.4 GHz 3GPP-38.901-UMa-NLOS-18 GHz Freespace-18 GHz 3GPP-38.901-UMa-NLOS-26 GHz Freespace-26 GHz 3GPP-38.901-UMa-NLOS-37 GHz Freespace-37 GHz

100 80 60

0

(b)

10

20

30 40 Distance in km

50

60

70

Figure 48.4 (a) Path loss variation for different propagation models at 26 GHz and (b) path loss variation at different frequencies

minimal for freespace propagation model and highest for NLOS models associated with rural macro (RMa), urban macro (UMa), and urban micro (UMi) deployment scenarios. The increase in path loss at higher frequencies can be verified by the graphs in Figure 48.4(b).

606

Advances in communications satellite systems 2: ICSSC-2019

y-coord (km)

North→

60 40 20 Area in which predicted signal strength into FSS exceeds threshold (–116 dBm/20 MHz)

0 –20 –40 –60 –60

Color key of P2MP signal strength received by FSS (in dBm/20 MHz) –40

–20

20 0 x-coord (km)

40

60 East→

Figure 48.5 Interference power map with a single BS, LOS propagation at 3.4 GHz and 0 dBi sidelobe of FSS-ES antenna

The interference power maps generated and discussed in Section 48.4 use a combination of RMa and freespace propagation models with the receive antenna at FSS ES modelled as an isotropic (0 dBi sidelobe), or a directional antenna as shown in Figure 48.3.

48.4 Interference power maps Using the system models in Section 48.2 and the simulation setup outlined in Section 48.3, the interference power experienced by a notional receive antenna at every point in a given region of interest is calculated. In other words, each point in the grid represents potential location of FSS ES. The received power at each point is calculated using arrival and departure angles between the interferer and the receiver in 3D coordinates. The resulting interference power map is plotted delineating regions where aggregate power due to simultaneous transmissions from interferers exceeds the interference threshold.

48.4.1 Single interferer A highly conservative analysis using freespace path loss model and an isotropic FSS ES antenna is shown in Figure 48.5. The region where the interference is below the threshold is assigned a dark blue color. The BS has two active sectors and the interference is maximum along the pointing direction of the boresight of the composite array pattern as indicated in the figure. An isotropic receive antenna located in this direction will intercept maximum unwanted signal power, whereas when it is located in the direction of the backlobe of the BS antenna, it will intercept minimum

Impact of antenna and propagation models on coexistence

607

North→

15 10

y-coord in (km)

5 0 –5

–10 –15 –15

–10

–5

5 0 x-coord (km)

10

15 East→

–15

–10

–5

5 0 x-coord (km)

10

15 East→

y-coord in (km)

North→

(a)

15 10 5 0 –5

–10 –15

(b)

Figure 48.6 Interference power map with RMa-NLOS propagation at 3.4 GHz, 0 dBi sidelobe of FSS-ES antenna, and a single BS with electronic downtilt of (a) 1 and (b) 6 interference power from BS. Hence, the exclusion zone is uneven, with distances exceeding 40 km in one direction and less than 20 km in the opposite direction. A similar analysis with a single BS but with RMa-NLOS propagation model is shown in Figure 48.6. The exclusion zone decreases when the electronic downtilt of BS is increased from 1 to 6 . The power map plot with a directional FSS antenna fixed at 10 elevation and 0 azimuth (pointing due east/right) is denoted in

North→

Advances in communications satellite systems 2: ICSSC-2019 15

y-coord in (km)

608

5

10

0 –5

–10 –15 –15

–10

–5

0 5 x-coord (km)

10

–15

–10

–5

0 5 x-coord (km)

10

North→

15

y-coord in (km)

(a)

5

15 East→

10

0 –5

–10 –15

(b)

15 East→

Figure 48.7 Interference power map with single BS, RMa-NLOS propagation at 3.4 GHz, and directional antenna of FSS-ES with receiver height of (a) 5 m and (b) 1.5 m

Figure 48.7. The decrease in exclusion zone between Figure 48.7(a) and (b) can be attributed to decrease in receiver height. The directionality of the receive antenna in general reduces the total exclusion zone area compared to an isotropic antenna, with maximum distance measured along the boresight of the directional antenna (pointing due east). The analysis indicates that the coordination distances with more realistic parameter considerations can be less than 7 km.

Impact of antenna and propagation models on coexistence

609

48.4.2 Multiple BS transmissions

North→

A single sector illumination of each of the BSs arranged in a hexagonal grid and transmitting away from the central region is shown in Figure 48.8. The BS antenna electronic downtilt is set to 1 . In Figure 48.9, the FSS receive antenna is fixed to a

20 15

y-coord (km)

10 5 0 –5 –10 –15 –20 –20

–15

–10

–5

5 0 x-coord (km)

10

15

20 East→

North→

Figure 48.8 Interference power map with BSs arranged in a hexagonal grid, RMa-NLOS propagation at 3.4 GHz, and 0 dBi sidelobe of FSS-ES antenna with electronic downtilt of 1

30

y-coord (km)

20 10 0 –10 –20 –30 –30

–20

–10

10 0 x-coord (km)

20

30 East→

Figure 48.9 Interference power map with BSs arranged in a hexagonal grid, RMa-NLOS propagation at 3.4 GHz, and directional antenna at FSS-ES receiver

610

Advances in communications satellite systems 2: ICSSC-2019 6

4

y-coord (km)

2

0

–2

–4

–6 –6

–4

–2

0 x-coord (km)

2

4

6

–4

–2

0 x-coord (km)

2

4

6

(a) 6 4

y-coord (km)

2 0

–2

–4 –6 –6 (b)

Figure 48.10 Interference power map with RMa-NLOS propagation at 26 GHz, 0 dBi sidelobe of FSS-ES antenna with electronic downtilt of 1 , and interference threshold of (a) 97 dBm/20 MHz and (b) 110 dBm/20 MHz 10 elevation and 0 azimuth (pointing due east/right). Just as in the previous case, all the transmit BSs have a single sector illumination and arranged in a hexagonal grid, with the sectors pointing away from the center. This configuration depicts the impact of the geometry with respect to the transmit and receive antenna pointing. The exclusion zone is characterized by larger separation distances when the boresight of the directional antenna shown in Figure 48.3 is aligned with the BS

6

4

4

4

2

2

2

0

y-coord (km)

6

y-coord (km)

y-coord (km)

6

0

0

–2

–2

–2

–4

–4

–4

–6 –6

–4

–2

2 0 x-coord (km)

4

6

–6 –6

–4

–2

2 0 x-coord (km)

4

6

–6 –6

–4

–2

2 0 x-coord (km)

4

Figure 48.11 Variation in interference power maps at 3.4 GHz due to change in UE transmit power: 23 dBm (left) and 26 dBm (center, right)

6

6

6

4

4

4

2 0 –2

(a)

0 –2 –4

–4 –6 –6

2

y-coord (km)

y-coord (km)

y-coord (km)

6

–4

–2

2 0 x-coord (km)

4

6

–6 –6 (b)

2 0 –2 –4

–4

–2

2 0 x-coord (km)

4

6

–6 –6 (c)

–4

–2

2 0 x-coord (km)

4

6

Figure 48.12 Variation in interference power contours at 26 GHz due to change in number of simultaneous UE transmissions: 20 (a), 100 (b), and 200 (c)

Impact of antenna and propagation models on coexistence

613

sectorial antenna transmissions. Careful planning during coordination process can enable coexistence of the two services without sacrificing coverage. The impact of transmission at Ka-band from a hexagonal grid of BSs is depicted in Figure 48.10. In accordance with trends exhibited at higher frequencies, the coordination zones are significantly smaller compared to that of C-band values shown in Figure 48.8. Additionally, it may be noted that the interference threshold value impacts the area of coordination zone. As the interference threshold decreases or in other words, when the permissible level of interference is lowered, the coordination distances between the FSS ES and BS/UEs increases.

48.4.3 Multiple UE transmissions The impact of uplink transmissions at Ka-band from a random distribution of UEs is depicted in Figures 48.11 and 48.12. The coordination areas increase when the transmit power or the number of transmitters increase [7,14].

48.5 Conclusion The above analysis serves as a baseline for evaluating various interference mitigation strategies, some of which have been discussed in [11]. An accurate simulation of the antenna and propagation models can aid in optimization of exclusion, coordination, and interference zones. Although the scenarios presented in this chapter represent fixed services and static nodes, they provide a good insight into the likely trends exhibited due to temporal changes in transmission and channel parameters. The simulations here can be expanded to include waveform and medium access features such as those being considered in recent 3GPP studies on non-terrestrialnetworks (NTN) [1]. The structure of waveforms plays a key role in predicting the interference levels as well as associated metrics useful in devising appropriate coexistence strategies. It is envisaged that the analysis here will provide a good basis to investigate and recommend enhancements to these wide-band waveforms.

References [1] 3GPP TR 38.811 v15.1.0. Study on new radio (NR) to support non-terrestrial networks (Release 15). 2019-06. [2] FCC 16-89. Report and order and further notice of proposed rulemaking. July 14, 2016. [3] Jaeckel, S., Rashkowski, L., Bo¨rner, K., Thiele, L., Burkhardt, F. and Eberlein, E. “QuaDRiGa – Quasi deterministic radio channel generator, user manual and documentation.” Fraunhofer Heinrich Hertz Institute, Tech. Rep. v2.0.0.0. 2017. [4] ITU-R S.465-5. Reference earth-station radiation pattern for use in coordination and interference assessment in the frequency range from 2 to about 30 GHz.

614

Advances in communications satellite systems 2: ICSSC-2019

[5] FCC Rules and Regulations. Title 47 – Telecommunication. [6] ITU-R SM.2028. Monte Carlo simulation methodology for the use in sharing and compatibility studies between different radio services or systems. 2017-06. [7] 3GPP TR 38.901 v14.1.0. Study on channel model for frequencies from 0.5 to 100 GHz (Release 14). 2017-06. [8] 3GPP TR 37.840 v12.1.0. Study of radio frequency (RF) and electromagnetic compatibility (EMC) requirements for active antenna array system (AAS) base station (Release 12). 2013-12. [9] Rebato, M., Resteghini, L., Mazzucco, C. and Zorzi, M. “Study of realistic antenna patterns in 5G mmWave cellular scenarios.” 2018 IEEE International Conference on Communications (ICC). Kansas City, MO. 2018. pp. 1–6. [10] ITU-R P.452-16. Prediction procedure for the evaluation of interference between stations on the surface of the Earth at frequencies above about 0.1 GHz. 2015-07. [11] Pahl, J. Interference analysis: Modelling radio systems for spectrum management. Wiley, Hoboken, NJ, USA; 2016. [12] ITU 2016. Radio Regulations. 2016 ed. Geneva. [13] GN Docket No.18-122. Further technical statement, C-band alliance. March 04, 2019. [14] 3GPP TR 38.101 v16.0.0. User equipment (UE) radio transmission and reception: Part 2: Range 2 Standalone (Release 16). 2019-06.

Chapter 49

Integrated space-enabled hybrid 5G-V2X communications link modeling Solomon Udeshi1, Mfonobong Uko1, Muazzam Zafar1, Arslan Altaf 1, Bamidele Adebisi1 and Sunday Ekpo1

Vehicle-to-everything (V2X) communication is the next innovative technology to transform the automotive industry. It implements the power of Internet of things (IoTs) connectivity and processing to improve the efficiency, performance, and safety of future vehicles. V2X enables vehicles to have a high awareness of the surrounding environmental factors, infrastructures, and other vehicles through the advanced sensing and communication technologies in place. Current communication platforms in place are not able to fully support the transmission of high priority safety critical data required to provide these services. Individually, the two main V2X architectures (DSRC and C-V2X) have limitations which inhibit them from fully supporting the V2X platform. As such, this chapter proposes the integration of 5G technology, which supports all types of communication technologies through the integration of reconfigurable devices and offers higher data rates and bandwidths than any preceding platform. The architecture proposed in the chapter is composed of three layers: DSR V2X for direct short-range communication; cellular V2X layer which will combine 5G cellular platform to provide backup communication when DSR V2X layer fails and multimedia capabilities; and satellite (SAT)-V2X which provides extra backhaul connectivity when both DSR and cellular V2X is unavailable. The system design, modeling, and simulation results yield a more reliable and sustainable V2X capacity for critical real-time vehicular communication applications. Key Words: V2X; 5G communications; DSRC; link budget

1

Department of Engineering, Manchester Metropolitan University, Manchester, UK

616

Advances in communications satellite systems 2: ICSSC-2019

49.1 Introduction The automotive industry has experienced significant innovation through the integration of information and communication technology (ICT) in vehicle subsystems. Modern vehicles contain a sophisticated network of electronic control units (ECUs), which constantly gather, process, and share data to improve the vehicles efficiency, safety, and performance. The next automotive advancement targets the inclusion of Internet of things (IoTs) services, enabling vehicles to obtain a greater awareness of its surrounds, producing applications such as forward collision warning, road works warning, platooning, etc. To support this, a robust vehicular communication system that can incorporate communication between vehicle to vehicle (V2V), vehicle to pedestrian (V2P), and vehicle to infrastructure (V2I) is required. This is known as vehicle to everything (V2X) communication. The radio access technologies (RAT); dedicated short range communication (DSRC) and cellular V2X (C-V2X) communication are two of the most prominent supporting access technologies for V2X. DSRC has been developed to enable direct low-latency communications between DSRC devices such as onboard units (OBUs) inside vehicles, roadside units (RSUs) located on roads, or handheld devices carried by pedestrians [1]. DRSC communication has been allocated the 5.9 GHz band, as per the IEEE 802.11p standard [2]. Its non-line-of-sight transmission makes it suitable for dynamic environments where obstacles quickly appear in the communication channels. Despite these ideal properties, DSRC is prone to link quality degradation in areas with high vehicle density which have a higher rate of channel collisions. C-V2X uses cellular LTE infrastructure for V2X communications, hence it can provide wide area coverage and is favorable for high bandwidth demand applications that require high data rate and reliability. C-V2X standards for LTE have been released by 3GPP [3]. C-V2X in its current form has a major drawback, as it experiences high end-to-end latency due to the long transition time interval. Furthermore, the current LTE platform is facing limitations in supporting the increasing bandwidth demands of other applications. These drawbacks can significantly hinder the transmission of real-time safety critical information, which is the key requirement of a V2X system. A hybrid communication platform that combines DSRC and C-V2X systems will be a more suited to cater for the dynamic and sophisticated nature of V2X communications. The key requirements for a V2X platform are: low latency, high bandwidth, and robust connectivity. In a hybrid DSRC-cellular platform, the drawbacks of the individual vehicular communication network are overcome by the strengths of the other to meet the key requirements of V2X communication. For instance, the sort range coverage of DSRC communication is not suitable for highly dynamic environmental conditions such as motorway driving, as vehicular (OBUs) would be disconnecting and reconnecting to RSUs at rapid rates that fragment the DSRC link. The cellular network can be used to patch the fragmented DSRC link as vehicles transition between nodes, making the V2X platform more robust and

Integrated space-enabled hybrid 5G-V2X communications link modeling

617

reliable. Cellular LTE technology in its current form does not have the infrastructure to fully support V2X communication. As such, research efforts are applied to develop V2X communication standards for the immerging 5G network, which promises greater bandwidths, higher data rates, and lower latency to mission-critical communications and massive IOT services. One of the key aspects of 5G is that it is the overall communications platform layer which supports all existing communication technologies such as sub-6 GHz frequency bands (e.g., DSRC, Wi-Fi, Bluetooth, etc.), VLC, LTE, etc., with the addition of new millimeter wave (mmWave) frequency bands [4]. This chapter will look at the architecture of the physical layer of a hybrid 5G cellular V2X platform, shown in Figure 49.1. Furthermore, a link budget model will be designed and evaluated for the proposed system. Section 49.2 investigates current DSRC as part of the updated IEEE 802.11p standard and C-V2X network topologies. Section 49.3 introduces the various technology areas of the 5G network. Section 49.4 examines the hybrid 5G and V2X platform link budget simulation, modeling, and results.

49.2 Existing V2X communication networks and architectures 49.2.1 DSRC topologies DSRC is a main intelligent transportation system (ITS) of V2X communication according to the IEE802.11p standard. It can be used for unidirectional or bidirectional transmission, operating in the 5.9 GHz sub-frequency band. The DSRC is an ad hoc network type, consisting of nodes that collaborate for information

Vehicle-cellular Vehicle--sat

Direct V2X (V2V/V2I)

Figure 49.1 Hybrid DSRC-cellular V2X

618

Advances in communications satellite systems 2: ICSSC-2019

transfer, with an operating range of 1000 m. Ad hoc networks have both static and mobile classes, where the position of the node is either permanent once added to the network, or can move arbitrarily. Mobile ad hoc networks (MANET) is a cluster of nodes that dynamically form a network to exchange information without any preexisting fixed network or centralized administration [5]. Communication between mobile units are established through infrastructurebased communications (Figure 49.2(a)) or ad hoc communications (Figure 49.2(b)). MANET systems topology constantly changes without prior notice, making routing through these networks very challenging. Mobile Ad hoc nodes onboard vehicles are known as VANET (vehicular ad hoc networks), where wireless mobile hosts are free to move randomly and often act as router at the same time. Ad hoc network traffic differ from wireless infrastructure network traffic, as they can contain: ●

●

●

Peer-to-peer networks: Communication between two nodes are in range and traffic network is usually consistent. Remote-to-remote networks: Communication between two nodes which are out of range but use an intermediate node to relay data. Dynamic traffic networks: Occurs when nodes are in dynamic motion, resulting in poor network activity and connectivity over short intervals.

49.2.2 C-V2X topologies C-V2X communication is an adaptation of cellular RAT for V2X communication, developed by 3GPP [6]. C-V2X takes advantage of the existing widespread cellular infrastructure but overcomes the lack of low latency communication required for V2X by introducing two new transmission modes: modes 3 and 4 stated in 3GPP Rel. 14. In an LTE network, user equipment (UE) communicates to an advanced base station known as eNodeB, which is responsible for scheduling, link adaptation, mobility management, modulation, demodulation, etc. The use of a PC5 air interface enables direct communication to occur between EUs both with and without the involvement of the eNodeB, which reduces latency making it suitable for safety critical applications.

MU

MU MU

MU

MU MU IN

(a)

MU

(b)

MU

MU

Figure 49.2 (a) Infrastructure based communications and (b) ad hoc communications

Integrated space-enabled hybrid 5G-V2X communications link modeling

619

Sidelink transmission mode 3 is used when UEs are in cellular coverage. In this mode, scheduling is implemented by the eNodeB, which manages the communication between UEs. Sidelink transmission mode 4 is used when UEs are outside cellular coverage, where each UE autonomously reserves resources using the resource reservation algorithm. In this algorithm, each UE checks the channel for 1 s to ensure neighboring UEs reserve orthogonal resources semipersistently to minimize packet collisions. Certain studies show that sidelink transmission modes 3 and 4 have a higher link budget than DSRC, which shows better performance. This, however, is not the case when traffic density increases, as C-V2X performance drops rapidly. This is especially true for mode 4, where the rate of frequency reuse enabled by the resource reservation algorithm decreases with traffic density increase, resulting in a greater interference level among C-2VX users.

49.3 5G infrastructure Both DSRC and C-V2X have limitations which do not make them suitable to support a full V2X platform. As such, a hybrid of these two systems must be implemented to create a stable platform that can meet the demands of V2X. The upcoming 5G network is designed to be heterogeneous, supporting numerous modes and a unified air interface which can be adapted to meet the needs of current and future applications [7]. Furthermore, 5G platform aims to enable intelligent matching of resources to applications, sharing of network assets and operations, while redesigning the air interface for efficient spectrum use to meet latency, capacity, and throughput demands of future applications like V2X. To meet these aims, research and development (R&D) is focused on several key research areas shown in Table 49.1 [7]. Significant technology development efforts are being applied toward beam-forming and multiple-input-multiple-output (MIMO) antennas. Beam-forming antennas focus radio signals to a narrow beam, the reduction of high frequency carriers during propagation. Furthermore, MIMO antennas are also developed to improve communication capacity and coverage, which is achieved using special multiplexing and coding to split the data stream into multiple channels [8]. Technology efforts are also applied to develop new network architectures that can support the heterogeneous nature of 5G communication. Qualcomm is one of the vendors working in this area describes a “user-centric design” where the user is “no longer the endpoint of the network, but an integral part of it” [9]. In this architecture, users can connect to one another directly to communicate and share resources. Furthermore, devices will be able to “multihop,” where devices cluster to form hubs and relays for other devices, acting as a mesh network.

49.4 Proposed hybrid DSRC-cellular 5G V2X platform overview The proposed hybrid V2X platform combines DSRC and cellular topologies with 5G infrastructure (such as beam-forming and MIMO antennas, new multiplexing

620

Advances in communications satellite systems 2: ICSSC-2019

Table 49.1 R&D technology areas to meet 5G requirements. Adapted from [7] Parameter

Technology efforts

Latency in the air link

New multiplexing schemes; new coding schemes, shorter transmit time interval (TTI) Latency end-to-end (device to New network architectures core) Connection density New multiplexing schemes; new coding schemes Area capacity density Higher frequency bands; beam-forming antennas; MIMO antennas System spectral efficiency New multiplexing schemes; new coding schemes; MIMO antennas Peak throughput (downlink) Beam-forming antennas; higher frequency bands; MIMO per connection antennas Energy efficiency New multiplexing schemes; new coding schemes, new control channel structures

and coding schemes, user centric and multihop connectivity, etc.) to create a stable V2X platform that meets the V2X requirements of low latency, high bandwidth, and robust connectivity. This platform has two main layers as follows: ● ●

Layer 1: DSRC and/or C-V2X between V2V and V2I Layer 2: Integrated space-enabled vehicle to satellite communication

The first layer uses DSRC communication to transfer safety critical data between vehicles and infrastructures. The safety critical data transmitted will be used to provide applications such as forward collision avoidance (FCA), blind spot warning (BSW), intersection collision warning (ICA), etc. The second layer (C-V2X) is complementary to the DSRC, offering enhanced safety features (such as situational awareness), driver to driver (D2D) communications, autonomous driving, and much more. The final layer is the direct vehicle to satellite communication which is allocated for cases where DSRC or C-V2X is not possible.

49.5 V2X link budget analysis Carrier data links margins [10,11] represent the two major links margins that characterize the uplink and downlink performances of radio communication systems. Link budget analysis is carried out to examine the gains and losses present in the V2X link. The link budget is expressed in (49.1) Pr ðdBmÞ ¼ Pt þGt þGr  L

(49.1)

where the received signal power (Pr) is given as the sum of the transmitting power (Pt), transmitting and receiving antenna gain (Gt and Gr, respectively), followed by the subtraction of any losses in the link.

Integrated space-enabled hybrid 5G-V2X communications link modeling

621

49.5.1 Signal attenuations There are many different types of losses that can be present in a link depending on the signal propagation environment as shown in (49.2) [12]. LðdBÞ ¼ Lp þ LS þ LF

(49.2)

In this equation, the propagational loss in the channel is characterized by three aspects: path loss (Lp), slow fading loss (LS), and fast fading loss (LF). Path loss is the attenuation of a signal as it propagates through space. As a result, it is determined by macroscopic parameters such as the carrier frequency, the distance between transmitter and receiver, and land profile. The fading losses during signal propagation are modeled as the random change in the amplitude and phase of the transmitted signal. Slow fading is caused by the variation in the propagation channel environment due to buildings, roads, and other obstacles in a relatively small area. Fast fading occurs as a result of scattering of the signal by objects near the transmitter. Unlike slow fading, which is long-term, fast fading takes place in short term periods. This analysis focuses on the most prominent type of loss in a link budget, the path loss. The basic formula for path loss is expressed in (49.3):
 4pdf 2 (49.3) Lp ¼ c where d is the distance of propagation, f is the frequency of the transmitted signal, and c is the speed of light (3  108 ms1). This model will be expanded for each platform layer to cater for the change in the macroscopic parameters.

49.5.2 Noise floor and SNR analysis The presence of noise in a communication system is the most limiting factor in the quality of signal received, as it directly impacts the reliability of a communications system. Receiver noise is introduced in a communication system either through interfering radiation captured by the antenna or generated in the receiver circuitry [13]. The noise figure (NF) is the measure of signal degradation of the signal-tonoise-ratio in a device, expressed in (49.4). NFðdBÞ ¼ MDS  ENP  10ðBWÞ

(49.4)

The noise figure is obtained by subtracting the bandwidth of the signal (BW) and effective noise power (ENP) from the minimum detectable signal (MDS). As per the IEEE 802.11p standard, the bandwidth allocated for DSRC and C-V2X communication is 75 MHz. The MDS is equal to the noise floor, which presumed to be 90 dBm. The ENP is calculated in (49.5)
 kT (49.5) ENP ¼ 10 log10 1 mW where k is the Boltzmann constant (1.381023), and T is the standard reference temperature (290ºK). Using this equation, ENP is calculated to be 174 dBm.

622

Advances in communications satellite systems 2: ICSSC-2019

V1

d V2

Figure 49.3 V2V stationary

Subtracting the noise power N from the received signal power Pr yields the signal to noise ratio (SNR), shown in (49.6). SNRðdBÞ ¼ Pr  N

(49.6)

49.6 Layer 1: DSRC link analysis Link budget analysis for this section is broken down into two cases. Case 1 deals with V2V communication when both vehicles are stationary. Case 2 examines the link parameters when one vehicle is in motion and the other is stationary.

49.6.1 Case 1: V2V stationary The V2V communication between two parked vehicles V1 and V2 is described in Figure 49.3, where the distance of propagation is given by d. The link budget calculation for this case is carried out using (49.1) and (49.2). This case is used to model V2V or V2I link where the position of the nodes is fixed. The results shown in Figures 49.4 and 49.5 show the impact of varying the distance of propagation on the path loss and received signal power. As the distance of propagation increases, the path loss increases exponentially and the received signal power decreases exponentially. The optimum link range (d) is less than 15 m where the signal power received (Pr) ranges from 42.5 to 20 dB.

49.6.2 Case 2: V2V stationary and dynamic This case looks at the link budget between a stationary vehicle (V1) and a vehicle in motion (V2), shown in Figure 49.6. As the location of the dynamic vehicle is not fixed, the frequency of the received signal will not be the same as the frequency of the source. As V2 approaches V1, the frequency of the received signal will be higher than the source

Integrated space-enabled hybrid 5G-V2X communications link modeling

623

Distance of propagation vs path loss

120 110

Path loss (dB)

100 90 80 70 60 50 40

0

200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 Distance (m)

Figure 49.4 Distance of propagation versus path loss

Distance of progation vs received power

50 40

Received power (dB)

30 20 10 0 –10 –20 –30

0

200 400 600 800 1,000 1,200 1,4001,600 1,800 2,000 Distance (m)

Figure 49.5 Distance of propagation versus received power frequency, and the opposite is exhibited when V2 moves past V1 resulting in a lower frequency than the source. This shift in frequency is known as the Doppler effect, as is expressed in (49.7) [14] fd ¼

vfc cos q c

(49.7)

624

Advances in communications satellite systems 2: ICSSC-2019

θ

V1

dmin

v dmax

θ V2

Figure 49.6 V2V stationary-mobile

where fd is the Doppler frequency (Hz), v is the relative velocity between V1 and V2 (ms1), c is the speed of EM wave (3  108 ms1), and Q is the angle between the distance of propagation and direction of travel of V2 ( ). fr ¼ fc  fd

(49.8)

The difference between the source frequency ( fc) and Doppler frequency or Doppler shift (fd) produces the received frequency (fr), expressed in (49.8). As such, the received frequency which accounts for the Doppler shift can be added to the path loss equation to evaluate the impact of Doppler shift in a direct V2X link. However, as the vehicle is mobile, the distance of propagation (d) constantly changes. The distance of propagation is a function of the linear distance between V1 and V2 (dmin) and the angle between the distance of propagation and direction of travel (Q) expressed in (49.9). d¼

dmin sin q

(49.9)

The results shown in Figures 49.7 and 49.8 exhibit the performance of the link between V1 (stationary) and V2 (mobile), as V2 drives past V1. As V2 approaches V1, azimuth angle of relative vehicles travel (Q) increases from 0º to 90º. When Q ¼ 90º, the location of V2 is directly perpendicular to V1, which is the minimum

Integrated space-enabled hybrid 5G-V2X communications link modeling

625

Path loss vs azimuth angle

95 90

Path loss, PI (dB)

85 80 75 70 65 60 55

0

20

40

60 80 100 120 Azimuth angle, phi (°)

140

160

180

Figure 49.7 Path loss versus azimuth angle

Received power vs azimuth angle

35

Received power, Pr (dB)

30 25 20 15 10 5 0 –5 0

20

40

60 80 100 120 Azimuth angle, phi (°)

140

160

180

Figure 49.8 Received power versus azimuth angle

distance of propagation (dmin). As V2 drives past V1, the angle of motion increases to 180º. When Q ¼ 0º or 180º, V2 is furthest away from V1 resulting in a maximum distance of propagation (dmax). Figures 49.7 and 49.8 show that the maximum received signal power is obtained when path loss is minimal at Q ¼ 90º. Furthermore, minimal received signal power is observed at maximal path loss, where Q ¼ 0º and 180º.

626

Advances in communications satellite systems 2: ICSSC-2019

Table 49.2 Doppler shift at different frequencies V2 velocity, v Doppler fre(mph) quency, fd (Hz)

Received frequency, Path loss, LP (dB) at maxfr (GHz) imum distance

NA 5 10 30 50 70

5.900000000 . . . 5.899999956 . . . 5.899999912 . . . 5.899999736 . . . 5.899999560 . . . 5.899999385 . . .

No Doppler shift 44 88 264 440 615

94.14795487 94.14795481 94.14795474 94.14795448 94.14795423 94.14795397

49.6.3 Analysis of the effects of vehicle motion on link performance The velocity of an object is a key component of its Doppler frequency; hence analysis is carried out to examine the impact of this component on the overall performance of the V2X link. Analysis is carried on a range of typical UK driving speeds from 5 to 70 mph, at a maximum distance of propagation, shown in Table 49.2. These results show that the Doppler frequency increases with vehicle velocity, resulting in a reduction of received signal frequency. At typical DSRC frequencies of 5.9 GHz, the effects of Doppler shift are negligible with a maximum Doppler shift of 52 Hz taking place when the V2 travels at 70 mph. Furthermore, the results show that the effects of Doppler shift are negligible on the path loss.

49.7 Layer 3: integrated space-enabled vehicle to satellite communication The satellite provides an advantage of all round connectivity and its deployment in the vehicular hybrid system provides an emergency substitute for a regular vehicle to infrastructure communication systems. Two scenarios are supported with the deployment of the satellite in the V2X systems, viz: When the satellite is aligned vertically with the vehicle, the distance between them is modeled as the direct distance between the earth and the LEO satellite, denoted as “d” in Figure 49.9. Figure 49.10 shows the change in the path loss as a result of the vertical distance between the satellite and earth. The “d” is the vertical distance between the vehicle and the satellite is approximately 800 km. Figure 49.11 shows the result of the received power with respect to the propagation distance. As distance increases, the receive power attenuates and degrades. Figure 49.12 shows the SNR response over the propagation distance of transmission.

Integrated space-enabled hybrid 5G-V2X communications link modeling

627

D

θ d

Figure 49.9 Vehicle to satellite illustration Distance of propagation vs path loss

110 100

Path loss (dB)

90 80 70 60 50 40

0

100

200

300

400 500 600 Distance km

700

800

900

Figure 49.10 Path loss versus distance of propagation When the vehicle is at an angle Q with the satellite, the distance, “D” which includes the earth’s radius and vertical distance is modeled as shown in (49.10): qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d ¼ ðRE þ hÞ2  RE 2 cos2 j  RE sin j (49.10) where RE is the earth’s radius and h is the distance of the LEO satellite from ground (800 km). The simulated vehicle to satellite network gives an added communication structure for data capacity enhancement with the availability of a substitute link for V2X. However, due to the large propagation distance between vehicles and satellites, an integration of available nodes and infrastructures suffice for the degradation due to propagation distance. With vertical positioning, the SNR is increased when compared with the nonvertical position model (Figures 49.13–49.15).

628

Advances in communications satellite systems 2: ICSSC-2019 Distance of propagation vs received power

50

Received power (dB)

40 30 20 10 0

–10 –20 0

100

200

300

400 500 600 Distance (km)

700

800

900

Figure 49.11 Received power versus distance of propagation Distance of propagation vs SNR

140 130

SNR (dB)

120 110 100 90 80 70

0

100

200

300

400 500 600 Distance (km)

700

800

900

Figure 49.12 SNR versus distance of propagation

49.8 Conclusion The current V2X DSRC and C-V2X infrastructures are not suitable to fully support future developments due to their limitations. As such, a hybrid V2X platform that combines these infrastructures would be more equipped to support V2X. Furthermore, the developments of the emerging 5G platforms will improve the

Integrated space-enabled hybrid 5G-V2X communications link modeling Distance of propagation vs path loss

120

Path loss (dB)

629

115

110

105 500

1,000

2,500 1,500 2,000 Distance (km)

3,000

3,500

Figure 49.13 Path loss versus distance of propagation nonvertical positioning

–16

Distance of propagation vs received power

Received power (dB)

–18 –20 –22 –24 –26 –28 –30 500

1,000

2,500 1,500 2,000 Distance (km)

3,000

3,500

Figure 49.14 Received power versus distance of propagation for nonvertical positioning capabilities of the hybrid V2X system, with the addition of technologies such as beam-forming and MIMO antennas, new types of multiplexing, and a user-centric design. The hybrid V2X platform proposed in this chapter investigates the link budget of a DSRC/C-V2X communications link, with the addition of an integrated direct

630

Advances in communications satellite systems 2: ICSSC-2019 Distance of propagation vs SNR

80 78

SNR (dB)

76 74 72 70 68 66 500

1,000

1,500

2,500 2,000 Distance (km)

3,000

3,500

Figure 49.15 SNR versus distance of propagation for nonvertical positioning vehicle-satellite link to be utilized as a back-up when DSRC/C-V2X communication is not possible. The results show the received power to be 21 dBm and path loss to be 108 dB at the optimal communication distance of 1 km. Furthermore, the impact of Doppler shift for DSRC and C-V2X communication was investigated and found to be negligible. Moreover, the link budget carried out for the vehiclesatellite shows an SNR of 79 dB, at a transmission distance of 900 km where the satellite is directly over the vehicle. Furthermore, a significant increase in attenuation is exhibited when data transmission is nonvertical, with maximum attenuation found at 3,300 km with an SNR of 66.5 dB.

Acknowledgments The authors wish to thank Innovate UK and the Jeff Gosling Hand Controls Ltd. for sponsoring this research in the Department of Engineering, Manchester Metropolitan University, Manchester, UK.

References [1] Abboud, K., Omar, H., and Zhuang, W. “Interworking of DSRC and cellular network technologies for V2X communications: A survey.” IEEE Transactions on Vehicular Technology. 2016; 65(12): 9457–9470. [2] Bey, T. and Tewolde, G. “Evaluation of DSRC and LTE for V2X.” 2019 IEEE 9th Annual Computing and Communication Workshop and Conference (CCWC). 2019.

Integrated space-enabled hybrid 5G-V2X communications link modeling

631

[3] Habibi, M., Nasimi, M., Han, B., and Schotten, H. “A comprehensive survey of RAN architectures toward 5G mobile communication system.” IEEE Access. 2019; 7: 70371–70421. [4] Rappaport, T., Sun, S., Mayzus, R., et al. “Millimeter wave mobile communications for 5G cellular: It will work!.” IEEE Access. 2013; 1: 335–349. [5] Gheorghiu, R., Cormos, A., Stan, V., and Iordache, V. “Overview of network topologies for V2X communications.” 2017 9th International Conference on Electronics, Computers and Artificial Intelligence (ECAI). 2017. [6] Naik, G., Choudhury, B., and Park, J. “IEEE 802.11bd & 5G NR V2X: Evolution of radio access technologies for V2X communications.” IEEE Access. 2019; 7: 70169–70184. [7] GSA. “The road to 5G: Drivers, applications, requirements and technical development.” GSA. 2015. [8] Yu, C., Jing, J., Shao, H., et al. “Full-angle digital predistortion of 5G Millimeter-wave massive MIMO transmitters.” IEEE Transactions on Microwave Theory and Techniques. 2019; 67(7): 2847–2860. [9] Qualcomm Technologies Inc. 5G – Vision for the next generation of connectivity. San Diego, CA: Qualcomm Technologies Inc.; 2015. [10] Ekpo, S. and George, D. “Impact of noise figure on a satellite link performance.” IEEE Communications Letters. 2011; 15(9): 977–979. doi: 10.1109/ LCOMM.2011.11.111073. [11] Ekpo, S. C. “Parametric system engineering analysis of capability-based small satellite missions.” IEEE Systems Journal. 2019; 13(3): 3546–3555. doi: 10.1109/JSYST.2019.2919526. [12] Agrawal, D. and Zeng, Q. Introduction to wireless and mobile systems. 4th ed. Boston, MA: Cengage Learning; 2016. pp. 64–81. [13] Fette, B., Aiello, R., Chandra, P., et al. RF & wireless technologies (Newnes know it all series). Newnes, Oxford, UK; 2008. pp. 90–93. [14] Aye, A., Lin, S., and Win, K. “Study on indoor RF propagation model with Doppler effect.” International Journal of Scientific and Research Publications (IJSRP). 2018; 8(12): 258–263.

This page intentionally left blank

Chapter 50

K/Ka-band transceiver sensitivity modeling and link characterization for integrated 5G-LEO communication applications Mfonobong Uko1, Muazzam Zafar1, Arslan Altaf 1, Solomon Udeshi1, Sunday Ekpo1 and Bamidele Adebisi1

This chapter presents a transceiver sensitivity modeling and link characterization over-the-air (OTA) for K/Ka-band frequency for next-generation 5G and low earth orbit (LEO) communication applications. The transceiver system is simulated using estimated link budget parameters. The simulated transceiver includes a DQPSK modulator, a power amplifier, a low noise amplifier, and mixers optimized to meet industry specifications. The transmitter front-end shows a 24.6 dBm output power with an input transmission power of 0 dBm. The receiver simulation shows a gain of 48 dBm and an intermediate frequency output power of 35.8 dBm. The simulated sensitivity of the receiver spans from
115 dBm to
110 dBm, indicating good coverage over the channel bandwidth. Key Words: 5G; RF communication link; link budget; satellite; transceiver; K/Ka-band

50.1 Introduction The fifth generation (5G) of communication network promises massive capacity, 1,000 times more than 4G, with super-fast data rate over 100 times more than 4G. This brings about the need for easier adaptability and performance of communication architectures, with the requirement of new technology development for efficient synchronization and deployment of 5G networks, in which the millimeterwave spectrum has been earmarked for it deployment [1]. Furthermore, 5G wireless networks are expected to be a combination of existing network tiers (2G, 3G, and 4G), with transmit powers, back-haul connections, and distinct radio access 1 Communication and Space Systems Engineering Team, Manchester Metropolitan University, Manchester, UK

634

Advances in communications satellite systems 2: ICSSC-2019

technologies (RATs) that can be accessed by an unprecedented number of smart and heterogeneous wireless devices. 5G will also require much higher bandwidth, making use of various frequency bands ranging from the sub-1 GHz, 1–6 GHz, and above 6 GHz bands. The 26 GHz and 28 GHz have emerged as two of the most important 5G mmWave bands as they offer the widest harmonization with minimized user equipment complexity. The European Union has designated the 24.25–27.5 GHz band as a pioneer 5G band. This high-frequency spectrum offers very high data capacity and speeds with vast communication resources, with satellite communication systems deployed in the Ka-band (26.5 GHz to 40 GHz). The last decade has seen tremendous advancement in the 4G LTE standards with an increase in the peak data rate up to 1 Gbps. These advancements, however, aren’t feasible to satisfy the growing demand for data due to the exponential mobile traffic growth. With the evolution of 5G standards, these existing networks are being incorporated with the emerging 5G standards to meet consumers’ needs of higher data rates and capacity, as shown in Figure 50.1. With advancement well in progress, 5G will be required to deliver high mobile broadband, reliable transmission of high-speed data from the transmitter front-end to the receiver front-end [2]. This requires a critical analysis of the application link performance [3], taking into consideration large-capacity communication, the noise level, power level, and sensitivity of the overall communication system, in order to cope with various communication demands of space and terrestrial applications [4–6]. In this chapter, a transceiver sensitivity modeling and link characterization for K/Ka-band frequency of 23–28 GHz is carried out for 5G and low earth orbit (LEO) communication applications. Section 50.2 presents an understanding of the 5G link behavior in terms of path loss. In Section 50.3, we consider a 5G mmWave

Satellite gateway and terrestrial backhaul

Connectivity for disaster-prone and hard-to-reach location

Connectivity for users on the move Connectivity for residential areas Connectivity for densely populated area (industries and commercial areas)

Figure 50.1 Satellite network integration for 5G

K/Ka-band transceiver sensitivity modeling

635

link budget for a K/Ka-band transceiver. A sensitivity modeling for the K/Ka-band transceiver is presented in Section 50.4. Section 50.5 presents the achieved results with their interpretation. Section 50.6 concludes the chapter.

50.2 5G link characterization In designing a reliable communication links, an understanding of the optimal link resource availability and link-quality-performance such as earth–space and space– earth (uplink and downlink) line-of-sight (LOS) and non-line-of-sight (NLOS) are important. The path loss determines the large-scale fading behavior of the communication channel with reference to the transmit power. The LOS path loss is close to free space path loss, whereas the NLOS path loss is significantly deviated owing to difference in frequency and path terrain. This involves the need for frequency planning based on the desired application. For LEO communication, the satellite provides a transceiver path for uplink and downlink data transmission to the ground station, operating at the K/Ka-band frequency. The challenge using this frequency includes the absorption of atmospheric gases, cloud attenuation, rain attenuation, and group delay due to the different travel times of different frequency components. Therefore a link characterization to meet performance in terms of high data rate, multiple-input multiple-output (MIMO) transmission, and adaptive beam-forming for 5G application is essential [1]. For integrated 5G-LEO communication application, this link characterization is enhanced through the employment of large antenna arrays for improved coverage, highorder modulation for required data rate, beam adaptation, and wide-band availability [7]. The design frequency for this work covers the 23–28 GHz bandwidth for 5G communication, which also is the passband of the BPF, at a design frequency of 26 GHz. Theoretically, the overall path loss is given by   4ld 2 (50.1) Lp ¼ l where d is the distance of propagation and l the propagation wavelength. The path loss design of any given communication application varies at any given distance with respect to the carrier frequency. A common path-loss model based on channel impulse responses (CIRs) is defined as Lp ð f ; dÞ½dB ¼ Lp ð f ; d0 Þ þ 10n log10 ðd=d0 Þ þ Xs

(50.2)

where Lp( f,d) is the path loss at different frequencies with various Tx–Rx separation distance, Lp is the path loss in dB at a close-in (CI) distance, d0, n is the pathloss exponent, and Xs is a zero-mean Gaussian-distributed random variable with standard deviation s. For a small distance of 10–300 m, the path loss at the design frequency (26 GHz) is shown in Figure 50.2.

636

Advances in communications satellite systems 2: ICSSC-2019 5G path loss vs small cells distance

120

5G path loss (dB)

110 100 90 80 70 60 50

50

200 100 150 Small cells distance, x (m)

250

300

Figure 50.2 Simulated 5G NR path loss at 26 GHz

50.3 5G mmWave link budget for a K/Ka-band transceiver To ensure adequate coverage, a link budget is essential. The 5G mmWave link budget involves the analysis of system components and parameters for a given communication link [3,7]. This takes into consideration the transceiver transmit and receive power, path loss, modulation scheme, data rate, bit error rate, uplink and downlink frequency, antenna gains as well as losses due to rain fade, shadowing, foliage, atmospheric absorption, humidity, and Fresnel blockage. The link budget is a vital area of focus for the 5G system design and components deployment. These components are carefully selected to achieve good performance and receiver sensitivity. For space communication, the carrier link margin and data link margins are the two key performance matrices in determining the link characterization [3]. A radio link budget based on theoretical assumptions is presented in Table 50.1 showing system design parameters and requirement for the K/Ka-band transceiver. The receiver sensitivity, S, determines the minimum received power required to attain a desired error probability. S ¼ SNRðdBÞ þ 10 log10 ðBÞ
174 þ NF

(50.3)

where NF is the noise figure of the receiver, B the bandwidth, SNR the signal-tonoise ratio. The noise figure for the receiver is vital for the sensitivity of a communication device and can be defined using the signal-to-noise (SNR) ratios of the input and output of the receiver. SNRoutput ¼ SNRinput
NF

(50.4)

K/Ka-band transceiver sensitivity modeling

637

Table 50.1 Link budget calculation at 26 GHz design frequency

A B C D E F G H I J K L M N O P Q R S T U V W X

Parameter

Value

Unit

Bandwidth B Receiver sensitivity limit S Antenna array elements Antenna array gain SNRmin Transmitter (Tx) EVM Receiver (Rx) SNR Noise density Thermal noise power Receiver (Rx) noise figure Transmitted power Pt Transmitter (Tx) gain Gt Receiver (Rx) gain Gr Transmitter-end losses Path loss (Lp) Link margin (Lm) Path loss coefficient Transmit EIRP Shadowing loss (Ls) Minimum detectable signal Distance (d) Rx signal bandwidth (RxSig) Foliage loss (Lf ) Rain fade (R )

100
84 16 17 24.49
26 29.83
174
94 6 17–40 20 20 5 133–156
9 to þ14 2.5 40–63 20–30
88 10–150
107 10–50 10–50

MHz dBm N/A dB dB dB dB dBm/Hz dBm dB dBm dBi dBi dB dB dB N/A dBm dB dBm m dBm dB dB

From the link budget in Table 50.1, the following link equation is obtained from theoretical analysis. The received power at the receiver front-end is calculated thus Pr ðdBmÞ ¼ Pt þ Gt þ Gr
Lp
Rf
Ls
Lf

(50.5)

EIRP ¼ S þ SNR
Gr þ Lp

(50.6)

Lm ¼ EIRP
Lp
Rx Sig

(50.7)

50.4 Sensitivity modeling for integrated 5G-LEO communication applications In this section, a description and analysis of the K/Ka-band transceiver frontend and sensitivity is presented. The transceiver designed operates in the 26 GHz frequency band earmarked for 5G communication applications in the UK. For the transceiver frequency planning, the parameters in Table 50.2 are used.

638

Advances in communications satellite systems 2: ICSSC-2019

Table 50.2 Transceiver frequency planning S/N

Parameter

Value

Unit

1 2 3

Pass-band frequency Intermediate frequency (IF) RF carrier frequency

23–28 350 26

GHz MHz GHz

MIXI

BPFI

AMP1

MIX2

BPF2

AMP2

PA

BPF3

IF input

RF output

LO1

LO2

Figure 50.3 26 GHz transmitter subsystems

50.4.1 Transmitter front-end modeling The transmitter is made up of two double stage up-conversion mixers as shown in Figure 50.3. The transmitter consists of local oscillators, up-converters (mixers), filters, and a power amplifier. The transmitter is modeled at an intermediate frequency (IF) of 350 MHz which is double up-converted to 26 GHz with transmission bandwidth of 550 MHz. An input power of 0 dBm is supplied to the transmitter. The power amplifier is used to increase the output power for transmission. For the transmitter, a very low spur emission and intermodulation distortion (IMD) products are required. To satisfy the spur emission requirement, the unwanted level of the signals should be at least
74 dB below the required signal.

50.4.2 Receiver front-end modeling The receiver is made up of a single stage down-conversion mixer as shown in Figure 50.4. It consists of a local oscillator, down-converter (mixer), filters, and a low noise amplifier. The band pass filter filters the out-band signals and the required signal is then amplified by the low noise amplifier followed by the mixer for RF to IF conversion. A IF amplifier is added to the back-end of the receiver chain to guarantee the noise figure and gain of the receiver.

50.5 Simulation result and analysis 50.5.1 Transmitter front-end analysis The power output spectrum of the transmitter is shown in Figure 50.5. An output power of 24.623 dBm after the second band pass filter is measured from the

K/Ka-band transceiver sensitivity modeling BPF1 LNA

MIX1

639

BPF2 AMP1 IF AMP BPF3

RF input

IF output

LO

Figure 50.4 26 GHz receiver subsystems

Power output spectrum (dBm)

0

–20

–40

–60

–80

–100 0 20 40 –140 –120 –100 –80 –60 –40 –20 Frequency (kHz)

60

80

100 120 140

Figure 50.5 Transmitter output spectrum for a DQPSK signal simulation (Figure 50.6). The cascaded voltage gain response of the transmitter subsystem components is shown in Figure 50.7. An output gain of 26.6 dB is achieved from the design. The transmitter path noise figure is shown in Figure 50.8. The band pass filter at the transmitter input (BPF1) contributes significantly to the noise level. However, this effect of noise is subsequently reduced at the output band pass filter (BPF2).

50.5.2 Receiver front-end analysis An output power of 35.8 dBm is measured at the IF frequency (350 MHz) receiver output end from the simulation (Figure 50.9). The simulated cascaded voltage gain in dB of each of the receiver subsystems is deduced from Figure 50.10. From Figure 50.11, the variation of the noise at each component is simulated. The mixer introduces significant noise into the receiver chain which is reduced by the second band pass filter.

640

Advances in communications satellite systems 2: ICSSC-2019 50

Output power levels

0 –50 –100 –150 –200 –250 0

10

40

20 30 Frequency (GHz)

50

Figure 50.6 Spectral response of the output power of the transmitter at 26 GHz

40

Voltage gain (dB)

20 0 –20 –40 –60 BPF2

Powerstage

Driverstage1

BPF3

MIX2

Driverstage

BPF1

MIX1

Figure 50.7 Simulated cascaded voltage gain in dB of the transmitter subsystems

50.5.3 5G NR receiver sensitivity modeling The reference receiver sensitivity estimating relationship, S, in dBm is given by S ¼ SNRðdBÞ þ 10 log 10ðBÞ
174 þ NF

(50.8)

where B is the channel bandwidth of the signal, NF is the noise figure of the receiver in dB, and SNR is the signal-to-noise ratio in dB. The sensitivity of the receiver spans from
115 dBm to
110 dBm (Figure 50.12). This performance gives the advantage

K/Ka-band transceiver sensitivity modeling

641

16

Noise figure (dB)

14 12 10 8 6 4 2 0 BPF2

Powerstage

Driverstage1

BPF3

MIX2

Driverstage

BPF1

MIX1

Figure 50.8 Simulated noise figure from system input to component output of the transmitter subsystems

Output power levels

50

0

–50

–100

–150 0

10

20 Frequency (GHz)

30

40

Figure 50.9 Spectral response of the output power of the receiver at 350 MHz IF frequency of wider coverage for 5G applications as well as reduces the need for power hungry power amplifier at the transmitter front-end.

50.6 Conclusion A 26 GHz transceiver sensitivity design and simulation has been carried out for 5G and LEO applications. The radio link analysis indicates an output power of 24.6 dBm in the transmitter front end and 35.8 dBm in the receiver front end.

642

Advances in communications satellite systems 2: ICSSC-2019 60

Voltage gain (dB)

50 40 30 20 10 0 –10 BPF1

LNA

MIX1

BPF2 Gain_Block IFAmp

BPF3

Figure 50.10 Simulated cascaded voltage gain in dB of the receiver subsystems

10

Noise figure (dB)

8 6 4 2 0 BPF1

LNA

MIX1

BPF2 Gain_Block IFAmp

BPF3

Figure 50.11 Simulated noise figure from system input to component output of the receiver subsystems The sensitivity of the receiver spans from
115 dBm to
110 dBm, indicating good coverage over the channel bandwidth.

Acknowledgments The authors wish to thank the Niger Delta Development Commission of Nigeria for sponsoring this research at the Manchester Metropolitan University, UK through its scholarship scheme.

K/Ka-band transceiver sensitivity modeling

5G NR receiver response (dBm)

–80

643

5G NR receiver response vs channel bandwidth

–90 –100 –110 –120 –130 –140 –150 0.5

1

2.5 3 1.5 2 Channel bandwidth, B(Hz)

3.5 ×108

Figure 50.12 Simulated 5G NR receiver sensitivity at 26 GHz from 50 to 400 MHz bandwidth

References [1] Leinonen, M.E., Destino, G., Kursu, O., Sonkki, M., and Pa¨rssinen, A. “28 GHz wireless backhaul transceiver characterization and radio link budget.” ETRI Journal. 2018; 40(1): 89-100. [2] Lin, J. “Synchronization requirements for 5G: An overview of standards and specifications for cellular networks.” IEEE Vehicular Technology Magazine. 2018; 13(3): 91–99. [3] Ekpo, S.C. and George, D. “Impact of noise figure on a satellite link performance.” IEEE Communications Letters. 2011; 15(9): 977–979. [4] Ekpo, S.C. “Thermal subsystem operational times analysis for ubiquitous small satellites relay in LEO.” International Review of Aerospace Engineering (IREASE). 2018; 11(04): 48. [5] Kaneko, K., Nishiyama, H., Kato, N., Miura, A., and Toyoshima, M. “Construction of a flexibility analysis model for flexible high-throughput satellite communication systems with a digital channelizer.” IEEE Transactions on Vehicular Technology. 2018; 67(3): 2097–2107. [6] Fenech, H., Amos, S., Tomatis, A., and Soumpholphakdy, V. “High throughput satellite systems: An analytical approach.” IEEE Transactions on Aerospace and Electronic Systems. 2015; 51(1): 192–202. [7] Ekpo, S.C. “Parametric system engineering analysis of capability-based small satellite missions.” IEEE Systems Journal. 2019; 13(3): 3546–3555.

This page intentionally left blank

Chapter 51

Link budget design for integrated 5G-LEO communication applications Mfonobong Uko1, Muazzam Zafar1, Arslan Altaf 1, Solomon Udeshi1, Sunday Ekpo1 and Bamidele Adebisi1

Fifth-generation (5G) promises an all-round connectivity scenario through the Internet of things, between people, and their environments as well as delivery of new communication levels and efficiency. To guarantee this goal of an all-round connectivity, the integration of satellite systems is vital based on the wide coverage area provided by satellite infrastructures. Hence a detailed analysis and definition of parameters and architectures for the seamless deployment of satellite systems within the 5G network is presented and captured in the design of a link budget for integrated 5G-low earth orbiting communication applications. Key Words: 5G; RF communication link; link budget; satellite; transceiver

51.1 Introduction The fifth-generation (5G) standards promise massive capacity, 1,000 times more than 4G, with super-fast data rate over 100 times more than 4G, and ultra-low latency applications such as connected car and massive machine-to-machine (M2M) and Internet of things (IoTs) applications. This brings the need for efficient communication structures to handle the exponential demand for data from users as well as provision of affordable physical structures for mobile network operators (MNO). To cater for these demands in terms of wide coverage and low-cost devices, low Earth orbit (LEO) satellite deployment and connectivity are vital for full coverage [1], especially to the inaccessible areas by terrestrial infrastructures. The integration for satellite systems also provides an all-round increase in the capacity and bandwidth for the integration of 5G [2–4] as shown in Figure 51.1. This leads to a lower cost of deployment of communication infrastructures for the 1 Communication and Space Systems Engineering Team, Manchester Metropolitan University, Manchester, UK

646

Advances in communications satellite systems 2: ICSSC-2019

Satellite gateway and terrestrial backhaul

Connectivity for disaster-prone and hard-to-reach locations

Connectivity for users on the move

Connectivity for densely populated are (industries and commercial areas)

Figure 51.1 Satellite integration for 5G communication network MNO as well as fast data rates and resources for users. Also, satellite network deployment will provide an integration of existing network tiers (2G, 3G, and 4G) with the new 5G structure, for access to an unprecedented number of smart and heterogeneous wireless devices [5,6]. With advancement well in progress, 5G-LEO communication integration requires a critical analysis of the application link performance [5]. A link budget design and analysis determine component (system and subsystem) parameters and the signal power at the receiver front-end needed to recover the information transmitted. In this chapter, we analyze the link budget for 5G and LEO communication applications. Section 51.2 presents an understanding of RF communication link budget design and calculation for 5G-LEO communication application. In Section 51.3, we consider a 5G mmWave link budget for a K/Ka-band satellite. Section 51.4 presents simulated results for the link budget. Section 51.5 concludes the chapter.

51.2 5G-LEO RF link budget design and calculation 51.2.1 Received power determination In designing a reliable communication link, a comprehensive budget based on the components, subsystems, and system parameters is required to characterize the quality of the wireless communication link, relating the power at the receiver with regard to the power at the transmitter [5]. It is a theoretical calculation of the end-to-end performance of the communications link. For integrated 5G-LEO communication application, this link characterization is enhanced through budget calculations generally influenced by the satellite’s trajectory and spatial conditions [6].

Link budget design for integrated 5G-LEO communication applications

647

Theoretically, the RF link budget equation for receiver power is given by ! Pt Gt Gr c2 Pr
(51.1) ð4pÞ2 R2 f 2 where Pr is the power received by the receiving antenna; Pt is the power applied to the transmitting antenna; Gt is the transmitter antenna gain; Gr is the receiver antenna gain; c is the speed of light (3  108 m/s); R is the distance between transmitter and receiver, and f is the frequency in Hz. In terms of decibel Pr ðdBÞ ¼ Pt þ Gt þ Gr
20 logð f  RÞ þ 147:6

(51.2)

51.2.2 Path loss modeling Path loss Lp describes signal attenuation between transmitter and receiver antenna as a function of the propagation distance, d. It is one of the mechanisms causing attenuation between the transmitter power amplifier (PA) and receiver front-end system low noise amplifier. The modeling of Lp, therefore, becomes an essential tool in predicting some essential performance criterion. For instance, we are able to determine the required output power of a PA to overcome Lp in a given satellite communication link at mmWave frequencies. The calculation of Lp is given by   4ld 2 (51.3) Lp ¼ l where d is the distance of propagation between the satellite and the ground station (km) and l is the propagation wave-length. The propagation distance, d, varies as a function of the satellite elevation angle, j, above the horizon and is given by qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (51.4) d ¼
RE : sin f þ R2E :ðsinfÞ2 þ h þ 2:RE :h This is further expressed as qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d ¼
RE : sin f þ ðRE þ hÞ2
R2E :cos2 fÞ

(51.5)

where RE is the radius of the earth ¼ 6,371 km; h is the altitude of the satellite trajectory (km). From (51.3), the satellite free space path loss is given by   4pf 2 (51.6) Lp ¼ 20log10 c

648

Advances in communications satellite systems 2: ICSSC-2019 5G path loss vs. small cells distance

120 110

5G path loss (dB)

100 90 80 70 60 50

50

100 150 200 Small cells distance, x (m)

250

300

Figure 51.2 Simulated 5G-LEO path loss for small cell distance

The received RF signal equation now becomes Pr ðdBÞ ¼ Pt þ Gt þ Gr
Lp

(51.7)

For a small terrestrial distance of 10 to 300 m, the path loss is shown in Figure 51.2.

51.3 5G mmWave link budget for a Ka-band satellite 5G-LEO integration depends on the uplink and downlink frequencies (Table 51.1) for seamless transmission of information with less interference (Figure 51.3). A radio link budget based on theoretical assumptions is presented in Table 51.2 showing system design parameters and requirement for the 5G-LEO Ka-band satellite link.

Table 51.1 Ka-band satellite link budget allocation Transmit parameter

Frequency (GHz)

Bandwidth (GHz)

Uplink Downlink

27.5–31 17.7–21.2

3.5 3.5

Link budget design for integrated 5G-LEO communication applications

649

Satellite ink

l

Up

k

lin

n ow

D

Earth station

Terrestrial base stations

Figure 51.3 An integrated 5G-LEO satellite link description for Ka-band

Table 51.2 5G-LEO link budget calculation

A B C D E F G H I J K L M N O P Q R

Parameter

Value

Unit

Bandwidth (B) Receiver sensitivity (S) SNRmin Transmitter (Tx) EVM Transmit EIRP Noise density Thermal noise power Satellite distance (d) 5G terrestrial distance (d) Receiver (Rx) noise figure Transmitted power (P) Transmitter (Tx) gain (Gt) Receiver (Rx) gain (Gr) Transmitter-end losses Path loss (Lp) Link margin (Lm) Foliage loss (Lf ) Shadowing loss (Ls)

3.5
100 20
26 40–63
174
94 600–900 10–150 6 17–40 20 20 5 133–156
9 to þ14 10–50 20–30

GHz dBm dB dB dBm dBm/Hz dBm km m dB dBm dBi dBi dB dB dB dB dB

The receiver sensitivity, S, determines the minimum received power required to attain a desired error probability. S ¼ SNRðdBÞ þ 10log10 ðBÞ
174 þ NF

(51.8)

where NF is the noise figure of the receiver, B is the bandwidth, and SNR is the signal-to-noise ratio. The noise figure for the receiver is vital for the sensitivity of a

650

Advances in communications satellite systems 2: ICSSC-2019 5G path loss vs. small cells distance

120 110

5G path loss (dB)

100 90 80 70 60 50

50

100 150 200 Small cells distance, x (m)

250

Figure 51.4 SNR versus 5G-LEO satellite link elevation angle

communication device and can be defined using the signal-to-noise ratios (SNRs) of the input and output of the receiver SNRoutput ¼ SNRinput
NF

(51.9)

From the link budget in Table 51.2, the following link equations are obtained: Pr ðdBmÞ ¼ Pt þ Gt þ Gr
Lp
Rf
Ls
Lf

(51.10)

EIRP ¼ S þ SNR
Gr þ Lp

(51.11)

51.4 Simulation result and analysis This section shows the simulated performances of the 5G-LEO satellite link. Figure 51.4 indicates SNR according to 5G-LEO satellite link elevation angle at different bandwidth. At 90 , the SNR of 50 MHz frequency band is 90 dB while SNR at 400 MHz is 80 dB. Hence, it is concluded that the SNR according to elevation angle decreases as the frequency increases. Similarly, as the elevation angle goes on increasing the path loss faced by the transmitted signal from 5G LEO satellite toward base station decreases which is shown in Figure 51.5. The minimum path loss is obtained at an angle of 90 . Figure 51.6 indicates the path loss according to the distance of 5G LEO satellite from ground base station. Minimum path loss obtained at 600 km while maximum path loss is obtained at 3,000 km. Hence, path loss increases as the

Link budget design for integrated 5G-LEO communication applications

651

Path loss vs. elevation angel

132 130 128

SNR (dB)

126 124 122 120 118 116

0

20

40

60 80 100 120 Elevation angel, phi (˚)

140

160

180

Figure 51.5 Path loss versus 5G-LEO satellite link elevation angle

–36

Received power vs. earth-to-satellite distance

–38

Lp (dB)

–40 –42 –44 –46 –48 –50 500

1,000 1,500 2,000 2,500 Earth-to-satellite distance, d (km)

3,000

Figure 51.6 Path loss versus 5G-LEO satellite distance

distance of the satellite from the ground base station increases. In the meanwhile, as the distance of the 5G LEO satellite increase from ground base station, the power received by the receiving antenna also decreases which is shown in Figure 51.7.

652

Advances in communications satellite systems 2: ICSSC-2019 –36

Received power vs. earth-to-satellite distance

–38

Pr (dB)

–40 –42 –44 –46 –48 –50 500

1,000 1,500 2,000 2,500 Earth-to-satellite distance, d (km)

3,000

Figure 51.7 Received power versus 5G-LEO satellite distance Maximum power is received by the receiving antenna at a distance of 600 km from ground base station while minimum power received by the receiving antenna at 2,800 km.

51.5 Conclusion A simplified link budget analysis for integrated 5G-LEO satellite communication is discussed in this chapter. The result analysis shows that as the distance of the 5G-LEO satellite from the terrestrial base station increases, the path loss of the transmitted signal from the satellite to the base station also increases with an attenuation to the received signal.

Acknowledgments The authors wish to thank the Niger Delta Development Commission of Nigeria for sponsoring this research at the Manchester Metropolitan University, UK through its scholarship scheme.

References [1] Luglio, M., Romano, S. P., Roseti, C., and Zampognaro, F. “Service delivery models for converged satellite-terrestrial 5G network deployment: A satellite-assisted CDN use-case.” IEEE Network. 2019; 33(1); 142–150.

Link budget design for integrated 5G-LEO communication applications

653

[2] Lin, J. “Synchronization requirements for 5G: An overview of standards and specifications for cellular networks.” IEEE Vehicular Technology Magazine. 2018; 13(3): 91–99. [3] Bai, L., Zhu, L., Zhang, X., Zhang, W., and Yu, Q. “Multi-satellite relay transmission in 5G: Concepts, techniques, and challenges.” IEEE Network. 2018; 32(5): 38–44. [4] Cioni, S., Gaudenzi, R. D., Herrero, O. D. R., and Girault, N. “On the satellite role in the era of 5G massive machine type communications.” IEEE Network. 2018; 32(5): 54–61. [5] Ekpo, S. C. and George, D. “Impact of noise figure on a satellite link performance.” IEEE Communications Letters. 2011; 15(9): 977–979. [6] Ekpo, S. C. “Parametric system engineering analysis of capability-based small satellite missions.” IEEE Systems Journal. 2019; 13(3): 3546–3555.

This page intentionally left blank

Section 13

New satellite components and transmitter and modem technologies

This page intentionally left blank

Chapter 52

Secret key agreement for satellite laser communications Hiroyuki Endo1 and Masahide Sasaki1

Free-space optical (FSO) communications offer high-capacity wireless transmission due to their plentiful properties originated from higher carrier frequency. These properties also yield a greater security advantage: the high directionality of the laser beam and the line-of-sight configuration can reasonably restrict an attack model launched by an eavesdropper (Eve). Secret key agreement over FSO links (FSO-SKA) employs this security advantage for key establishment between two distant parties, which is secure against Eve even with unbounded computer resources. In this chapter, we numerically evaluate the performance of FSO-SKA for satellite laser communications under the given power constraint. We also compare the performance of FSOSKA and quantum key distribution (QKD). Our result shows that FSO-SKA can generate a key even for the distance between geostationary orbit satellite and ground station. We anticipate that FSO-SKA has a potential to extend the secure satellite networks to global scale, which is hard only with QKD. Key Words: physical layer cryptography; secret key agreement; quantum key distribution; information theoretic security; generalized on off keying

52.1 Introduction Physical layer cryptography (PLC) [1,2] realizes secure communications against an eavesdropper even with unbounded computer resources (i.e., information-theoretic security), under the reasonable assumption on the attack model launched by the eavesdropper. The inherent natures of wireless communication channels are employed to satisfy this assumption. In the case of free-space optical (FSO) [3] communications, the line-of-sight (LoS) configuration and the high directionality of the laser beam make any suspicious attempts in the middle of the FSO link difficult. Thus, the eavesdropper is forced to tap the communication at the sidelobe of

1

National Institute of Information and Communications Technology, Koganei, Tokyo, Japan

658

Advances in communications satellite systems 2: ICSSC-2019

the beam. PLC in the FSO domains (FSO-PLC) can realize high-speed and information-theoretically secure communications thanks to these security advantages. FSO-PLC has been studied both theoretically [4–9] and experimentally [10–15] during the last decade. Quantum key distribution (QKD) [16–18] can be regarded as an extreme example of PLC. This scheme ensures the security against an eavesdropper even with any physically implementable attacks as well as unbounded computational resources. Unfortunately, recent theoretical studies clarified that an upper bound on the maximum achievable key rate of QKD is just a few times larger than the key rates of existing QKD protocols [19,20]. This is an urgent issue especially in satellite QKD. Numerical calculation [6] reveals that the transmission distance of current QKD schemes is limited to the distance between a low earth orbit (LEO) satellite and a ground station: QKD between a ground station and a medium earth orbit (MEO) or a geostationary orbit (GEO) satellite seems challenging. FSO-PLC can become an attractive option to overcome this difficulty. With the above-mentioned security advantages in FSO communications, the security assumption on QKD that the eavesdropper can be everywhere in the universe and can do everything is reasonably relaxed. Therefore, FSO-PLC is expected attaining higher key rate and longer transmission distance compared with QKD. The relation between FSO-PLC and QKD was first discussed and numerically examined in our previous paper [6], showing the possibility of the global scale network with information theoretic security by FSO-PLC. In this chapter, however, wiretap channel (WTCh) coding [21,22] is considered. WTCh coding and QKD have different purpose: WTCh coding aims at secure message transmission, while QKD aims at key establishment between distant two parties. Moreover, in QKD, the legitimate parties can use additional error-free channel—an authenticated public channel—to create a key from the random bit sequence shared over a quantum channel. For a fair comparison, FSO-PLC scheme aiming at key establishment should be considered. In the present chapter, instead of WTCh coding, we consider PLC scheme aiming at key establishment—secret key agreement—proposed by Maurer [23] and Ahlswede and Csisza´r [24], and numerically compare its performance with QKD. In this scheme, the legitimate parties share correlated randomness by observing a common randomness source or transmitting a random number sequence. After then, they create a key from the shared randomness via the discussion over an authenticated public channel, just like QKD. This scheme has some practical advantages on WTCh coding thanks to the public channel. The secret key rate of SKA is higher than the secure message rate of WTCh coding, as shown later. Moreover, legitimate parties can opportunistically select the “good” events, which is preferable to FSO communications suffering from atmospheric turbulence. We have demonstrated SKA over FSO links—FSOSKA—in a 7.8-km terrestrial FSO link testbed, showing that roughly a rate of 7 Mbps is possible for 10-MHz repetition rate [14,15]. Therefore, we are at the point to consider the implementation of FSO-SKA in satellite communications. The remainder of this chapter is organized as follows. In Section 52.2, we introduce SKA over WTCh model. In Section 52.3, channel model and the key generation

Secret key agreement for satellite laser communications

659

performance are specified. Section 52.4 is dedicated to numerical investigation of the behaviors of key generation performance. In particular, we compare the performances of SKA and WTCh coding. In Section 52.5, we compare the key generation performance of FSO-SKA and QKD. Finally, Section 52.6 summarizes the chapter.

52.2 Secret key agreement SKA is modeled by the WTCh model shown in Figure 52.1. Alice X transmits a random bit sequence (RBS) xn of length n to Bob Y over the main channel WB and to Eve Z over the wiretapper channel WE . Let yn and zn denote the RBSs output via the main and wiretapper channels, respectively. In general, yn and zn differ from xn due to the transmission errors. After sharing the correlated RBSs, Alice and Bob perform the signal processing for key creation via the public channel. This processing—called key-distillation processing—contains two steps. The first step is information reconciliation [23], in which the discrepancies between xn and yn are corrected by exchanging the information for error correction (correction information) over the public channel. The second step is privacy amplification [24], in which the information leaked to Eve is wiped out by compressing reconciled sequence. Regarding information reconciliation, we focus on the simple case: Bob discloses the correction information to Alice, and she retrieves yn from xn using the correction information as side information. This scheme is referred to as reverse reconciliation. Our previous experiment [15] showed that reverse reconciliation surpasses its “direct” counterpart where Alice discloses the correction information. The key generation performance of SKA is measured with the secret key rate RK defined as RK ¼ max½IðPA ; WB Þ  IðPB ; WBE Þ

(52.1)

PA

xn Alice X

yn Main channel WB Wi ret app er cha nne lW

Bob Y

zn

E

Eve Z

Authenticated public channel

Figure 52.1 Schematic diagram of WTCh model for SKA

660

Advances in communications satellite systems 2: ICSSC-2019

where I(P, W) denotes the mutual information between two random variables with an input probability distribution P and with an output probability via a channel W. PA and PB denote the probability distributions at Alice and Bob, respectively, and WBE denotes the channel between Bob and Eve. The first term represents the information rate reconciled via the reverse reconciliation and the second term represents the leaked information to Eve. The maximum of all achievable secret key rate is called secret key capacity CK. Deriving CK is still an open problem. Instead, (52.1) serves as an achievable lower bound below CK. So, this equation is enough to evaluate the performance of SKA.

52.3 Channel model 52.3.1 Generalized on-off keying As a modulation scheme appropriate for satellite laser communication, we consider generalized on-off keying (GOOK) scheme [25,26]. Alice transmits a laser pulse with a width of Dp and an average photon number of nA in each time slot D when x ¼ 1. Otherwise, she transmits a vacuum pulse. In contrast to usual OOK, she can adjust the pulse generation probability q ¼ PA ðx ¼ 1Þ so that the secret key rate is maximized. Besides the input probability q, the average photon number of nA also should be optimized. We should note that q and nA cannot be optimized independently: they should be optimized under the constraint that the average optical power over whole transmission must be lower than P. This constraint—referred to as power constraint—is formulated in our previous paper [6] as following: q

nA hfc P D

(52.2)

where h is Planck’s constant and fc is the carrier frequency which is set to be 200 THz in the present chapter. Bob and Eve receive the optical pulse with on-off photon counting detectors with dark count rates (DCRs) of lB and lE, respectively. The unit of DCR is count per second (cps). They determine y ¼ 1 (or z ¼ 1) when the detector fires. Otherwise, they set y ¼ 0 (or z ¼ 0). The loss in main and wiretapper channels is characterized by the channel transmittance hB  1 and hE  1, respectively. Detector efficiencies are renormalized into these channel transmittances.

52.3.2 Secret key rate for GOOK The secret key rate of SKA over GOOK channel is obtained by solving the following optimization problem: RK ¼ max½IðPA ; WB Þ  IðPB ; WBE Þ q;nB

subject to q

nB hfc  hB P D

(52.3)

Secret key agreement for satellite laser communications

661

which is the power constraint translated in terms of nB ¼ hBnA, the average photon number of received pulse at Bob. To derive a closed form expression of this maximization is difficult. So, we perform numerical optimization. The mutual information in (52.3) are calculated as IðPA ; WB Þ ¼

X X

PA ðxÞWB ðyjxÞlog2

x¼0;1 y¼0;1

IðPB ; WBE Þ ¼

X X

WB ðyjxÞ S PA ðx0 ÞWB ðyjx0 Þ x0

PB ðyÞWBE ðzjyÞlog2

y¼0;1 z¼0;1

WBE ðzjyÞ Syx PB ðx0 ÞWBE ðzjy0 Þ

(52.4)

(52.5)

We should specify the transition probabilities WB(y|x) and WBE(z|y) as functions of nB to evaluate these quantities. We assume that the main channel is a two-input– two-output discrete memoryless channel. For such channel model, the transition probability WB(y|x) is calculated based on Poisson probability distribution as WB ð0j0Þ ¼ elB D

(52.6)

WB ð0j1Þ ¼ eðnB þlB DÞ

(52.7)

Obviously, WB ð1j0Þ ¼ 1  WB ð0j0Þ and WB ð1j1Þ ¼ 1  WB ð0j1Þ. Similarly, the transition probability of the wiretapper channel WE ðzjxÞ, which is also assumed to be a two-input–two-output discrete memoryless channel, is given as WE ð0j0Þ ¼ elE D

(52.8)

WE ð0j1Þ ¼ eðnB hEB lE DÞ

(52.9)

To evaluate ðPB ; WBE Þ, however, the channel WBE from Bob to Eve should be specified. The transition probability of this channel can be written as X e B ðxjyÞ WBE ðzjyÞ ¼ WE ðzjxÞW (52.10) x¼0;1

eBW e from Bob to e B ðxjyÞ is the transition probability of the channel W where W e Alice. Since this channel is the inverse channel of WB, W B ðxjyÞ is calculated by the Bayes’ rule as e B ðxjyÞ ¼ WB ðyjxÞPA ðxÞ W PB ðyÞ P where PB ðyÞ ¼ x WB ðyjxÞPA ðxÞ.

(52.11)

52.4 Numerical investigation of secret key rate First, we compare secret key rate RK —key generation performance of SKA—and secure message rate RM —secure transmission performance

662

Advances in communications satellite systems 2: ICSSC-2019

of WTCh coding. Secret message rate RM for OOGK channel is calculated as [6] RM ¼ max½IðPA ; WB Þ  IðPA ; WE Þ q;nB

(52.12)

where the maximization is taken subject to the power constraint. The difference between RK and RM is in the second term which corresponds to the leaked information. In SKA based on reverse reconciliation, Alice and Bob share Bob’s RBS yn. Thus, the information leakage is measured by the mutual information between Bob and Eve (PB, WBE). In WTCh coding, on the other hand, Bob should retrieve Alice’s RBS x . Thus, the information leakage is measured by the mutual information between Alice and Eve (PA, WE). We should note that the inequality IðPA ; WE Þ  IðPB ; WBE Þ

(52.13)

holds for the second terms in these formulae, because the errors occurring in channel WBE are the superposition of errors in main channel WB and in wiretapper channel WE. Therefore, the relation RK  RM immediately follows. Similarly IðPA ; WB Þ  IðPB ; WBE Þ

(52.14)

also holds. Therefore, RK always has nonzero positive value, while RM can be lower than zero especially when lB  lE . Figure 52.2 shows the secret key rate RK (blue lines) and the secure message rate RM (red lines) as functions of the main channel attenuation h1 B —the inverse of the main channel transmittance—for various transmittance ratio hEB. We here assume that Bob’s detector has a large DCR, whereas Eve’s detector is almost DCR-free. We further assume that the transmission ratio hEB ¼ hE =hB is fixed, which corresponds to the scenario that the footprint of the laser beam is physically secured and Eve wiretaps from the sidelobe. As discussed above, RK is always larger than RM for the same hEB and has nonzero positive values. In contrast, RM rapidly decreases as hEB is close to 1 and drastically falls to 0 at a certain threshold of h1 B . Surprisingly, RK achieves a few Mbps even when hEB is larger than 10, namely, the case that the received power at Eve is ten times larger than that at Bob. This result shows a great advantage of SKA on WTCh coding. As discussed in our previous paper [6], these rates can be characterized by two regions. For smaller h1 B , the rates keep unchanged. We refer to this region as the loss-independent region. For larger h1 B , on the other hand, they monotonically decrease as h1 B increases. We refer to this region as the noise-limited region. In Figure 52.3, to further understand the behaviors characterized by two regions, we summarize the optimal parameters—the optimum photon number per received pulse nB at Bob and the optimum input probability q*—that give the secret key rate RK. As with RK, hB is unchanged in the loss-independent region and slightly decreases in the noise-limited region. This value is determined under the

Secret key agreement for satellite laser communications 10 M Rate (bps)

Rate (bps)

1G 100 M 10 M 1M 60

663

1M 100 k 10 k 110

70 80 90 (dB) Channel attenution η–1 B

120 130 140 Channel attenution η–1 (dB) B

Rate (bps)

1G

1M

WTCh, ηEB = 0.5 WTCh, ηEB = 0.99 SKA, ηEB = 0.5

1k

SKA, ηEB = 0.99 SKA, ηEB = 10

1

0

30

60

90

120

150

180

Channel attenution η–1 (dB) B

Figure 52.2 Secret key rates RK (red lines) for SKA and the secure message rate RM (blue lines) as functions of the main channel attenuation h1 B . The solid, dashed, and chained lines correspond to the case for hEB ¼ 0.5, 0.99, and 10, respectively. The upper left and right insets are the magnifications for the loss-independent and the noise-limited regions. Parameters: P ¼ 10 mW, lB ¼ 1 kcps, lE ¼ 1 cps, and D ¼ 1 ns trade-off between the transmission error and the information leakage: if nB becomes large, transmission error decreases, whereas the information leakage to Eve increases. Actually, hB decreases as hEB increases. The optimum input probability q* is roughly at 0.5 in the loss-independent region. In this region, the product of q* and n* is constant and the power constraint in (52.5) is satisfied with strict inequality. This implies that Alice regulates the input power to avoidable information leakage. In the noise-limited region, q* monotonically decreases as h1 B increases, meaning that the power constraint is satisfied with equality. In this situation, Alice should decrease q* to keep nB optimum.

52.5 FSO-SKA versus QKD In Figure 52.4, we compare the secret key rate of QKD, the secure message rate RM of WTCh coding, and the secret key rate RK of SKA. The curve labelled with “Decoy BB84” shows a secure key rate of BB84 protocol [27] employing the

Advances in communications satellite systems 2: ICSSC-2019 10

Optimum input probability q*

Optimum photon number n*B

664

1 0.1 0.01 0.001

(a)

SKA, ηEB = 0.5 SKA, ηEB = 0.99 SKA, ηEB = 10

0

30

60 90 120 150 Channel attenution ηB–1 (dB)

180

1

10–3

10–6 SKA, ηEB = 0.5 SKA, ηEB = 0.99 SKA, ηEB = 10

10–9

(b)

0

30

60 90 120 150 Channel attenution ηB–1 (dB)

180

Figure 52.3 Optimum parameters for the secret key rate RK in Figure 52.2. (a) Optimum average photon number n* of the received pulse at Bob. (b) Optimum input probability q* Channel capacity (without eve) SKA, ηEB = 0.99 SKA, ηEB = 10 WTCh, ηEB = 0.99

1M

B8

yB

co

De

1k

1

4

Key rate or secrecy rate (bps)

1G

0

30

60 90 120 150 Main channel attenution αB (dB)

180

Figure 52.4 Comparison of the secret key rate RK (blue solid line), the secret message rate RM of WTCh coding (red dashed line), the channel capacity without Eve (green solid line), and secure key rate of decoy BB84 protocol (green chained line). Parameters for GOOK: P ¼ 10 mW, lB ¼ 1 kcps, lE ¼ 1 cps, and D ¼ 1 ns. Parameters for QKD: pulse generation rate ¼ 1 GHz and DCR of a detector ¼ 100 cps decoy-pulse. As with GOOK, we assume an ideal linear attenuation channel, 1 GHz pulse repetition rate, and 100 cps DCR, which is a typical value in the current QKD systems. This figure indicates that the secure key rate rapidly drops at h1 EB ¼ 40 dB, which roughly corresponds to the best link budget for a LEO to ground distance in FSO communications. The curve labeled with “WTCh coding” is the secure message rate RM. It can cover a wider range in which QKD hardly generates the key even for the high transmittance ratio, hEB ¼ 0.99. However, it dramatically drops at h1 B ¼ 120 dB. As opposed to these schemes, SKA can generate a key even at h1 EB ¼ 120 dB. Moreover, key generation can be realized when Eve can receive more intense light

Secret key agreement for satellite laser communications

665

than Bob. Consequently, SKA is a promising candidate for realization of the global scale secure network based on FSO communications.

52.6 Conclusion In this chapter, we numerically investigated the secret key rate of FSO-SKA over GOOK. Our results show that this scheme can generate key even when Eve has an almost DCR-free detector and the received power at Eve is much higher than that at Bob. We also show that FSO-GKA can be implemented in GEO-Ground communications. We anticipate that FSO-GKA is a promising scheme that extends the secure satellite network to continental or global scale.

Acknowledgments This work was funded in part by JSPS KAKENHI Grant Numbers 17H01281 and 19K14992. This work was also partly supported by “Research and Development of the Quantum Cryptography Technology for Satellite Communications” in “Research and Development of Information and Communications Technology” of Ministry of Internal Affairs and Communication (MIC), Japan.

References [1] Zhou, X., Song, L., and Zhang, Y. Physical layer security in wireless communications. Boca Raton, FL: CRC Press; 2013. [2] Schaefer, R. F., Boche, H., Khisti, A., et al. Information theoretic security and privacy of information systems. Boston, MA: Cambridge University Press; 2017. [3] Khalighi, M. A., and Uysal, M. “Survey on free space optical communication: A communication theory perspective.” IEEE Communications Surveys & Tutorials. 2017; 16(4): 2231–2258. [4] Wang, N., Song, X., Cheng, J., et al. “Enhancing the security of free-space optical communications with secret sharing and key agreement.” Journal of Optical Communications and Networking. 2014; 6(12): 1072–1081. [5] Lopez-Martinez, F. J., Gomez, G., and Garrido-Balsells, J. M. “Physicallayer security in free-space optical communications.” IEEE Photonics Journal. 2015; 7(2): 7901014. [6] Endo, H., Han, T. S., Aoki, T., et al. “Numerical study on secrecy capacity and code length dependence of the performances in optical wiretap channels.” IEEE Photonics Journal. 2015; 7(5): 7903418. [7] Sun, X., and Djordjevic, I. B. “Physical-layer security in orbital angular momentum multiplexing free-space optical communications.” IEEE Photonics Journal. 2016; 8(1): 7901110.

666

Advances in communications satellite systems 2: ICSSC-2019

[8] Zou, D., and Xu, Z. “Information security risks outside the laser beam in terrestrial free-space optical communication.” IEEE Photonics Journal. 2016: 8(5): 7804809. [9] Chen, C., and Yang, H. “Shared secret key generation from signal fading in a turbulent optical wireless channel using common-transverse-spatial-mode coupling.” Optics Express. 2018; 26(13): 16422–16441. [10] Endo, H., Fujiwara, M., Kitamura, M., et al. “Free-space optical channel estimation for physical layer security.” Optics Express. 2016; 24(8): 8940–8955. [11] Wang, T., Gariano, J. A., and Djordjevic, I. B. “Employing Bessel–Gaussian beams to improve physical-layer security in free-space optical communications.” IEEE Photonics Journal. 2018; 10(5): 7907113. [12] Wang, T., and Djordjevic, I. B. “Physical-layer security of a binary data sequence transmitted with Bessel–Gaussian beams over an optical wiretap channel.” IEEE Photonics Journal. 2018; 10(6): 7908611. [13] Ji, J., Huang, Q., Chen, X., et al. “Performance analysis and experimental investigation of physical-layer security in OCDMA-based hybrid FSO/fiber wiretap channel.” IEEE Photonics Journal. 2019; 11(3): 7903420. [14] Fujiwara, M., Ito, T., Kitamura, M., et al. “Free-space optical wiretap channel and experimental secret key agreement in 7.8 km terrestrial link.” Optics Express. 2018; 26(15): 19513–19523. [15] Endo, H., Fujiwara, M., Kitamura, M. et al. “Free space optical secret key agreement.” Optics Express, 2018: 26(18): 23305–23332. [16] Bennett, C. H., and Brassard, G. “Quantum cryptography: Public key distribution and coin tossing.” Proceedings of IEEE International Conference on Computers, Systems and Signal Processing. December 1984. pp. 175–179. [17] Ekert, A. K. “Quantum cryptography based on Bell’s theorem.” Physical Review Letters. 1991; 67(6): 661–663. [18] Gisin, N., Ribordy, G., Tittel, W., et al. “Quantum cryptography.” Reviews of Modern Physics. 2002; 74(1): 145–195. [19] Takeoka, M., Guha, S., and Wilde, M. M. “Fundamental rate-loss tradeoff for optical quantum key distribution.” Nature Communications. 2014; 5: 5235. [20] Pirandola, S., Laurenza, R., Ottaviani, C., et al. “Fundamental limits of repeaterless quantum communications.” Nature Communications. 2017; 8: 15043. [21] Wyner, A. D. “The wire-tap channel.” Bell System Technical Journal. 1975; 54(8): 1355–1387. [22] Csisza´r, I., and Ko¨rner, J. “Broadcast channels with confidential messages.” IEEE Transactions on Information Theory. 1978; 24(3): 339–348. [23] Maurer, U. M. “Secret key agreement by public discussion from common information.” IEEE Transactions on Information Theory. 1993; 39(3): 733–742. [24] Ahlswede, R., and Csisza´r, I. “Common randomness in information theory and cryptography. I. Secret sharing.” IEEE Transactions on Information Theory. 1993; 39(4): 1121–1132. [25] Boroson, D. M. “A survey of technology-driven capacity limits for freespace laser communications.” Proceedings of SPIE. 2007; 6709: 670918.

Secret key agreement for satellite laser communications

667

[26] Waseda, A., Sasaki, M., Takeoka, M., et al. “Numerical evaluation of PPM for deep-space links.” IEEE Journal of Optical Communications and Networking. 2011; 3(6): 514–521. [27] Lo, H.-K., Ma, X., and Chen, K. “Decoy state quantum key distribution.” Physical Review Letters. 2005; 94(23): 230504.

This page intentionally left blank

Chapter 53

Methods for securing spacecraft tasking and control via an enterprise Ethereum blockchain David Hyland-Wood1,2, Peter Robinson1,2, Roberto Saltini3, Sandra Johnson1 and Christopher Hare1

Integration of space-based communications infrastructure within 5G networks presents specific challenges for spacecraft, namely a necessary rationalization of currently patchy communications security and the assurance of identity when conducting high-level spacecraft tasking and control operations. This chapter presents approaches to addressing both issues via the deployment of an enterprise Ethereum blockchain modified with a consensus algorithm appropriate for access by spacecraft. We discuss the applicability of enterprise Ethereum blockchains to the problem of spacecraft communication security, analyze the properties of blockchain consensus algorithms suitable for use with spacecraft, and suggest information architectures to allow secure spacecraft integration into 5G networks. Key Words: blockchain; Ethereum; spacecraft; communications; security

53.1 Introduction Communication security for existing spacecraft has been incompletely implemented, leaving significant attack vectors related to spacecraft control*. Communications security has been widely deployed for military satellites, newer telecommunications satellites in geosynchronous orbit, and newer deep space probes. Relatively few of the new breed of CubeSats and other small satellites in low Earth orbit (LEO) have launched with fully encrypted communications, 1

PegaSys, ConsenSys AG, Brisbane, Australia School of Information Technology and Electrical Engineering, The University of Queensland, Brisbane, Australia 3 PegaSys, ConsenSys AG, Sydney, Australia * Based upon personal discussions with industry participants; few are willing to publicly acknowledge communication security limitations. 2

670

Advances in communications satellite systems 2: ICSSC-2019

including on channels used for spacecraft control. Practical exploitation of such lax security measures has been limited due to the relative complexity and cost of satellite ground stations. The recent implementations of ground-station-as-a-service offerings from Amazon and other vendors have slashed the costs of ground station access and exposed such communications vulnerabilities. Motivations to close security vulnerabilities on satellites are generally synonymous with motivations for securing services on the public Internet. Attacks may be conveniently separated into two types: attempts to gain unauthorized control (colloquially known as “hacking”) and attempts to deny service (“jamming” in the context of radio communications [1]). Spacecraft and ground-based systems that control them are at risk of both active hacking and denial-of-service attacks. Although few spacecraft operators publicly acknowledge cybersecurity incidents, governmental transparency regulations in the United States have allowed evidence of some incidents to be acknowledged. Examples include attacks by Chinese state actors that led to unauthorized access to “networks that control spacecraft” at NASA JPL [2] and acknowledgment that U.S. Air Force satellites are “jammed by commercial equipment easily acquired by state and nonstate actors” [3]. One can reasonably assume that commercial satellite operators and space assets controlled by other national governments have had and continue to face similar challenges. Integration of satellites into 5G networks would further ease network access to those assets and increase potential for software bugs due to the amount of the technical stack implemented in software. Both aspects increase pressure to provide adequate communication security measures. This chapter presents an approach to securing spacecraft command communications against attempts to gain unauthorized control. Antijamming techniques are beyond the scope of this chapter.

53.2 Literature review Researchers with government, especially military, connections in China, Russia, and the United States are actively investigating uses of blockchains for access security and data integrity of Earth orbiting satellites [4–9]. Blockchains are essentially distributed, append-only databases of transactions, and are considered a strict subset of distributed ledgers [10]. Administrative responsibility and trust are typically shared among the operators of blockchain nodes. As opposed to traditional database architectures, blockchains (of whatever form) require consensus to be formed among participants for information to be added. Information is collected into blocks which are cryptographically linked to preceding blocks. These features combine to make the blockchain tamper-evident. In other words, a key property of blockchains is that they are difficult to compromise because a successful attack requires the attacker to be able to successfully gain control of many participants. Published Chinese military interest seems to be focused on preventing “cyber and physical attacks” [5] against space assets, and to allow “multiple departments to participate in the maintenance and update of equipment status” [11].

Securing spacecraft tasking and control via Ethereum blockchain

671

The Roscosmos State Corporation for Space Activities in Russia is developing a “digital platform for control spacecrafts” [sic] and for “use of ground stations” focused on the Roscosmos orbital group of satellites based on blockchains [9]. Existing public literature does not specify in detail how these systems are intended to function. One might reasonably anticipate that best practices from the field of computer science could be borrowed and extended to secure spacecraft communications. Relevant approaches used to control remote access to cloud computing resources and weapon systems include multifactor authentication [12] and multiparty authorization [13]. A blockchain could readily serve as a source for multiparty authorization in the case where its nodes are operated by multiple parties and they are required to form consensus on an impending action. The notional heart of a blockchain is its consensus algorithm [10]. A blockchain consensus algorithm defines the steps necessary for blockchain participants to agree on information to be added to the distributed ledger. It is how the nodes in the network agree (come to consensus on) the next block to be added to the chain. Existing blockchain consensus algorithms have been recognized by many as limiting the applicability of blockchains to space system operations (e.g., [5–7,11,14]). Several researchers have suggested the applicability of Ethereum as a possible blockchain framework but noted that the consensus algorithms currently used on the public production Ethereum blockchain (known as “Ethereum MainNet”) are inappropriate for use in space system operations [6,7,15]. Neither the traditional proof of work (PoW) algorithm nor the forthcoming proof of stake (PoS) consensus algorithm used on Ethereum MainNet or its public test networks provide the properties needed for space system operations. Ethereum PoW is intentionally designed to be computationally intensive. Limitations of computation, power, and heat rejection capabilities make the adoption of PoW consensus algorithms impractical. It is clearly undesirable for spacecraft to perform unnecessary computation, especially when those computations may not result in a successful transaction. Ethereum PoS relies upon the blockchain having an economically meaningful cryptocurrency to be used for internal operations. Changing the consensus algorithm of an Ethereum system creates a blockchain that is incompatible with Ethereum MainNet, at least under the current state of the art. Those taking this path (e.g., the Enterprise Ethereum Alliance [16] and its members) are thus proposing “private”, “consortium,” or “enterprise” Ethereum blockchains with consensus algorithms and perhaps other properties they deem appropriate for operations in their contexts [17]. The authors follow by proposing blockchain properties appropriate for securing spacecraft communications. Three groups have suggested the practical Byzantine fault tolerance (PBFT) algorithm [18] as a possible consensus algorithm for near-Earth orbital space system operations [5,7,15]. One presumes those researchers meant to suggest PBFT as modified for use as a blockchain consensus algorithm, for example, Istanbul Byzantine fault tolerance (IBFT) [19] or IBFT 2.0 [20] since PBFT was not itself defined with blockchains in mind. PBFT is a so-called proof of authority (PoA) algorithm, in that certain network nodes, called validators, are given authority to act as proxies for many other nodes.

672

Advances in communications satellite systems 2: ICSSC-2019

The PBFT family of consensus algorithms is an imperfect fit for blockchains nodes running on orbital spacecraft given their reliance on time and connectivity as critical algorithmic components. Timeouts resulting from communications delays, occultation, radio interference, and other communication disruptions are all too common with spacecraft but are used to determine error conditions in the PBFT family of consensus algorithms. PBFT algorithms would be an even less perfect fit for deep space operations where such communication disruptions are routinely expected. PBFT message sizes also tend to be large in practice, which works against the bandwidth, processing, and storage capabilities of most extant and proposed spacecraft. However, blockchains using PBFT algorithms located terrestrially may still be a reasonable component of integrated space systems. Current spacecraft are highly isolated following launch. One option may be to treat them as authorities for their own transactions since access to their private keys may be strictly controlled. Physical access to a spacecraft following launch is currently quite unlikely, and detection of such an event is likely to be noticed or could be designed into new architectures. This is an argument for the use of an algorithm in the PoA family. PoA protocols may be conveniently separated into those that guarantee immediate finality of created blocks, those that guarantee eventual finality, and those that do not guarantee finality. A block is said to be final only if it has been added to the blockchain of an honest node and both its position and content may not be changed under any future circumstance (e.g., a network fork, or a rebalancing of the blockchain at a later time). A blockchain consensus protocol provides immediate finality only if any block is final as soon as it is added to the blockchain provided that there are no more faulty nodes than the maximum threshold allowed by the protocol. In other words, immediate finality guarantees that the blockchain cannot fork. A blockchain consensus protocol guarantees eventual finality if blocks become final only after they have been on the chain for “sufficient long” time. In the case that blocks added to the chain never become final, such as in the Clique PoA protocol [21], then the consensus protocol is said not to guarantee finality. Another property to be considered when choosing PoA protocols, and consensus protocols more in general, is whether they are tolerant to nodes that may act maliciously; such nodes are commonly called Byzantine [22]. The number of Byzantine nodes that a protocol can withstand is called BFT tolerance. Finally, the type of network that a consensus protocol is designed for must be considered. Networks can be divided into three types: synchronous, asynchronous, and partially synchronous [23]. The maximum message delay in synchronous networks is bounded by a known amount of time, and unbound in asynchronous networks. In partially synchronous networks, either the maximum message latency is finite but unknown or the network is guaranteed to reach a state of synchrony in a finite time after experiencing an initial state of asynchrony. We argue that the properties of immediate finality, BFT, and the ability to operate on partial synchronous networks are very compelling features for space system operations because they allow for immediate reads of the information known to be on the chain, provide a high level of security, and allow for patchy

Securing spacecraft tasking and control via Ethereum blockchain

673

communication which is usually the case for space communication. The Enterprise Ethereum Alliance is considering adoption of immediate finality and BFT in their client standard [24]. Naturally, spacecraft in Earth orbit suffers less from extreme time delays or communication loss than those in deep space. It is possible, and probably preferred, to treat orbiting spacecraft as edge devices instead of expecting them to run blockchain node software. In such a scenario, spacecraft may read transactions from, or propose transactions to, a blockchain operated terrestrially. The property of immediate finality is still anticipated to be a requirement for terrestrially operated blockchains supporting space systems operations. PoA protocols that provide immediate finality include Honey Badger [25], Tendermint [26], DBFT (Democratic BFT) [27], and IBFT 2.0 [20]. Honey Badger [25] ensures immediate finality but relies on a probabilistic common coin protocol to ensure liveness. Liveness is a property that guarantees that transactions submitted to the system will eventually be included in the blockchain. Therefore, the addition of new blocks to the local blockchain of honest nodes is a probability function. This makes Honey Badger better suited to asynchronous networks than to networks that can guarantee partial synchrony. Tendermint [26] was primarily designed as PoS blockchain consensus protocol but can be easily adapted to fit the PoA category. This would produce a PoA blockchain consensus protocol with immediate finality that ensures liveness under the assumption that messages are always eventually delivered. DBFT [27] ensures optimal Byzantine-fault-tolerant persistence and liveness under the assumption that the network is reliable, that is, all messages that are sent are always eventually delivered. The IBFT 2.0 protocol is an iteration of the IBFT protocol [19] that was developed around early 2017 by AMIS Technologies [19] and was fully implemented in Quorum [28] by around November 2017. IBFT was created to provide an alternative consensus protocol for the Ethereum blockchain that was better suited for either private or consortium blockchains, where deterministic immutability of the blockchain is often a requirement and spending significant computational effort is less desirable. However, as proven by Saltini and Hyland-Wood [29,30], IBFT is neither safe nor live when operated in eventually synchronous networks. IBFT 2.0 was developed as a response to these findings to provide a BFT PoA blockchain consensus protocol with immediate finality that is safe and live in partially synchronous networks which may experience an initial period of asynchrony where messages may be lost. Tendermint, IBFT, and IBFT 2.0 adapt the state machine replication protocol (PBFT) for application on a blockchain. IBFT and IBFT 2.0 integrate the PBFT concepts with the dynamic validator set voting mechanism originally designed for Clique to allow validators to be added to, or removed from, the set. We deliberately decided not to include Algorand [31] in this list as Algorand provides only probabilistic finality under some conditions. Forks are possible in the case where a network experiences periods of asynchrony. Such forks may eventually be resolved once a network becomes synchronous again. Given that

674

Advances in communications satellite systems 2: ICSSC-2019

spacecraft communication is more likely to be partially synchronous or asynchronous rather than synchronous, we suggest that Algorand is not appropriate for spacecraft blockchain communication. Mandl, at NASA Goddard Space Flight Center, has proposed using smart contracts on blockchains to create tasking services abstracted from which space-based assets might fulfil a sensing requirement (a “Remote Sensing as a Service” offering) [14]. In Mandl’s conception, a single Earth observation requirement could be obtained by multiple platforms conducting multiple observations under a variety of conditions until a desired goal is achieved. Such a service would allow a space systems operator to state an abstract goal (e.g., image a given area on Earth between given times on a given day in given frequencies), and for one or more spacecraft to cooperate without explicit coordination to fulfill that abstract requirement. Any spacecraft capable of reading the portion of the requirement that remains unfilled has the capability to both satisfy all or a portion of the requirement and has the capability to pass the results of its efforts to fulfill the smart contract could participate. Such a service would allow satellites owned and operated by different agencies, companies, or militaries to fulfill portions of a requirement without needing to communicate. Many such services could be simultaneously deployed to one or more blockchains.

53.3 Methodology It seems reasonable to ask why blockchains should be considered to address spacecraft communication security at all. We, therefore, begin by ensuring that we match stated domain requirements to a theoretical framework for blockchain applicability. Several researchers (e.g., [32,33]) have proposed decision trees to help determine the applicability of blockchains to particular domains. We will follow Wu¨st and Gervais [32] to suggest blockchain properties that could be used to satisfy the stated goals of [4–9]. Following the choice of general blockchain properties, we will compare PoA consensus protocols with the property of immediate finality to determine which algorithms are most appropriate for the stated goals. We will focus on comparing the various protocols by looking at their message traffic volumes and average message sizes with an aim to reduce both metrics. Several logical blockchain-satellite relationships are possible, and each relationship implies a different overall systems architecture. Figure 53.1 illustrates the various relationships between satellite and blockchain. These possible relationships imply very different computation capacity and radio bandwidth onboard spacecraft, especially due to overall message traffic and average message size. We will suggest a satellite-blockchain relationship suitable for the stated domain. Finally, we will address the communications security requirement by proposing to use an enterprise Ethereum blockchain as a proxy for multiparty agreement that a command should be executed. We suggest that partially or fully automated multifactor authentication and/or multiparty authorization is achievable with minimal software changes to deployed satellites.

Securing spacecraft tasking and control via Ethereum blockchain

675

(a)

(b)

(c)

(d)

Figure 53.1 Possible blockchain-satellite relationships: (a) satellite as a regular blockchain node; (b) satellite as a mining (validator) blockchain node; (c) Satellite may read from the blockchain; and (d) satellite requests transactions to be written to the blockchain

53.4 Results Following Wu¨st and Gervais (see the flowchart presented in Figure 53.1), we evaluated the applicability of a blockchain space systems operations by analyzing:

676

Advances in communications satellite systems 2: ICSSC-2019 The need to store state. We determined that a logged history of commands issued to spacecraft was desirable in order to allow for the possibility of future audits. Command logs are generally kept, although not generally in blockchains. The need for multiple writers. We determined that although extant spacecraft are generally controlled by single organizations, multiple operators are most often involved in issuing commands. The availability of a trusted third party that is always online. Although extant spacecrafts are indeed controlled by third parties (their owners and operators), and blockchains make sense “when multiple mutually mistrusting entities want to interact and change the state of a system” [32], we contend that the controlling authorities should not, in fact, be trusted due to poorly controlled information technology environments currently in place. The removal of inappropriate trust is exactly the problem we are attempting to solve. Are all writers known? We assumed that all spacecraft operators are known and can be provided with permission to fulfill their roles. Are all writers trusted? We assumed that an untrusted hacker could penetrate networks that control spacecraft.

Our contention that the owners and operators of spacecraft should not be trusted by the information system may require some further explanation: It is not that the controlling entities are not trustworthy in relation to their own spacecraft but that the information system cannot know which parties to trust in a poorly structured or maintained IT environment. In fact, the U.S. Department of Homeland Security recommends that we not trust network participants because any such trust “could result in long-term exposure without detection” [34]. We therefore believe that a blockchain may be used to reduce the available trust boundaries to something more manageable in the face of incomplete IT security based on the above, the process of Wu¨st and Gervais argues for the use of a private (or consortium) blockchain with user permissions to satisfy the stated use cases. We argued in the literature review that PoA consensus protocols with the property of immediate finality are of most apparent applicability to space system operations due to the immediate clarity of information flow. Such a choice has become commonly mirrored in enterprise blockchain deployments, especially among banks, insurance companies, and other organizations with an expectation of immediate finality in their IT systems. We now turn to the evaluation of PoA consensus protocols with the property of immediate finality in relation to the stated goals of [4–9]. Table 53.1 lists PoA algorithms with immediate finality and compares their total message traffic, their average message sizes, and their resilience to lost messages. Entries for the columns related to messages are given in Big O notation [35], in which the exact values are unknown but the values are limited by some (possibly large) integer multiple of the stated formulae. The number of nodes in a blockchain network participating in consensus formation is represented by n, the size of committee of validators is represented by k, and the size of a given block is represented by b.

Securing spacecraft tasking and control via Ethereum blockchain

677

Table 53.1 Proof of authority consensus algorithms with immediate finality Algorithm

Message traffic to reach consensus

Average message size

Resilience to lost messages

Honey Badger Tendermint DBFT IBFT 2.0

O(n2)

O(b)

No

O(n2) O(n2) O(n2)

O(b) O(b) O(b)

No No Yes

Table 53.2 Comparison of possible blockchain-satellite relationships Architecture

Message traffic to/from satellite

Changes needed to onboard software

Regular node Mining node Read-only Write request

High High Low Low

High High Low–medium Medium

Table 53.1 suggests that all algorithms have the same order of message traffic and average message size. Given the resilience of IBFT 2.0 toward lost messages, we suggest this algorithm as optimal in the state of the art for our use cases. Our choice of blockchain is therefore narrowed to a private (or consortium) blockchain with user permissions using the IBFT 2.0 consensus algorithms. We next analyzed each of the possible blockchain-satellite relationships shown in Figure 53.1. Table 53.2 summarizes the message traffic to and from a satellite in each of the four relationships to a blockchain and estimates the degree of operational complexity inherent in changing satellite software for each relationship. Running a blockchain node is nontrivial in terms of computation, memory, storage, power consumption, and communications. Limitations in computation, memory, and storage onboard extant satellites alone would argue against upgrading existing systems to run blockchain node software. We do note that some new systems have been specially designed for that purpose. There seems to be little reason to command existing satellites from other satellites specifically for the purpose of securing their communications, and therefore little reason to modify extant satellites to propose blockchain transactions for the considered use cases. By far the most compelling relationship between a satellite and a terrestrial blockchain is for the satellite to read information from the blockchain, as illustrated in Figure 53.1(c) and compared in Table 53.2. The costs for such read operations are minimal in terms of both the message traffic and the degree of software changes. We, therefore, focused on the evaluation of technical solutions in which Earth

678

Advances in communications satellite systems 2: ICSSC-2019

orbiting satellites would act as edge devices for a terrestrial blockchain, with a read-only relationship. Two possible ways to secure satellite communication may be borrowed from experiences with securing cloud computing and weapons systems, as mentioned in the literature review: multifactor authentication and multiparty authorization. Either may be used to secure edge devices by using call-backs to secure external information. Multifactor authentication is used to ensure that a user is who they say they are. For example, one may provide credentials to log onto their bank’s IT systems, and then subsequently be asked to confirm their login via an email, message to their mobile phone, or use of a separate hardware token. The second, hopefully independent, confirmation of their identity significantly increases the challenges facing a remote attacker attempting to gain unauthorized access. Similarly, multiparty authorization requires a separate party to validate an operation you wish to perform before you are allowed to proceed. In the case of your banking system, your bank may wish to confirm an attempt to close a joint account with the other owner before taking action. Figure 53.2 illustrates the communication paths for simple command passing (Figure 53.2(a)) and for either the multifactor authentication or multiparty authorization scenarios. Step 2 may be inserted as an additional check (a confirmation of either an authentication or authorization) prior to command execution. The use of an enterprise Ethereum blockchain as a confirming system allows for some interesting and useful concepts to be employed. A blockchain is a naturally distributed system that must come to consensus on new information in order to operate. The addition of arbitrary smart contract execution allows users of an Ethereum blockchain to encode whatever business logic is appropriate for a given use case. A smart contract may be written to require actions taken by blockchain users, off-blockchain processes, other smart contracts, or any combination thereof. There is no theoretical limit to the business logic that may be so encoded (although implementations clearly have many practical limitations, e.g., inability of hardware or operating system to execute business logic with high algorithmic complexity). 3

3

1 (a)

1

2

(b)

Figure 53.2 Communications execution: (a) in common usage, a command is sent (1) and executed onboard (3) and (b) using multifactor authentication or multiparty authorization: a command is sent (1), validated via a blockchain read (2), then executed only if validated (3)

Securing spacecraft tasking and control via Ethereum blockchain

679

For example, a smart contract may be written so a command destined for a spacecraft will not be echoed to the blockchain until an authenticated user confirms the validity of a command issued to a satellite (multifactor authentication) or multiple authenticated users confirm the command’s validity (multiparty authorization). Commands may also be checked for correctness of form, usefulness in an operational context, or any other automated checks that may be encoded in a smart contract. Adjusting onboard satellite software to read from a remote system prior to command execution should require minimal changes for those systems that allow for remote software updates. The bulk of the work to implement multifactor authentication and/or multiparty authorization would fall to a terrestrial blockchain, where it can be reached, extended, maintained, and managed. Implementation of additional computation, communication, implementation, etc, for the purpose of improving cybersecurity is often and rightly viewed as an economic cost. It is therefore important to note that a spectrum of options exists to improve the security of satellite call-backs to match perceived protection to perceived risks. Reading a command verification from a blockchain may be itself sufficient to protect against a single account disclosure, but only if the communication channel is secure and the blockchain node returning the information is not in itself compromised. Security could be improved by, for example, having a satellite query more than one node on the blockchain, using a so-called trusted oracle cryptographically sign the command verification at the smart contract, using some verifiable computing scheme [36] to produce a proof that the command verification has actually been included in the blockchain and cannot be removed (except, perhaps, with negligible probability), or even (ultimately) running a “light” blockchain client on the satellite so that, by providing the Merkle path to the command verification, the satellite can itself securely verify its inclusion in the blockchain. Based on the above analysis, we recommend using an enterprise Ethereum blockchain to implement multifactor authentication, multiparty authorization, or both via a smart contract on a private or consortium network with user permissions and an IBFT 2.0 consensus algorithm. Figure 53.3 illustrates an example workflow that combines multifactor authentication and multiparty authorization in which an operator command to a spacecraft is first confirmed by multiple parties on a blockchain. It is important to keep in mind that the various parties are making calls to the same smart contract although the figure shows the parties interacting with different copies of the smart contract stored on different nodes. The blockchain, as a distributed system, is required to come to consensus between its nodes each time a write operation results in an addition of information to the blockchain. Smart contracts in this scenario would maintain a command approval table, indexed by the cryptographic hash of the command, which maps to a Boolean value indicating whether a command has been approved. An arbitrary number of automated processes or human actions may be required by a given smart contract for complete execution. Arbitrarily complex business requirements may therefore be satisfied.

680

Advances in communications satellite systems 2: ICSSC-2019 g

e

f

b a

d

c

Figure 53.3 Multifactor authentication and multiparty authorization example: (a) an operator proposes a command to be sent to a spacecraft; (b) some number of automated processes (zero or more) confirm command syntax and perhaps applicability in the operational context; (c) some number of humans (zero or more) confirm the command should proceed; (d) the smart contract sets the entry of the command approval table associated with the hash of the command to the Boolean value True; (e) the operator sends the command to the spacecraft. Where verifiable computing is used, the operator also sends a proof that the entry of the command approval table has been set to True. Where the satellite runs a light client, the operator also sends the Merkle path to the contract state for the approval table; (f) the spacecraft hashes the command and verifies that the entry of the approval table associated with the hash is set to True using one of the techniques listed above; and (g) the spacecraft executes the command if and only if the command verification was successful. NB: The blockchain nodes come to consensus after each write to any version of the smart contract

The key benefit of this approach is to provide a much higher level of authentication and/or authorization security. An attacker would need to gain control over an arbitrary number of user accounts and be able to use those accounts to perform actions on the blockchain in order to confirm an inappropriate command. There are limitations to using any multifactor authentication or multiparty authentication system, including this proposal. Specifically, the need for a confirmation

Securing spacecraft tasking and control via Ethereum blockchain

681

of each issued command increases the time required between command issuance and execution, which might not be advantageous in some circumstances. Additionally, any disruption in communications between command issuance and confirmation, or other ways of disallowing confirmation, could cause command failure. These concerns should be traded off against the benefits of enhanced security in practical environments. Mitigations to such limitations may include creating a separate smart contract to require multiparty authorization by humans in the event of a failure of a multifactor authentication system (possibly via a separate communications channel), a mandatory entering of a safe mode in the event of communications disruption, or other fallbacks. Given the high economic value of spacecraft, it may also be wise to allow a command override to be used in dire emergencies, such as by assigning a one-time pad of secret keys prior to launch. The necessity of keeping such emergency keys physically secure naturally cannot be overstated.

53.5 Conclusion Integration of space-based communications infrastructure within 5G networks will require many spacecraft to improve communications security implementations and ensure the identity of operators issuing commands. We analyzed limitations of the subset of Earth orbiting communications satellites with the ability to remotely update software and provided recommendations that would allow them to improve their operational security with minimal changes. Specifically, we suggested the implementation of a terrestrial enterprise Ethereum blockchain to implement multifactor authentication, multiparty authorization, or both via a smart contract on a private or consortium network with user permissions and an IBFT 2.0 consensus algorithm. We introduced, we believe for the first time, the concepts of fully-or partially automated multifactor authentication and multiparty authorization systems in which a blockchain is a source of consensus from multiple parties. Both concepts have existed for the last human generation, but have not been applied to a blockchain to the authors’ knowledge. We note that a smart contract deployed on an enterprise Ethereum blockchain configured to use a POA consensus algorithm with immediate finality would be capable of addressing Mandl’s requirements at the blockchain level. The approach suggested in this chapter already requires participating spacecraft to read from such a blockchain. The only additional feature required to fulfill Mandl’s requirements onboard spacecraft would be the ability for satellites to be able to provide their sensor data to a smart contract on the blockchain. Thus, the system outlined in this chapter represents a subset of Mandl’s requirements and would be a reasonable stepping stone toward the implementation of Mandl’s requirements. It also seems likely that economies of scale could develop using Mandl’s conception in that spacecraft owners could be compensated financially for fulfilling smart contract Earth observation requests. Existing crypto-economic

682

Advances in communications satellite systems 2: ICSSC-2019

features of Ethereum and other blockchains could be useful in developing such economies. The approaches outlined in this chapter can reduce the likelihood of command exploitation but not denial of service (jamming) attacks.

Acknowledgments The authors gratefully acknowledge ConsenSys AG for financially supporting this research, especially its PegaSys division.

References [1] Sowmya, S., and Malarchelvi, P.D.S.K. “A survey of jamming attack prevention techniques in wireless networks.” Proceedings of International Conference on Information Communication and Embedded Systems (ICICES2014). Chennai, India. 2014. pp. 1–4. [2] NASA Office of Inspector General, Office of Audits. Cybersecurity Management and Oversight at the Jet Propulsion Laboratory, IG-19-022. NASA. June 2019. [3] Creedon, M. Space and cyber: Shared challenges, shared opportunities. Air Force Research Institute. November 2011. [4] Beldavs, V. “Blockchains and the emerging space economy.” Available from http://www.thespacereview.com/article/3077/1 [Accessed August 23, 2019]. [5] Cheng, S., Gao, Y., Li, X., et al. “Blockchain application in space information network security.” In Q. Yu (ed.). Space information networks. vol. 972. Singapore: Springer; 2019. pp. 3–9. [6] Xu, R., Chen, Y., Blasch, E., et al. “Exploration of blockchain-enabled decentralized capability-based access control strategy for space situation awareness.” Optical Engineering. 2019; 58(4): 041609. [7] Molesky, M. J., Cameron, E. A., Jones, J., et al. “Blockchain network for space object location gathering.” Proceedings IEEE 9th Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON). 2018. pp. 1226–1232. [8] Jennath, H. S., Adarsh, S., and Anoop, V. S. “Distributed IoT and applications: A survey.” In A.N., Krishna, K.C., Srikantaiah, and C., Naveena (eds.). Integrated intelligent computing, communication and security. Singapore: Springer; 2019. pp. 333–341. [9] Skobelev, P.O., and Lakhin, O.I. “Towards the digital platform and smart services for managing space traffic.” International Journal of Design and Nature and Ecodynamics. 2018; 13(2): 187–198. [10] Hyland-Wood, D., and Khatchadourian, S. “A future history of international blockchain standards.” Journal of the British Blockchain Association. 2018; 1(1): 3724.

Securing spacecraft tasking and control via Ethereum blockchain

683

[11] Gao, Y., Hu, S., Tang, W., et al. “Situational awareness in space based blockchain wireless networks.” In Q. Yu (ed.). Space information networks, 972. Singapore: Springer; 2019. pp. 15–20. [12] Moussa, M.A., and Chan, C.S. “Plurality-factor security system.” U.S. Patent 6,035,406. March 2000. [13] Carley, J.A. “Near real-time multi-party task authorization access control.” U.S. Patent 7,519,826B2. April 2009. [14] Mandl, D. “Bitcoin, blockchains and efficient distributed spacecraft mission control.” Available from https://ntrs.nasa.gov/search.jsp?R=20170009470 [Accessed March 13, 2019]. [15] Mital, R., de La Beaujardiere, J., Mital, R., et al. “Blockchain application within a multi-sensor satellite architecture.” NASA Technical Report 20180006549. April 2019. [16] Enterprise Ethereum Alliance. Available from https://entethalliance.org [Accessed August 23, 2019]. [17] Enterprise Ethereum Alliance. EEA Client Specification v3. 2019. [18] Castro, M., and Liskov, B. “Practical Byzantine fault tolerance.” Proceedings of Third Symposium on Operating Systems Design and Implementation. New Orleans, LA. February 1999. [19] Lin, Y.T. “Istanbul Byzantine fault tolerance.” Available from https://github.com/ethereum/EIPs/issues/650 [Accessed March18, 2019]. [20] Saltini, R., and Hyland-Wood, D. “IBFT 2.0: A safe and live variation of the IBFT blockchain consensus protocol for eventually synchronous networks.” Available from https://arXiv:1909.10194v1 [cs]. 2019. [21] Szila´gyi, P. “Clique PoA protocol and Rinkeby PoA testnet. EIP 225.” Available from https://github.com/ethereum/EIPs/issues/225 [Accessed August 23, 2019]. [22] Lamport, L., Shostak, R., and Pease, M. “The Byzantine generals problem.” ACM Transactions on Programming Languages and Systems. 1982; 4(3): 382–401. [23] Dwork, C., Lynch, N. and Stockmeyer, L. “Consensus in the presence of partial synchrony.” Journal of the ACM. 1988; 35(2): 288–323. [24] The enterprise enhanced BFT specification. Available from https://github.com/EntEthAlliance/enhanced-bft [Accessed August 23, 2019]. [25] Miller, A., Xia, Y., Croman, K., et al. “The honey badger of BFT protocols.” Proceedings of 2016 ACM SIGSAC Conference on Computer and Communications Security -CCS’16. Vienna, Austria. 2016. pp. 31–42. [26] Buchman, E., Kwon, J., and Milosevic, Z. “The latest gossip on BFT consensus.” Available from https://arXiv:1807.04938 [cs]. 2018. [27] Crain, T., Gramoli, V., Larrea, M., et al. “DBFT: Efficient leaderless Byzantine consensus and its application to blockchains.” Proceedings of 2018 IEEE 17th International Symposium on Network Computing and Applications (NCA). Cambridge, MA. 2018. pp. 1–8. [28] Quorum. Available from https://github.com/jpmorganchase/quorum [Accessed August 23, 2019].

684 [29] [30] [31]

[32]

[33]

[34]

[35] [36]

Advances in communications satellite systems 2: ICSSC-2019 Saltini, R. “IBFT liveness analysis.” Proceedings of 2019 IEEE International Conference on Blockchain (Blockchain 2019). Atlanta, GA. July 2019. Saltini, R. and Hyland-Wood, D. “Correctness analysis of IBFT.” Available from https://arXiv:1901.07160 [cs]. 2018. Gilad, Y., Hemo, R., Micali, S., et al. “Algorand: Scaling Byzantine agreements for cryptocurrencies.” Proceedings of 26th Symposium on Operating Systems Principles - SOSP ’17. Shanghai, China. 2017. pp. 51–68. Wu¨st, K., and Gervais, A. “Do you need a Blockchain?” Proceedings of 2018 Crypto Valley Conference on Blockchain Technology (CVCBT). 2017. pp. 45–54. Xu, X., Weber, I., and Staples, M. “Design process for applications on blockchain.” In X., Xu, I., Weber, and M. Staples (eds.). Architecture for blockchain applications. Cham: Springer; 2019. pp. 93–111. Stouffer, K. K., and Falco, J. “Recommended practise: Improving industrial control systems cybersecurity with defense-in-depth strategies.” Department of Homeland Security, Control Systems Security Program, National Cyber Security Division. 2009. p. 3. Kosmala, W.A.J. A friendly introduction to analysis. New Jersey: Pearson Prentice Hall; 2004. Gennaro, R., Gentry, C., and Parno, B. “Non-interactive verifiable computing: Outsourcing computation to untrusted workers.” Proceedings of the 30th Annual Conference on Advances in Cryptology. Santa Barbara, CA 2010.

Chapter 54

PAPR reduction and digital predistortion for 5G waveforms in digital satellite payloads Ovais Bin Usman1, Thomas Delamotte1 and Andreas Knopp1

Satellite systems will play an important role in the coming fifth generation (5G) of mobile communications. For a smooth integration of satellite networks into the terrestrial ones, the standardization bodies are pushing for shared spectrum. Therefore, it is of interest to study satellite specific scenarios, the applicability of multicarrier waveforms that have already shown promise to meet the requirements of the future mobile networks. 5G candidate waveforms such as filtered orthogonal frequency division multiplexing (f-OFDM), filter bank multicarrier (FBMC), and universal filtered multicarrier (UFMC) offer sharper out-of-band characteristics, significantly increasing the spectral efficiency. However, like OFDM, these waveforms exhibit a high peak to average power ratio (PAPR). A high PAPR saturates the nonlinear high power amplifier (HPA) which leads to nonlinear distortions in the on-board HPA’s output. Moreover, signal clipping is often proposed in the literature to reduce the PAPR. However, clipping itself introduces nonlinear distortions within the signal bandwidth. Digital predistortion (DPD) can be applied to the clipped signal to remove the added nonlinear distortions while keeping the overall PAPR low. This chapter provides the simulation results on the application of the aforementioned waveforms to a satellite communication chain and presents the gains achieved by implementing DPD and clipping together in terms of PAPR, power spectral densities (PSDs), and bit error rates (BERs). Key Words: peak to average power ratio (PAPR); filtered-orthogonal frequency division multiplexing (f-OFDM); filter bank multicarrier (FBMC); universal filtered multicarrier (UFMC); high power amplifier (HPA); digital predistortion (DPD); 5G waveforms; satellite payloads

1

Institute of Information Technology, Munich University of the Bundeswehr, Neubiberg, Germany

686

Advances in communications satellite systems 2 : ICSSC-2019

54.1 Introduction Over the past years, efforts have been made to converge satellite systems toward the terrestrial systems, especially in the context of physical layer. As a result, the standardization bodies have pushed for the need of shared spectrum [1]. Furthermore, the potential integration of satellite systems with terrestrial networks has been recently receiving significant attention [2]. Therefore, the recent developments made in the terrestrial communication systems should also be analyzed for high throughput satellite (HTS) systems. For example, the performance evaluation of the 5G New Radio (NR) waveform in HTS. Some of the candidate waveforms that have been studied for 5G NR include filtered orthogonal frequency division multiplexing (f-OFDM), windowed OFDM (W-OFDM), filter bank multicarrier (FBMC), and universal filtered multicarrier (UFMC). The aforementioned waveforms achieve lower out-of-band (OOB) emissions which significantly increase spectral efficiency, but they also suffer from high peak to average power ratio (PAPR) and increased computational complexity [3]. The previously applied OFDM waveform in standards like the “3rd Generation Partnership Project (3GPP) and Long Term Evolution (LTE)” has been seriously questioned in terms of its suitability to meet the increased requirements of future communication systems especially in terms of flexibility, spectral and power efficiency, and robustness to synchronization errors [4]. Analysis performed in [5] indicates that the f-OFDM is the most promising candidate for 5G NR networks. It is shown that f-OFDM is a flexible multicarrier waveform that exhibits low OOB radiation while retaining most of the features of the legacy OFDM, and now is a part of the 5G standard. This chapter analyzes the performance of f-OFDM along with the other mentioned waveforms in a satellite communication chain in terms of power spectral densities (PSDs), PAPR, and bit error rates (BER). The aforementioned multicarrier waveforms suffer from a high PAPR [6,7]. This is because the closely packed multicarrier signals overlap causing severe amplitude fluctuations. A higher PAPR causes nonlinear distortions in the high power amplifier’s (HPA) output as it saturates the HPA. Near saturation HPA output exhibits higher intermodulation noise, clustering and warping of signal constellation, and spectral regrowth. To keep such nonlinear distortions low, the on-board HPA needs to be operated at a larger input back-off (IBO). However, this reduces the power efficiency [8]. PAPR reduction schemes can be applied to operate the HPA more efficiently in terms of power. Several PAPR reduction schemes exist in the literature such as selective mapping [9], partial transmit sequencing [10], and linear block coding [11]. However, these schemes increase the complexity of the transmitter and receiver, and require huge look-up tables (LUTs) for encoding and decoding purposes [12]. This chapter considers a much simpler and an effective PAPR reduction method, that is, signal clipping [13,14] which can be directly applied in the satellite’s transponder. The high peaks of the multicarrier signal are clipped before the signal passes through the HPA. However, clipping itself is a nonlinear operation. It introduces in-band and out-of-band (OOB) distortions which

PAPR reduction and digital predistortion for 5G waveforms

687

lead to a loss in spectral efficiency and BER performance. Filtering after clipping can reduce the OOB radiation, however, it may cause some peak growth within the signal bandwidth [15]. To counteract the effects of clipping, and the on-board nonlinear HPA, this chapter considers a combined on-board PAPR reduction and digital adaptive predistortion implementation for the discussed multicarrier waveforms. Digital predistortion (DPD) can not only linearize the HPA but also reduces the in-band distortion introduced by signal clipping. The considered DPD method, detailed in [16], is adaptive in nature and requires a feedback loop. Adaptive nature not only compensates for the nonlinearities introduced by the HPA and clipping but also tracks changes in signal characteristics such as number of carriers, bandwidth, etc. Conventionally, the HPA’s output signal bandwidth reaches five times that of the input signal bandwidth due to spectral regrowth. However, due to the constraints on sampling rates of the on-board processors (OBPs) and the analog to digital converters (ADCs), only a certain portion of the HPA’s output is fed to the OBP to compute the predistorter coefficients. Therefore, the abstract only considers bandlimited DPD. Moreover, low sampling rate requirements also keep the computational and hardware complexity of the on-board digital architecture low. The rest of the chapter is organized as follows: Section 54.2 explains briefly the system model. Signal clipping, DPD, and HPA models are introduced in Section 54.3. Simulation results are presented in Section 54.4. Finally, conclusions are drawn in Section 54.5.

54.2 System model Figure 54.1 presents the considered system model. Transmitter block produces the multicarrier signals such as f-OFDM, FBMC, or UFMC. It is assumed that the power amplifier at the transmitter is operated deep in the linear region, such that the high PAPR of the produced multicarrier signals does not saturate the transmitter’s HPA. The received signal at the satellite is initially fed to the input-multiplexer (IMUX) to capture the bandwidth of interest, that is, bandwidth spanned by the multicarrier signal and rejects all other channels. As mentioned earlier, due to higher power efficiency requirements, the satellite’s HPA has to be operated much closer to saturation compared to the transmitter’s HPA. Therefore as proposed, the multicarrier signal is clipped for PAPR reduction. The clipped signal is then predistorted before being fed to the HPA. The output multiplexer (OMUX) filters out the OOB II. System model η1

Tx

ηI

IMUX

PAPR red.

DPD

HPA

OMUX

Figure 54.1 System model

Rx

688

Advances in communications satellite systems 2 : ICSSC-2019

emissions from the HPA output before the amplified signal is transmitted back to the receiver on earth.

54.3 PAPR reduction and predistortion method This section details the PAPR reduction technique and the DPD method implemented to linearize the HPA and remove the nonlinear effects present in the considered satellite communication chain. Figure 54.2 presents the satellite transponder model considered in more detail. In the following different blocks of Figure 54.2 are explained.

54.3.1 HPA model The HPA model used in this abstract is called the Saleh model [17], and it models the amplitude (AM-AM) and phase (AM-PM) distortion effects as follows: yAM ½rðtÞ ¼

aa rðtÞ ba rðtÞ2 þ 1

ðaÞ

yPM ½rðtÞ ¼

ap rðtÞ2 bp rðtÞ2 þ 1

ðbÞ

(54.1)

where r(t) is the envelope of the input signal
x ðtÞ to the HPA. The Saleh model is a memoryless HPA model and is widely used to model TWTAs. aa, ap, ba, and bp are the Saleh model coefficients and are also given in reference [17]. Even though the considered HPA model here is memoryless, but the memory effects still prevail due to the transponder filters, that is, IMUX and OMUX. Therefore, the feedback signal for DPD is taken from the output of OMUX to include the memory effects into account for the computation of predistorter coefficients.

54.3.2 PAPR reduction PAPR is defined as the ratio between peak power and the average power of the signal. For the multicarrier signals considered in the abstract, PAPR reduction is needed to avoid the saturation of the transponder’s HPA. The proposed PAPR reduction method is called signal clipping. It is detailed in [14] and is briefly described here as well. The output of the clipping block is given as   xðnÞ if jxðnÞj  g2 b (54.2) x ðnÞ ¼ g2 xðnÞ=jxðnÞj if jxðnÞj  g2 OBP x(n)

ADCs

xˆ(n)

Clipping

x˜(t)

x(n)

DPD

DACs

HPA

Adaptive DPD ADCs estimation yd(n)

Down conv

BPF

y(t)

OMUX

Figure 54.2 Proposed satellite transponder model

PAPR reduction and digital predistortion for 5G waveforms

689

where g is the clipping parameter. A smaller value of g implies a higher PAPR reduction. It should be noted that clipping does not change the phase of the signal, that is, ff b x ðnÞ ¼ ff xðnÞ. As mentioned earlier, clipping is a nonlinear operation and introduces both out-of-band (OOB) and in-band distortions. While the OOB distortions can be eliminated by the OMUX filtering, additional signal processing is required to mitigate the in-band distortions. If not compensated for, the in-band distortions can cause severe nonlinear ISI, and lead to a poor BER performance. The chapter proposes the use of a nonlinear predistorter to remove such nonlinear effects. The proposed predistorter is described in Section 54.3.3.

54.3.3 Digital predistortion Predistortion is one of the most effective techniques used to compensate for the nonlinear effects present in the system, for example, due to the signal clipping and operation of the HPAs closer to saturation. Predistortion intentionally distorts the input signal such that the nonlinearities present can be compensated. It can reduce the intermodulation noise and output-backoff (OBO) leading to more powerefficient HPAs. Furthermore, if modeled appropriately, DPD can remove linear distortions introduced by the transponder filters given that the feedback signal is taken from the OMUX output. Reference [14] also presents a predistortion method alongside the proposed PAPR reduction. However, the suggested DPD does not take into account the bandlimitation due to the OMUX and the bandpass filter in the feedback path. Furthermore, the presented predistortion method in [14] is sample-based approach. Although the block-based DPD techniques are more complex when compared to sample-based DPD methods; however, in general, they yield an overall better linearization performance, especially under bandlimitation constraint. Bandlimited block-based digital predistortion (DPD) detailed in [16] is considered here and is described in the following. Memory polynomial model is used to compute the predistorter coefficients [18]. Note that in this chapter, the Saleh model [17] is used as a realistic transponder HPA model, while the MP model is used for the state-of-the-art DPD algorithm. However, the predistortion technique detailed in this abstract is valid for other realistic HPA models as well. This is because first the true HPA is modeled using an MP model in the digital domain, and then that is used to obtain the DPD coefficients in the implemented presdistortion scheme. Using the MP model, the nth observation of the DPD output and the bandlimited HPA output is given by
x ðnÞ ¼

Q K X X

h i ckq;BL b x ðn  q  iÞjb x ðn  q  iÞjk1

k¼1 q¼0

"

yd ðnÞ ¼

L X

yðn  1ÞgðiÞ

(54.3a)

# (54.3b)

i¼0

where cBL is the unknown bandlimited MP model coefficients for the DPD. Furthermore, K and Q represent the nonlinearity order and maximum memory

690

Advances in communications satellite systems 2 : ICSSC-2019

depth, respectively. g(.) is the feedback filter. In the proposed DPD method, a bandlimited MP model for the HPA given by the coefficients wBL is also computed. This model is used inside the OBP to run additional iterations to obtain a set of DPD coefficients which exhibit the least mean squared error (MSE). Initially, the bandlimited DPD coefficients cBL can be set to the trivial solution [1 0 0]T. In terms of the HPA model coefficients (wBL), the approximated bandlimited HPA output b y d ðnÞ in digital domain can be written as b y d ðnÞ ¼

Q K X X

h i wkq;BL x
ðn  q  iÞj
x ðn  q  iÞjk1

(54.4)

k¼1 q¼0

Note that the wBL already incorporates the effects of the bandlimiting filter g(.), hence it is not included in (54.4). Defining bx ðn; k; qÞ ¼ b x ðn  qÞjb x ðn  qÞjk1 and k1
x ðn; k; qÞ ¼
x ðn  qÞj
x ðn  qÞj , the models in (54.3a) and (54.4) can be written as x Tn cBL xðnÞ ¼ b y d ðnÞ ¼

(54.5a)

b x Tn wBL

(54.5b)

x ðnÞ ¼ ½bx ðn; 1; 0Þ b x ðn; 1; 1Þ . . . b x ðn; 1; QÞ . . . b x ðn; K; QÞT . x ðnÞ also where b has the same definition as x ðnÞ. Gathering N samples of the input, that is, x ¼ ½xð0Þ xð1Þ . . . xðN  1ÞT , and the clipped signal, that is, b x ¼ ½b x ð0Þ bx ð1Þ . . . b x ðN  1ÞT , we can write the MP models in matrix form as b cBL x ¼X

(54.6a)

b wBL ; b yd ¼X

(54.6b)

b ¼ ½x ð0Þ b x ð1Þ . . . b x ðM  1ÞT , X ¼ ½x ð0Þ x ð1Þ . . . x ðM  1ÞT , where X b ¼ ½b y d ðN  1ÞT . The solution of (54.6) can be derived by and Y yd ð1Þ . . . b yd ð0Þ b applying the least squares (LS) algorithm. yd wBL ¼ ðX X Þ1 X b H

H

b HY b Þ1 Y b Hx ðaÞ cBL ¼ ðY

ðbÞ

(54.7)

Note that in (54.7a), b y d is the approximated HPA output in the digital domain, while in the first iteration we actually get a measured value of the bandlimited HPA output yd. Therefore, the measured value, that is, yd is used in (54.7a) to obtain the HPA coefficients for the proposed DPD method. Table 54.1 details the steps to obtain the DPD coefficients.

54.4 Simulations results Simulation results in terms of BER, PAPR, and PSDs are provided here for f-OFDM, UFMC, and FBMC multicarrier waveforms. Simulated waveforms have a 20 MHz bandwidth (BW) consisting of 64 subcarriers with a subcarrier spacing of

PAPR reduction and digital predistortion for 5G waveforms

691

Table 54.1 Bandlimited DPD algorithm

312.5 kHz. A sinc filter with a Hann time window is used as the filter in f-OFDM. Sub-band-wise filtering with Dolph–Chebyshev window is performed in the case of UFMC. A Hermite filter is used for FBMC with an overlapping factor of 4. Modulation schemes simulated are 4-QAM and 16-QAM. Feedback path BW is equal to the OMUX bandwidth.

54.4.1 Analysis with only DPD Figures 54.3 and 54.4 present the PAPR and BER analysis, respectively, for the case when only the DPD implementation is considered. The respective output backoffs (OBOs) for the cases with and without DPD are labeled in Figure 54.3. From Figure 54.4, it can be observed that both f-OFDM and UFMC have almost a similar BER performance, while FBMC performs a little worse when no DPD is considered for the given simulation parameters. Furthermore, note that implementing the proposed DPD improves the BER performance for all of the waveforms, but also drastically increases the PAPR (see Figure 54.3). Therefore, clipping is required to not only reduce the PAPR of the multicarrier waveforms in the first place but also to overcome the PAPR gain when DPD is implemented. Note that the increase in PAPR due to DPD is much higher for lower OBO.

54.4.2 Analysis with only clipping Section 54.4.4 highlighted the need of PAPR reduction, especially when the DPD is implemented. Figures 54.5 and 54.6 present the normalized PSD of the clipped

692

Advances in communications satellite systems 2 : ICSSC-2019

A.

Analysis with only DPD PAPR-no DPD/dB

PAPR-with DPD/dB

WaveForm

PAPR-no DPD/dB

PAPR-with DPD/dB

WaveForm

FBMC-OQAM

9.97

11.5

FBMC-OQAM

9.90

16.0

10.0

16.5

10.3

16.9

f-OFDM

10.1

11.8

f-OFDM

UFMC

10.4

12.1

UFMC

(a)

Modulation: 16-QAM, OBO with/without: 10.1 dB

(b)

Modulation: 4-QAM, OBO without/with DPD: FBMC, f-OFDM 3.6/4.6 dB OBO without/with DPD: UFMC 3.9/4.8 dB

10−1

10−1

10−2

10−2

10−3

10−4

10−5 10 (a)

Bit error ratio

Bit error ratio

Figure 54.3 PAPR analysis for the discussed waveforms with only DPD implementation (no clipping)

FBMC-No DPD FBMC-With DPD f-OFDM-No DPD f-OFDM-With DPD UFMC-No DPD UFMC-With DPD

12

14

16

10−3 FBMC-No DPD FBMC-With DPD f-OFDM-No DPD f-OFDM-With DPD UFMC-No DPD UFMC-With DPD

−4

10

18

SNR (dB)

20

10−5

22 (b)

6

8

10

12

Signal-to-noise ratio (dB)

Figure 54.4 BER analysis with only DPD implementation (no clipping). Waveforms: FOFDM, FBMC, and UFMC output and the BER curves for the proposed PAPR reduction method, respectively. Both figures provide the curves for the f-OFDM waveform. A much higher OBO is simulated to only observe and study the signal clipping effects on the PAPR and BER performance. The curves are labeled with the final reduced PAPR value, that is, lower the labeled PAPR value, severe the signal clipping was (a lower g was used). As mentioned earlier, even though clipping can effectively reduce the PAPR of the signal, however, it also introduces out-of-band (OOB) and in-band distortions. OOB distortions can be observed in Figure 54.5, where a higher OOB radiation is observed for more severe clipping, that is, a lower g. Moreover, the effects of in-band distortions are observed from Figure 54.6(a) and (b) where the BER curves have been plotted for 16-QAM and 4-QAM modulation schemes, respectively. Furthermore, it should be noted that signal clipping has a more significant effect on the BER performance at higher modulation order while a low modulation order is more robust to the clipping effects. Therefore, if 4-QAM is to be used, a much lower PAPR can be achieved without a significant loss in the BER

PAPR reduction and digital predistortion for 5G waveforms

Transmit signal PAPR 10.1 dB (no clip) γ :1.5, PAPR 7.1 dB γ :1.3, PAPR 4.9 dB γ :1.2, PAPR 3.8 dB γ :1.1, PAPR 2.9 dB

0 Power spectral density (dB)

693

−20

−40

−60

−80 0

10

20

30

40

50

Frequency (MHz)

Figure 54.5 PSD analysis for the proposed PAPR reduction method only (no DPD). Waveform: f-OFDM, modulation: 16-QAM, and OBO: 14.7 dB

10−1 10−1

Bit error ratio

Bit error ratio

10−2 10−2

10−3

10−4 (a)

PAPR 10.1 dB (no clip) γ : 1.5, PAPR 7.1 dB γ : 1.3, PAPR 4.9 dB γ : 1.2, PAPR 3.8 dB γ : 1.1, PAPR 2.9 dB

6

8

10

12

14

SNR (dB)

10−3 PAPR 10.0 dB (no clip) γ : 1.5, PAPR 7.1 dB γ : 1.3, PAPR 4.9 dB γ : 1.2, PAPR 3.8 dB γ : 1.1, PAPR 2.9 dB γ : 1.0, PAPR 2.1 dB

−4

10

16

18

10−5

20

5

(b)

6

7

8

9

10

11

12

SNR (dB)

Figure 54.6 BER analysis for the proposed PAPR reduction method only (no DPD). Waveform: f-OFDM and OBO: 14.7 dB performance. Similar trends are observed for the curves of other waveforms under clipping.

54.4.3 Analysis with clipping and DPD Figures 54.7 and 54.8 present the PSD, PAPR, and BER performance for the cases when the proposed PAPR reduction and DPD are implemented together using different clipping parameters (g). Figure 54.8(c) and (d) provides the PAPR observed at different values of g for the PSD analysis presented in Figure 54.7 and the BER analysis presented in Figure 54.8. The respective OBO and modulation schemes are labeled on the figures. Best BER performance is observed for the case when no clipping is introduced and DPD

694

Advances in communications satellite systems 2 : ICSSC-2019 10

−10

No DPD, γ:1.2 No DPD (no clip) With DPD, γ:1.2 With DPD (no clip) Transmit signal

0 Power spectral density (dB)

0 Power spectral density (dB)

5

No DPD, γ:1.3 With DPD, γ:1.3 No DPD, γ:1.5 With DPD, γ:1.5 No DPD (no clip) With DPD (no clip) Transmit signal

−20 −30 −40

−5 −10 −15 −20 −25

−50

−30 0

10

20

30

40

50

60

0

10

Frequency (MHz)

20

30

40

Frequency (MHz)

Figure 54.7 PSD analysis for the proposed PAPR reduction and DPD method when implemented together. Waveform: f-OFDM 10−1

10−1

10−3

10−4 10 (a)

Bit error ratio

Bit error ratio

10−2 10−2

No DPD (no clip) With DPD (no clip) No DPD, γ : 1.3 With DPD, γ : 1.3 No DPD, γ : 1.5 With DPD, γ : 1.5

12

14

10−3

10

16

18

10−5

20

Signal-to-noise ratio (dB)

No DPD (no clip) With DPD (no clip) No DPD, γ : 1.1 With DPD, γ : 1.1 No DPD, γ : 1.2 With DPD, γ : 1.2

−4

(b)

6

8

10

12

14

Signal-to-noise ratio (dB)

Clipping parameter-γ

PAPR-no DPD/dB

PAPR-with DPD/dB

Clipping parameter-γ

PAPR-no DPD/dB

PAPR-with DPD/dB

No clip

10.1

11.8

No clip

10.0

16.5

1.3

4.9

5.4

1.1

2.9

3.2

1.5

7.1

7.9

1.2

3.8

4.6

(c)

(d)

Figure 54.8 BER and PAPR analysis for the proposed PAPR reduction and DPD method when implemented together. Waveform: f-OFDM is performed. DPD removes the nonlinearities such as intermodulation noise, in-band and the OOB distortions to provide a gain in BER. DPD implementation without clipping also implies maximum linearization performance. This can be observed in

PAPR reduction and digital predistortion for 5G waveforms 18

20

16 Total degradation (dB)

22

TD (dB)

18 16 14 No DPD, γ:1.3 No DPD, γ:1.5 No DPD (no clip) With DPD, γ:1.3 With DPD (no clip) With DPD, γ:1.5

12 10 8 10

15

(a)

20

14 12 10 8

4 25

2

4

6

8

(b)

10

12

14

16

18

IBO (dB)

With DPD

No DPD

Clipping Min. Opt. Min. Opt. parameter TD/dB IBO/dB TD/dB IBO/dB

(c)

No DPD, γ:1.1 No DPD, γ:1.2 No DPD (no cLip) With DPD, γ:1.1 With DPD (no clip) With DPD, γ:1.2

6

IBO (dB)

No DPD

695

Clipping

Min.

Opt.

With DPD Min.

Opt.

parameter TD/dB IBO/dB TD/dB IBO/dB

No clip

9.2

10

5.7

7

No clip

16.4

1.3

12

12

6.8

6

1.3

20.1

1.5

10.4

11

5.4

6

1.5

16.5

(d)

18

9.0

19

11.5

10 10

18

8.7

10

Figure 54.9 TD analysis for the proposed PAPR reduction and DPD method when implemented together. Waveform: f-OFDM Figure 54.7 which provides the PSD curves for the HPA output. However, this is an impracticable scenario, as the PAPR rises even further by 1.7 dB and 6.5 dB for 16QAM and 4-QAM modulation schemes in the simulated scenario, respectively (see Figure 54.8(c) and (d)). Therefore, clipping is introduced to reduce the PAPR, but this leads to a loss in BER performance. The lower the clipping parameter g, more is the PAPR reduction. However, a lower g worsens the BER more as much severe in-band and OOB distortions (see Figure 54.7) are introduced. Nonetheless, when DPD is implemented along with signal clipping, the BER and linearization performance improves significantly and the increase in PAPR is rather minimal. Therefore, signal clipping can be performed to significantly reduce the PAPR, and DPD can be performed to remove the nonlinear effects introduced by clipping and the HPA. The curves for other waveforms exhibit similar trends and gains when clipping and DPD are implemented together.

54.4.4 Total degradation analysis Figure 54.9 presents the total degradation (TD) curves plotted against IBO for different clipping parameters (g) in conjunction with the proposed DPD method. The curves are provided for f-OFDM only. The total degradation metric is defined as follows: TD ¼ OBO þ ðSNRNonLin  SNRLin Þ

(54.8)

696

Advances in communications satellite systems 2 : ICSSC-2019

where SNRNonLin and SNRLin are the required SNRs to achieve a certain BER threshold (104 in this case) in a nonlinear and linear scenario, respectively. A TD curve presents a trade-off between power efficiency (related to IBO) and system BER performance. Furthermore, the TD curve provides the optimal operating point of the system, that is, the IBO exhibiting the lowest TD. The respective PAPRs for different clipping parameters are provided in Figure 54.8(c) and (d). From Figure 54.9, it can be observed that implementing DPD with or without clipping reduces the TD and leads to a lower operating point in terms of IBO, that is, more power-efficient amplifiers. A lower IBO implies a lower OBO. Moreover, when severe clipping is introduced without the DPD implementation, the TD rises drastically, for example, for g ¼ 1.3 (16-QAM) and g ¼ 1.1 (4-QAM). However, when the proposed DPD is considered along with clipping, there is a significant reduction in TD and the operating point of the HPA. Tables in Figure 54.9(c) and (d) provide the optimal IBOs and the respective TD. The BER and PSD analysis for optimal operating points is left as future work.

54.5 Conclusion This chapter highlighted the application of the multicarrier waveforms to satellite specific scenarios. However, the high PAPR of the discussed multicarrier signals is undesirable, especially for the satellite transponder’s HPA which has to be operated closer to saturation due to much stricter power efficiency requirements. Therefore, this study proposes signal clipping as a simple yet effective PAPR reduction method, in combination with digital predistortion to remove the nonlinear effects introduced by the onboard HPA and the clipping itself. The simulation results presented show that the proposed DPD, when implemented together with signal clipping, leads to a near-optimal performance with a much lower PAPR. Applying forward error correction (FEC) codes will improve the system performance further. The BER, PSD, and TD analysis incorporating FEC codes is left as future work.

References [1] Joint consortium 5G-PPP. “5G: Challenges, research priorities, and recommendations.” Technical Report. September 2014. [2] Kyrgiazos, A., Evans, B., Thompson, P., Mathiopoulos, P., and Papaharalabos, S. “A terabit/second satellite system for European broadband access: A feasibility study.” International Journal of Satellite Communications and Networking. 2014: 32(1): 63–92. [3] Ijaz, A., Zhang, L., Xiao, P., and Tafazolli, R. Analysis of candidate waveforms for 5G cellular systems, towards 5g wireless networks – A physical layer perspective. [4] Banelli, P., Buzzi, S., Colavolpe, G., Modenini, A., Rusek, F., and Ugolini, A. “Modulation formats and waveforms for 5G networks: Who will be the heir of OFDM?: An overview of alternative modulation schemes for

PAPR reduction and digital predistortion for 5G waveforms

[5]

[6]

[7]

[8]

[9]

[10]

[11] [12]

[13] [14]

[15]

[16]

[17]

[18]

697

improved spectral efficiency.” IEEE Signal Processing Magazine. 2014; 31 (6): 80–93. Zhang, X., Jia, M., Chen, L., Ma, J., and Qiu, J. “Filtered-OFDM—Enabler for flexible waveform in the 5th generation cellular networks.” 2015 IEEE Global Communications Conference (GLOBECOM). December 2015. pp. 1–6. Kollar, Z., Varga, L., and Czimer, K. “Clipping-based iterative PAPRreduction techniques for FBMC.” OFDM 2012; 17th International OFDM Workshop 2012 (InOWo’12). August 2012. pp. 1–7. Chafii, M., Palicot, J., and Gribonval, R. “Closed-form approximations of the PAPR distribution for multi-carrier modulation systems.” 2014 22nd European Signal Processing Conference (EUSIPCO). September 2014. pp. 1920–1924. Thompson, S. C., Proakis, J. G., and Zeidler, J. R. “The effectiveness of signal clipping for PAPR and total degradation reduction in OFDM systems.” GLOBECOM ’05. IEEE Global Telecommunications Conference, 2005. Vol. 5. November 2005. pp. 2807–2811. Bauml, R. W., Fischer, R. F. H., and Huber, J. B. “Reducing the peak-toaverage power ratio of multicarrier modulation by selected mapping.” Electronics Letters. 1996; 32(22): 2056–2057. Muller, S. H. and Huber, J. B. “OFDM with reduced peak-to-average power ratio by optimum combination of partial transmit sequences.” Electronics Letters. 1997; 33(5): 368–369. Tellambura, C. “Multicarrier transmission peak-to-average power reduction using simple block code.” Electronics Letters. 1998; (34)17: 1646. Han, S. H. and Lee, J. H. “An overview of peak-to-average power ratio reduction techniques for multicarrier transmission.” IEEE Wireless Communications. 2005; 12(2): 56–65. Li, X. and Cimini, L. J. “Effects of clipping and filtering on the performance of OFDM.” IEEE Communications Letters. 1998; 2(5): 131–133. Piazza, R., Bhavani Shankar, M. R., and Ottersten, B. “Generalized direct predistortion with adaptive crest factor reduction control.” 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). April 2015. pp. 3242–3246. Ann, P. P. and Jose, R. “Comparison of PAPR reduction techniques in OFDM systems.” 2016 International Conference on Communication and Electronics Systems (ICCES). October 2016. pp. 1–5. Yu, C., Guan, L., Zhu, E., and Zhu, A. “Band-limited Volterra series-based digital predistortion for wideband RF power amplifiers.” IEEE Transactions on Microwave Theory and Techniques. 2012; 60(12): 4198–4208. Saleh, A. A. M. “Frequency-independent and frequency-dependent nonlinear models of TWT amplifiers.” IEEE Transactions on Communications. 1981; 29(11): 1715–1720. Morgan, D. R., Ma, Z., Kim, J., Zierdt, M. G., and Pastalan, J. “A generalized memory polynomial model for digital predistortion of RF power amplifiers.” IEEE Transactions on Signal Processing. 2006; 54(10): 3852–3860.

This page intentionally left blank

Chapter 55

Effects of differential oscillator phase noise in precoding performance Liz Martı´nez Marrero1, Juan C. Merlano Duncan1, Jorge Querol1, Symeon Chatzinotas1, Adriano J. Camps Carmona2 and Bjo¨rn Ottersten1

Satellite precoding is a promising technique to meet the target data rates of the future high throughput satellite systems and the costs per bit as required by 5G applications and networks, but it requires strict synchronization among the transmitted waveforms, in addition to accurate channel state information. Most of the published work about this topic consider ideal oscillators, but in practice, the output of an oscillator is not a single spectral line at the nominal frequency. This chapter proposes a model for the oscillator phase noise and analyzes the resulting received signal to interference plus noise ratio (SNIR) in a satellite communication system using precoding. Simulations of a communication satellite system with a two-beam transponder and two receivers were performed to compute the effective SNIR. This work uses a simulator which also considers practical impairments such as time misalignment, errors in the channel state information, interference, thermal noise, and phase noise masks for satellite oscillators. The precoding methods used for the analysis are zero forcing (ZF) and minimum mean square error (MMSE). The obtained results prove that there is a degradation in the performance due to the use of independent oscillators but this effect is compensated by the precoding matrix. Key Words: phase noise; oscillator model; linear precoding; system performance

1

SnT, University of Luxembourg, Luxembourg, Luxembourg Department of Signal Theory and Communications, Universitat Polite`cnica de Catalunya, Campus Nord, Barcelona, Spain 2

700

Advances in communications satellite systems 2: ICSSC-2019

55.1 Introduction Satellite precoding is a promising strategy to meet the target data rates of the future high throughput satellite systems (HTS) and the costs per bit as required by 5G applications and networks [1,2]. Recent research activities have a special focus on multibeam precoding for multicast communication to achieve higher energy efficiencies [3] and to design better scheduling algorithms [4]. Precoding, which is a multiuser multiple input multiple output (MIMO) technique, requires strict synchronization among the transmitted waveforms in addition to channel state information, a requirement shared by all coherent distributed MIMO techniques [5]. A considerable amount of literature introducing new synchronization methods has been published [6]. Besides, other works have analyzed the effects of phase errors on signal coherence [7,8]. Specifically, authors in [8] consider errors resulting from imperfect clock alignment and platform spatial measurements in an open-loop system. As a result, they obtain beamforming’s tolerance to synchronization errors depending on the number of nodes and the desired coherence gain. The studies mentioned above consider ideal oscillators, but in practice, the output of an oscillator is not a single spectral line at the nominal frequency f 0, but it has sideband power as is shown in Figure 55.1 [9]. These phase and frequency instabilities affect precoding performance. Using a common local oscillator (LO) as clock reference might seem like a solution to this problem but it is not an alternative in distributed systems, such as network MIMO [10] and cloud radio access network (C-RAN) [4,11]. In this context, the main challenge is to coordinate the transmission of multiple geographically distant antennas that cannot use a common LO. The lack of a common oscillator also appears in practical satellite systems, due to technical constraints, such as independency between payloads, autonomy, robustness, cross-interference between RF channels and redundancy, where the whole system should not rely on the same oscillator [5,12]. Some authors have dealt with this problem during precoding design and implementation. For example, in [13], Gharanjik et al. propose a robust design by considering the time-varying phase noise introduced by oscillators onboard the satellite. The robustness is imparted by modeling the phase uncertainty as a random process and ensuring that the outage probability is maintained at desired levels. While in [14], Taricco considers the phase instability of the local oscillators driving the antenna feeds at the satellite payload as one of the phase offset causes in the

Ideal spectrum Real spectrum f0

Figure 55.1 Spectrum of the oscillator

Effects of differential oscillator phase noise in precoding performance

701

precoding implementation. Both papers model the phase uncertainty as Gaussian random process with zero mean and standard deviation s, 2
< s < 20
. A very interesting paper recently explores the effects of nonideal oscillators in a multiantenna hybrid digital-analog beamforming transceiver architecture [15]. The authors modeled the phase noise as Wiener and Gaussian processes in three different architectures: common LO, independent Los, and a block-based architecture. Through simulations, they arrived at the conclusion that the phase noise has more impact on the system performance when it is modeled as a Wiener process in an independent LOs architecture. In that case, for a phase noise variance of 10
, there is an error of more than 7
at the beam pointing and the sidelobe level increases in almost 2 dB respect to the common LO architecture. However, in practice oscillator noise is affected by additional phenomena that are not included in the aforementioned models. Many authors have studied this topic, searching for advanced models to characterize oscillator near-carrier power spectral density (PSD) [16,17]. Empirical models based on measurements suggest that the phase noise PSD can be described as a sum of power-law processes ha| f |a with a [ {4, 3, 2, 1,0} [9]. According to this idea, the random walk frequency noise, h4| f |4, continues increasing infinitely while frequency approaches oscillator’s nominal value. However, more recent researches consider an additional Gaussian session segment [16], instead of the spectral line at the carrier frequency shown in Figure 55.1, which is more similar to the real characteristic. This model, represented in Figure 55.2, includes the frequency drifts of a practical system, which is similar to a frequency modulation or spreading of the main carrier. In this chapter, we propose a model for the oscillator phase noise and analyze the resulting received signal to interference plus noise ratio (SNIR) in a satellite communication system using precoding. The work takes advantage of the simulations of a system with two beams and two receivers using a simulator developed by the Sigcom group. This allows considering practical impairment sources in a satellite communication such as time misalignment, errors in the channel state information, nonlinearities in the transmitter, interference, thermal noise, and phase

PSD (dB/(rad/s))

Gaussian 1Τ f 4 1Τ f 3 1Τ f 2 1Τf flat 0

ω (rad/s)

Figure 55.2 Oscillator PSD characteristic near the oscillator frequency. Adapted from [16]

702

Advances in communications satellite systems 2: ICSSC-2019

noise masks for consumer reception systems from ETSI standards for digital video broadcasting (DVB) [18]. The precoding methods used for the analysis are zero forcing (ZF) and minimum mean square error (MMSE) which are the simplest and most popular among the scientific community. This work analyzes independent LO for each beam and modeled the phase noise according to the two-state model proposed by Galleani in [19].

55.2 Two-state noise oscillator model The output voltage u0(t) of a generic oscillator with a nominal frequency f0 is u0 ðtÞ ¼ ½A þ aðtÞcos ð2pf0 t þ fðtÞÞ where A is the mean amplitude of the oscillator output, a(t) is the zero-mean amplitude noise, and f(t) is an error term due to oscillator phase noise. In this work, we consider that the effects of amplitude noise are overshadowed by the effects of phase noise, which is a common assumption in published work in this field. From this expression, we can obtain two fundamental quantities used to characterize clocks: phase and frequency deviation. The frequency fðtÞ . deviation (t) is defined as the derivative of the phase deviation, which isxðtÞ ¼ 2pf 0 As was mentioned above, phase noise PSD, Sf( f ), can be described by 8 0 X > > < ha f a 0 < f < fh (55.1) Sf ðf Þ ¼ a¼4 > > : 0 f  fh where fh is the high-frequency cut-off of an infinitely sharp low-pass filter [20]. These ha f a terms are related to random walk FM, flicker FM, white FM, flicker and white phase noise respectively [16]. A simpler implementation is the two-state clock noise model which considers just white FM phase noise (a ¼ 2) and random walk FM phase noise (a ¼ 4). Experimental evidence shows that the frequency deviation of a cesium clock is made by these two noises, namely, a white noise and a Wiener process. The last one is responsible for the random walk nature of the frequency deviation, while the white noise accounts for the local oscillations. Therefore, the frequency deviation can be written as yðtÞ ¼ x1 ðtÞ þ x2 ðtÞ

(55.2)

where x1(t) denotes a zero-mean Gaussian random process, x1 ðtÞeN ð0; q1 Þ, and x2(t) is a Wiener process. The Wiener process, also known as Brownian motion, is the prototype of random walks. It is characterized because the increments between two consecutive samples are normally distributed independent processes, Dweð0; tÞ. In continuous time: ðt (55.3) x2 ðtÞ ¼ x2 ðt_ Þ d t_ 0

where x2 ðt_ ÞeNð0; q2 Þ:

Effects of differential oscillator phase noise in precoding performance

703

To obtain the two-state model of the phase noise we substitute in (55.2) the phase deviation and the expression (55.3) for the Wiener process: ðt dxðtÞ ¼ x1 ðtÞ þ x2 ðt_ Þ d t_ (55.4) dt 0 Taking the derivative for both sides: ðt ðt xðtÞ ¼ x1 ðtÞ ¼ x1 ðt Þ þ x2 ðt_ Þ d t_ dt 0

(55.5)

0

Equation (55.5), shown in graphical form in Figure 55.3, describes the twostate clock noise model [17].

55.2.1 Discrete-time implementation To implement the system in Figure 55.3, it is useful to express (55.4) in the statespace form " # " #" # " # x01 ðtÞ 0 1 x1 ðtÞ x 1 ðtÞ ¼ þ (55.6) x02 ðtÞ x 2 ðtÞ 0 0 x2 ðtÞ where the inputs x1 and x2 are two independent Gaussian random pro zero-mean  q1 0 dðtÞ; x1 and x2 are the states cesses with a correlation matrix Rx 1 x 2 ðtÞ ¼ 0 q2 0 0 and x1 , x2 refers to their derivatives [17]. A discrete-time equivalent expression for (55.6) was obtained in [19] and has the form:   1 Ts x½n ¼ x½n  1 þ h½n  1 (55.7) 0 1 

   h1 ½n x1 ½n ; h½n ¼ and Ts sampling period. with x½n ¼ x2 ½n h2 ½n

[2(t)

x2(t)

y(t)

[1(t)

Figure 55.3 Two-state noise clock model

x1(t)

704

Advances in communications satellite systems 2: ICSSC-2019 The covariance matrix of h[n] is given by 2 3 Ts3 Ts2 q T þ q q 1 s 2 2 6 3 2 7 6 7 Ch1 h2 ¼ 6 7 4 5 2 Ts q2 q2 Ts 2

(55.8)

According to [19,21], q1 and q2 are directly related to the Allan variance s2y ðtÞ through s2y ðtÞ ¼

q1 q2 t þ t 3

(55.9)

This is a typical tool used to characterize the noise in oscillators and could be obtained from experimental measurements. Besides, the Allan variance is related to the noise PSD in (55.1) by [20] s2y ðtÞ ¼

h4

2p2 1 t þ h3 2 ln 2 þ h2 3 2t

1:038 þ 3 lnð2pfh tÞ 4p2 t2 3fh þh0 2 2 4p t þh1

(55.10)

For the two-state model analyzed in this chapter, we consider just the first and the third terms in (55.10). Then, equalling (55.9) and (55.10), we obtain q1 ¼

h2 2

(55.11)

q2 ¼ 2p2 h4 Using these equivalences, it can be generated the two-state model for any real or theoretical phase noise PSD.

55.3 Satellite precoding system with different clock references This work studies the effects of the oscillators phase noise in a satellite precoding system detailed in Section 55.4. To simplify the analysis, unicast communication between the satellite and two user terminals on Earth is considered. The satellite has two antennas feeds in single feed per beam configuration and uses precoding to avoid interference between the beams. It is well known that precoding operates by generating the transmitted signal c(t)  by multiplying the input symbol vector with the precoding matrix  w11 w12 . In a vector form: W¼ w21 w22

Effects of differential oscillator phase noise in precoding performance

705

c p ¼ Ws

(55.12)     c p1 s where c p ½n ¼ is the precoded transmitted signal and s½n ¼ 1 reprec p2 s2 sents the input data symbols. It should be noted that the phase noise is a fastvarying process that cannot be captured in the CSI loop. To evaluate the error introduced by different clock references during the upconversion, the terms ejf(t) were included in (55.12):      c p1 w11 e jf1 w12 e jf1 s1 ¼ (55.13) c p2 w21 e jf2 w22 e jf2 s2 The precoding matrix is calculated using the channel state information (CSI) obtained from each receiver. Basically, each user sends its estimation of the e and the precoding matrix is calculated in the gateway by two downlink channel H methods: zero-forcing (ZF) or minimum mean square error (MMSE). The former eH e H Þ1 , where H eH e H ðH uses the pseudo-inverse of the channel matrix, W ¼ H e . Otherwise, MMSE takes into account both the means the Hermitian matrix of H interference and the noise in order to improve the system performance also in eH e H þ aIÞ1 , with a being a regularization e H ðH noise-limited scenarios, W ¼ H parameter inversely proportional to the SNR and I the identity matrix [22]. The received signal at the user terminal equals r ¼ Hc p þ z (55.14)     h h12 z is the experienced where z½n ¼ 1 is the Gaussian noise and H¼ 11 z2 h21 h22 channel on the downlink. It should be noted that in a system with a common clock e ¼H), then applying reference (phase drift e jf), instant and perfect channel estimation (H 1 jf ZF yields W ¼ H and r ¼ se þ z. However, substituting (55.13) in (55.14)     r1 g s þ g12 s2 þ z1 ¼ 11 1 (55.15) r2 g21 s1 þ g22 s2 þ z2 with ðh11 h22 ejf1  h12 h21 ejf2 Þ detðHÞ ðh12 h11 ejf2  h11 h12 ejf1 Þ ¼ detðHÞ ðh21 h22 ejf1  h22 h21 ejf2 Þ ¼ detðHÞ ðh22 h11 ejf2  h21 h12 ejf1 Þ ¼ detðHÞ

g11 ¼ g12 g21 g22

706

Advances in communications satellite systems 2: ICSSC-2019

One of the most used parameters to measure performance in precoding systems is the signal to noise plus interference ratio (SNIR). According to [23], the SNIR at each receiver when ZF is used can be computed as: SNIRi ¼

jgii j2

(55.16)

jgij j2 þ s2z

This equation considers channel noise power s2 and interference power jgij j2 . It is evident that the use of different clock references will decrease the SNIR at the receivers due to an increment in the interference between beams. A similar approach can be followed for MMSE but it is omitted here due to space limitations.

55.4 System implementation The satellite communication system simulated is represented in Figure 55.4. The system is composed of a two-beams transparent satellite transponder, the communication channel, and the ground segment with the gateway and two user terminals. The transmitted signal was simulated according to the framing structure in the DVB-S2x standard [24]. Precoding is used for the downlink communication between the transponder and the user terminals, identified as Rx1 and Rx2 in Figure 55.4. The channel is assumed static in all simulations, and the return and feeder links are considered ideal. The right side of Figure 55.4, represents how was included the oscillator model in the simulations: u1(t) and u2(t) are the clock

s1(t)

s2(t)

X

Space segment

u1(t)

X u2(t)

H

y1(t)

Ground segment Gateway

y2(t)

Rx1 Rx1

Rx2

Rx2

Figure 55.4 Simplified diagram of a satellite communication system using precoding with independent clock references for each beam in the downlink channel

Effects of differential oscillator phase noise in precoding performance

707

reference for each beam at the transponder. Section 55.5 analyzes the effects of using two independent oscillators, as shown in Figure 55.4, against the use of a common clock reference, u1 ðtÞ ¼ u2 ðtÞ ¼ uðtÞ. The standard DVB-S2x defines a fixed framing structure to increase resilience to co-channel interference and to support synchronization algorithms [24]. For precoding purposes, the specifications state 9 bundled payload frames at each superframe. Each bundled frame contains a header, a precoded pilot field, 71 nonprecoded pilots, and 64,800 payload symbols. This structure is represented in Figure 55.5. In the performed simulations, 80 MBaud of baud rate was considered. Then, the transmission of a pilot takes 0.45 ms while a superframe lasts for 7.66 ms. Precoding systems use nonprecoded pilots to estimate the channel response at the receivers. Each pilot is formed by 36 BPSK symbols carrying Walsh–Hadamard sequences. The receivers compute an estimation of the distortion introduced at each beam by the channel. This information is sent to the Gateway, where the CSI,  e e e ¼ h 11 h 12 , is estimated. In order to increase robustness, the receiver averaH e h 21 e h 22 e estimates originating from several consecutive pilot sequences. The ges over H number of averaged estimates varies according to the SNIR of the system. The simulations, evaluate different values for this parameter. The gateway uses these CSI to calculate the precoding matrix. There are multiple algorithms to calculate W, but this work uses only ZF and MMSE, which are well known by the community. It is worth noting that the estimation from one superframe is used to compute the precoding matrix for the next superframes. The time gap between both of them depends on the delay of the communication link. For the simulations, it is considered zero-delay and 500 ms, which is the approximate delay for a closed-loop between the ground and a GEO satellite. Closed-loop includes both links, the direct one, gateway-transponder-receivers, shown as solid line arrows in Figure 55.4, and the return link, receivers-transponder-gateway, with dashed line arrows.

6 PLS code replica

180 precoded pilot symbols

Pilot fields, 36 symbols Scrambler RESET

Scrambler RESET SOSF

SFFI

PLH P2

1664 symbols 720 symbols

1

71

P

P

639 PLH P2

P

Pilots always ON

956 symbols [including 36 pilots]

Bundled PL-frame and 71 SF-pilot fields (384+180+64800+71×36=67920 symbols) 9 bundled PL-frame

540 Dummy symbols

Superframe length = 612,540 symbols

Figure 55.5 Superframe structure from the DVB-S2x standard. Adapted from [25]

708

Advances in communications satellite systems 2: ICSSC-2019

The gateway-satellite link is considered ideal, which is a valid assumption since it is usually a direct link with frequency division multiplexing and high SNIR. However, it introduces a considerable delay that cannot be omitted. For that reason, in the channel model, the response of the link between the satellite and the user terminals and the delay of both sections gateway-satellite and satellite-user terminals were included. The channel is modeled with coefficients hij provided through a European Space Agency (ESA) project [25]. The channel coefficients used for the selected locations are   0:4016 þ 0:0064j 0:0071 þ 0:0277j H¼  106 0:1911  0:1533j 0:2501  0:3269j

55.5 Simulations results Figure 55.6 shows two realizations, 8.4 s, of the phase noise model implemented. They were used to generate the phase drift at each clock reference in the satellite. As was discussed in Section 55.2, the phase noise model output contains a Wiener process plus an integrated Wiener process. That makes the phase noise have a smoother behavior than that of a simple Wiener process, so its variance grows quadratically with time, while the variance of a Wiener process increases linearly with time. The phase noise model output is a nonstationary stochastic process which implies that its PSD varies with time. The time-frequency representation of this PSD is shown in Figure 55.7. Besides, the estimation of the PSD represented in Figure 55.7 to verify the correspondence with the desired phase noise mask was computed. The phase noise 0.3 0.2

Phase drift [rad]

0.1 0 –0.1 –0.2 –0.3 –0.4

0

1

2

3

4 5 Time [s]

6

7

8

Figure 55.6 Phase drift from two independent oscillators

9

Effects of differential oscillator phase noise in precoding performance

709

60

100 0

50 –200

40

–300 8

30

Magnitude [dB]

–100

6 20

4

10–2 100

2 Time [s]

102

10

Frequency [kHz]

Figure 55.7 Time-frequency representation of the spectral characteristic for one realization of the phase noise 0

PSD [dBc/Hz]

–50

–100

–40 dB/dec

–20 dB/dec

–150

–200

–250

–300 100

101

102

103 104 Frequency [Hz]

105

106

Figure 55.8 Estimated PSD for the phase noise used in the simulations masks are generated through measurements of real oscillators PSD. For the simulations, a PSD with the 1/f 2 starting in 10 Hz at 75 dB was chosen. Figure 55.8 shows the estimated PSD of the clock references used in the simulations. The first simulation models the performance degradation due to the phase noise in an ideal system with perfect and updated CSI. The SNIR obtained with ZF and MMSE for a range of transmission power was compared when a

710

Advances in communications satellite systems 2: ICSSC-2019 18 16 14

ZF one oscillator ZF two oscillators MMSE one oscillator MMSE two oscillators

SNIR [dB]

12 10 8 6 4 2 0 10

15

20

25

Transmission power [dBW]

Figure 55.9 Open-loop simulation of phase noise PSD

common clock reference is used, dashed lines in Figure 55.9, against independent oscillators for each beam, solid line in Figure 55.9. As can be seen in the figure, there is an SNIR gap that grows with the transmission power. That is an expected result since the interference in (55.16) is directly related to the transmission power. It means that for low transmission power the channel noise s2z has more influence in the system performance, but as the transmission power increases the system is more affected by the interbeam interference due to the oscillators phase noise. However, the SNIR degradation is lower than 4 dB in the worse scenario. Besides, the closed-loop system was modeled: first, CSI is estimated using the pilots from three superframes and with this estimation, the system calculates the precoding matrix to send the payload symbols of the next superframe. Zero and 500 ms of delay were considered and the SNIR at the receivers for a set of transmission power was computed. The results are shown in Figure 55.10. This simulation shows similar results to the open-loop one; for low transmission power, MMSE performs better than ZF and the use of a common oscillator has less impact. However, for higher transmission power, the closed-loop showed better results since all curves converge to the same value. This is an expected result since phase deviation due to the oscillators are part of the CSI estimated by the receivers and therefore they are compensated with W. The delay has a very small effect in this performance, but it is important to note that this simulation considers a static communication channel. Results may be different when we include the dynamic behavior of the channel.

Effects of differential oscillator phase noise in precoding performance

711

18 ZF one oscillator, instant CSI ZF two oscillators, instant CSI MMSE one oscillator, instant CSI MMSE two oscillators, instant CSI ZF one oscillator, delayed CSI ZF two oscillators, delayed CSI MMSE one oscillator, delayed CSI MMSE two oscillators, delayed CSI

16 14

SNIR [dB]

12 10 8 6 4 2 0

10

15

20

25

Transmission power [dBW]

Figure 55.10 Closed-loop simulation of phase noise PSD

55.6 Conclusion This chapter deals with the effects of the phase noise from oscillators in the performance of satellite communication systems with precoding. Using the two-state phase noise model, the degradation in the SNIR at the receivers for a 22 system with common and independent clock references was analyzed. The simulations included the effects of the delay and the CSI estimation errors. The obtained results prove that there is a degradation in the performance of precoding systems due to the use of independent oscillators. In the open-loop system, the SNIR gap between a common clock reference and independent oscillators increases with the transmission power but never was bigger than 4 dB. However, in a closed-loop system, this effect is compensated by the calculation of the precoding matrix. According to the simulation results, the delay does not have much impact, but it may change if we include the dynamic characteristic of the channel which is part of our future work. Other open questions in this work are the effect of other impairments such as the initial phase offset between oscillators and the extension to more complex systems, bigger than 22.

712

Advances in communications satellite systems 2: ICSSC-2019

Acknowledgments This work was supported by the Fond National de la Recherche Luxembourg, under the CORE project COHESAT: Cognitive Cohesive Networks of Distributed Units for Active and Passive Space Applications.

References [1] Schwarz, R. T., Delamotte, T., Storek, K.-U., and Knopp, A. “MIMO applications for multibeam satellites.” IEEE Transactions on Broadcasting. 2019: 1–18. [2] Fatema, N., Hua, G., Xiang, Y., and Member, S. “Massive MIMO linear precoding : A survey.” IEEE Systems Journal. 2018; 12(4): 3920–3931. [3] Qi, C. and Wang, X. “Precoding design for energy efficiency of multibeam satellite communications.” IEEE Communications Letters. 2018; 22(9): 1826–1829. [4] Hou, Q., He, S., Huang, Y., Wang, H., and Yang, L. “Joint user scheduling and hybrid precoding design for MIMO C-RAN.” in 2017 9th International Conference on Wireless Communications and Signal Processing, WCSP 2017 - Proceedings. Vol. 2017. January. pp. 1–6. [5] Vazquez, M. A., Perez-Neira, A., Christopoulos, D., et al. “Precoding in multibeam satellite communications: Present and future challenges.” IEEE Wireless Communications. 2016; 23: 88–95. [6] Jayaprakasam, S., Rahim, S. K. A., and Leow, C. Y. “Distributed and collaborative beamforming in wireless sensor networks: Classifications, trends, and research directions.” IEEE Communications Surveys and Tutorials. 2017; 19(4): 2092–2116. [7] Quitin, F., Irish, A. T., and Madhow, U. “A scalable architecture for distributed receive beamforming: Analysis and experimental demonstration.” IEEE Transactions on Wireless Communications. 2016; 15(3): 2039–2053. [8] Nanzer, J. A., Schmid, R. L., Comberiate, T. M., and Hodkin, J. E. “Openloop coherent distributed arrays.” IEEE Transactions on Microwave Theory and Techniques. 2017; 65(5): 1662–1672. [9] Rutman, J. “Characterization of phase and frequency instabilities in precision frequency sources: Fifteen years of progress.” Proceedings of the IEEE. 1978; 66(9): 1048–1075. [10] Song, C. and Jeon, Y. “Weighted MMSE precoder designs for sum-utility maximization in multi-user SWIPT network-MIMO with per-BS power constraints.” IEEE Transactions on Vehicular Technology. 2018; 67(3): 2809–2813. [11] Femenias, G. and Riera-Palou, F. “Multi-layer downlink precoding for cloud-RAN systems using full-dimensional massive MIMO.” IEEE Access. 2018; 6: 61583–61599. [12] Yang, Y., Wang, W., and Gao, X. “Distributed RZF precoding for multiplebeam MSC downlink.” IEEE Transactions on Aerospace and Electronic Systems. 2018; 54(2): 968–977.

Effects of differential oscillator phase noise in precoding performance

713

[13] Gharanjik, A., Bhavani Shankar, M. R., Arapoglou, P. D., Bengtsson, M., and Ottersten. B. “Robust precoding design for multibeam downlink satellite channel with phase uncertainty.” in ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. 2015. vol. 2015-August. pp. 3083–3087. [14] Taricco, G. “Linear precoding methods for multi-beam broadband satellite systems.” in European Wireless 2014: Proceedings of 20th European Wireless Conference. 2014. pp. 1–6. [15] Aminu, M. U., Lehtoma¨ki, J., and Juntti, M. “Beamforming and transceiver optimization with phase noise for mmWave and THz bands.” in 16th International Symposium on Wireless Communication Systems ISWCS 2019. 2019. [16] Chorti, A. and Brookes, M. “A spectral model for RF oscillators with powerlaw phase noise.” IEEE Transactions on Circuits and Systems I: Regular Papers. 2006; 53(9): 1989–999. [17] Mcneill, J., Razavi, S., Vedula, K., and Richard, D. “Experimental characterization and modeling of low-cost oscillators for improved carrier phase synchronization.” in 2017 IEEE International Instrumentation and Measurement Technology Conference (I2MTC). 2017. [18] ETSI. “Digital video broadcasting (DVB); Second generation framing structure, channel coding, and modulation systems for broadcasting, interactive services, news gathering and other broadband satellite applications; Part 1: DVB-S2.” vol. 1. 2014. pp. 1–80. [19] Galleani, L. “A tutorial on the two-state model of the atomic clock noise.” Metrologia. 2008; 45(6). [20] 1139-2008 IEEE Standard Definitions of Physical Quantities for Fundamental Frequency and Time Metrology–Random Instabilities. [21] Zucca, C. and Tavella, P. “The clock model and its relationship with the Allan and related variances.” IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. 2005; 52(2): 289–296. [22] Alodeh, M., Spano, D., Kalantari, A., et al. “Symbol-level and multicast precoding for multiuser multiantenna downlink: A state-of-the-art, classification and challenges.” IEEE Communications Surveys and Tutorials. 2018; 20(3): 1733–1757. [23] Chatzinotas, S., Ottersten, B., and de Gaudenzi, R. “7.2 multiuser MIMO communications.” in Cooperative and cognitive satellite systems. vol. 1. Elsevier Science & Technology; 2015. pp. 220–224. [24] ETSI. Digital Video Broadcasting (DVB); Second generation framing structure, channel coding and modulation systems for broadcasting, interactive services, news gathering and other broadband satellite applications part 2: DVB-S2 Extensions (DVB-S2X). 2014. [25] LiveSatPreDem – Live satellite precoding demonstration. [Online]. Available from https://wwwfr.uni.lu/snt/research/sigcom/projects/liv esatpredem_live_ satellite_precoding_demonstration [Accessed May 11, 2019].

This page intentionally left blank

Chapter 56

GNSS-assisted acquisition technique for LTE over satellite Xiangdong Liu1 and Dilip Gokhale1

In this chapter, we describe an acquisition method for LTE over satellite, based on the standard 3GPP LTE PRACH waveform and processing procedures. The method is applicable when the user terminals are equipped and can utilize a global navigation satellite system (GNSS, e.g., GPS) receiver. The method requires modification of LTE user equipment (UE) only in software at layer 2 and reuse the PRACH preamble processing (detection and estimation) at the base station (eNodeB). Key Words: LTE; mobile satellite system (MSS); PRACH; synchronization

56.1 Introduction There is significant interest in SATCOM networks for using the 4G long term evolution (LTE) physical layer over satellite links. Using the widely deployed field-proven LTE physical layer in conjunction with associated layer 2 and 3 radio access technologies can be very beneficial for satellite networks to include seamless support for 4G and/or evolved packet system (EPS)-based services, a range of standardized modulation and coding schemes, and standardized resource definitions (e.g., frequency and time) that are well suited for 4G and/or EPS services. LTE technology further supports dynamic control of modulation/coding depending upon the link and terminal characteristics, dynamic resource allocation depending on user traffic needs and is compatible with the evolved packet core (EPC) network via the S1 interface. Adherence to LTE technology also provides an upgrade path to compatibility with 5G services and networks. There are, however, several areas where LTE standard protocols and/or procedures need to be modified for use in a satellite environment, primarily due to the significantly longer propagation delay and larger cell (beam) sizes. In particular, the LTE return direction acquisition is an important step to establish user equipment (UE) to base station (eNodeB) synchronization. LTE accommodates a variation of UE to

1

Lockheed Martin Space, Rockville, MD, USA

716

Advances in communications satellite systems 2: ICSSC-2019

eNodeB delay of no more than about 0.33 ms. In a satellite system, the variation of delay in a beam is ordinarily several milliseconds. In creating an acquisition method for LTE-over-Satellite, it is nonetheless advantageous to reuse the standard LTE design to the maximum extent to allow maximal reuse of terrestrial LTE equipment and product components, which minimizes system development and deployment cost. An “LTE over satellite” radio access network (RAN) as shown in Figure 56.1 consists of a multibeam satellite base station eNodeB transmitting at least one downlink OFDMA carrier paired with an uplink SC-FDMA carrier in each satellite beam. In each beam, there may be zero, one, or more user terminals or devices. On the service link (to end-users), radio carriers may be in, for example, L or S band, and on the feeder link side, carriers may be in, for example, C, Ku, Ka, or Q/V band. In this chapter, we describe an acquisition method developed at Lockheed Martin, as part of GMR2-4G technologies. The rest of this chapter is organized as follows. In Section 56.2, we give a brief overview of standard LTE acquisition method and point out why it is insufficient for a satellite system with much longer path delays and delay variations. In Section 56.3, we review some of the related work in the literature that extends terrestrial LTE coverage. In Section 56.4, we will detail one GMR2-4G acquisition method. In Section 56.5, we conclude the chapter. In the remainder of this chapter, we use GPS as a specific example of a GNSS. However, it should be noted that the described method works with any GNSS as long as microsecond level timing accuracy can be achieved.

56.2 LTE acquisition and synchronization background 56.2.1 LTE acquisition method overview In LTE, a UE must be time-synchronized with an eNodeB in the return direction, to be able to transmit information to the eNodeB. Time synchronization means that

E LT ers rri ca

LTE carriers

Internet

Satellite eNodeB S1 Core network

Figure 56.1 LTE-over-satellite system

PSTN

PLMN

GNSS-assisted acquisition technique for LTE over satellite

717

Cell radius

Cell

eNodeB

Max differential delay =

Cell radius Speed of light

Figure 56.2 Differential delay within an LTE cell the UE first obtains and then maintains information of the precise time delay between itself and the eNodeB. With the precise delay information, the UE can subsequently determine the precise instance of time to start transmitting a burst (a subframe) to the eNodeB such that the burst arrives at the eNodeB at a time that the eNodeB expects the burst, and the burst does not overlap with bursts from other UE that use the same frequency. A UE obtains the delay between itself and the eNodeB through acquisition. The acquisition mechanism specified in the 3GPP standard allows for a minimum delay of 0 and a maximum delay of about 0.33 ms. This is the reason why the maximum cell radius supported by terrestrial LTE is about 100 km for an eNodeB tower centered in the cell. The differential delay between two UEs can be as large as 0.33 ms. Figure 56.2 illustrates the relationship between LTE cell radius and maximum in-cell differential delay. LTE radio resources are divided into 10-m frames in the time dimension, with each frame further subdivided into 1 ms subframe. In the uplink, a 1.08 MHz by n subframe block of resources in a frame, n ¼ 1, 2, or 3, may be configured to be used for acquisition. The 1.08 MHz  n ms block carries the physical random access channel (PRACH). Table 56.1 shows LTE PRACH parameters for four different formats for the frequency division duplex (FDD) operation mode. The four formats offer different levels of robustness to signal deterioration and, relevant to this chapter, support different maximum terminal uplink timing uncertainty at acquisition, or equivalently, the maximum cell radius. The maximum timing uncertainty is the lesser of CP duration and guard time duration. The table also shows the PRACH slot duration to be n ¼ 1, 2, or 3 subframes. It is noted that the relationship among CP duration, GT duration, and Max cell radius is approximately min(CP, GT)  speed of light/2 ¼ max cell radius There are other considerations in designing the exact duration of CP and GT, such as fading channel time spread to be accommodated [1]. The LTE acquisition process is as follows, as partially illustrated in Figure 56.3: 1.

An eNodeB transmits a downlink carrier that embeds (frame and subframe boundary) timing information and contains PRACH (RB) frequency and (frame and subframe) schedule information.

718

Advances in communications satellite systems 2: ICSSC-2019

Table 56.1 LTE PRACH format parameters (all duration in milliseconds) Format CP Z-C Guard time duration duration duration

Slot Max timing duration uncertainly

Max cell radius (km)

0 1 2 3

1 2 2 3

14.6 77.4 29.6 102.6

0.103 0.684 0.203 0.684

0.800 0.800 1.600 1.600

0.097 0.516 0.197 0.716

0.097 0.516 0.197 0.684

PRACH observation window eNodeB subframe Z-C sequence duration

CP

eNodeB Tx subframes

CP

Z-C sequence

PRACH preamble Terminal Rx subframes

CP

Z-C sequence

PRACH slot

GT

Terminal Rx subframe

Figure 56.3 LTE PRACH observation window 2.

3.

4.

5.

A UE monitors and locks (i.e., synchronizes) on to the received downlink carrier frames and subframes and retrieves the PRACH frequency and schedule information. Due to the distance between the terminal and the eNodeB, the UE receives frames and subframes later in time than (i.e., to the right of) those at the eNodeB by an offset equal to the delay between the terminal and the eNodeB. The UE transmits a PRACH burst (in LTE terminology, a PRACH preamble) starting at the very beginning of the receive subframe of a frame, in which PRACH is scheduled. The eNodeB receives the PRACH burst in the PRACH time slot starting at the scheduled subframe and frame and measures its arrival time against eNodeB downlink subframe starting time. Due to the distance between the UE and the eNodeB, the PRACH burst arrives at a time that is twice the delay between the eNodeB and the terminal, after the start of the eNodeB subframe. The eNodeB sends a message addressed to the specific UE with the measured two-times delay between the UE and the eNodeB, called timing advance (TA).

The UE can next apply the TA in its subsequent transmissions to the eNodeB, by shifting the start of a burst by TA amount of time earlier than the start of the intended subframe, so that the burst arrives at the eNodeB at start of the intended

GNSS-assisted acquisition technique for LTE over satellite

719

B Beam (angular) diameter

M ax dif fer en tia l Satellite eNodeB

dis tan ce

Pnear Nadir

Pfar Beam

Max differential delay =

Max differential distance Speed of light

Figure 56.4 Differential delay within a satellite beam subframe. The eNodeB continues to measure terminal burst arrival time variation from eNodeB subframe boundaries and to continuously feeds back the measured timing variation to the UE, allowing the UE to maintain its precise timing in transmission to the eNodeB.

56.2.2 Need for modification to operate over a satellite In a mobile satellite communications system, the differential delay between users in a beam can be much larger 0.33 ms. The maximum differential delay within a beam is a function of beam size, distance of beam center to the satellite nadir, and satellite altitude. This is illustrated in Figure 56.4 in a simple manner without showing the effect of earth curvature. For a GEO satellite that can maintain its orbit with near-0 inclination, and with a typical beam size of 1 diameter, the maximum differential delay is about 2.84 ms within a beam centered at 45 latitude and at the same longitude as that of the satellite. Even with a much smaller beam size of 0.2 diameter, a beam centered at 50 latitude and same satellite longitude leads to a maximum differential delay of 0.70 ms. To conserve fuel for station keeping, a GEO communications satellite for mobile users is typically operated to allow the orbit to be inclined over the life of the satellite, for example, by as much as 7 . With this orbit inclination, the maximum differential delay is about 3.92 ms within a beam of 1 diameter that is centered at 45 latitude and at the same longitude as that of the satellite. Even with a diameter of 0.2 , a beam centered at 50 latitude and same satellite longitude still has a maximum differential delay of 0.97 ms. With larger beam size and/or beams placed further from satellite nadir, the maximum differential delay within a beam can be larger than the examples above.

720

Advances in communications satellite systems 2: ICSSC-2019

Given the larger differential delay, the terrestrial LTE acquisition mechanism is not directly applicable, and adaptations are needed for an LTE over the satellite LTE system.

56.3 Review of prior work in the literature Wei et al. [2] proposes a method that concatenates two Z–C sequences to extend the resolvability of terrestrial cell size. The method requires HW-level change at both UE and eNodeB due to the change of burst format. While not explicitly stated, the included simulation suggests max cell size twice that of LTE max size, that is, about 200 km or 0.67 ms one-way delay. Seo et al. [3] propose a method that has eNodeB baseband equipment precompensate signals by a fixed delay corresponding to a certain distance D, such that the effective cell coverage shifts from normal [0, R] to [D, D þ R] from the eNodeB transmitter, where R is the normal cell radius of a particular PRACH preamble format. For example, if D is set to 100 km, and LTE PRACH preamble Format 3 is used with R ¼ 100 km, then the effective coverage is an area from 100 km to 200 km away from the eNodeB transmitter, provided the link closes. The in-coverage differential delay is still R/c, where c is the speed of light. The maximum in-coverage differential delay is still about 0.33 ms. Lim et al. [4] describe a specific method to precompensate for distance D by shifting an eNodeB FFT time window by D/c. The FFT window is used to capture the content of the LTE PRACH preamble. The effective coverage is the same as that in [3]. Park et al. [5] discuss a two-eNodeB method whereby one eNodeB provides coverage from 0 km to 100 km from the coastline out to sea and the other provides coverage from 80 km to 180 km from the coast, such that between the two eNodeBs, continuous coverage is provided from coastline to 180 km away from the coast. Ha et al. [6] apply similar idea of [3] to the NB-IoT NPRACH process, such that a [D, D þ 35] km coverage can be achieved for NB-IoT service, 35 km is deemed by the authors as the maximum distance that NB-IoT NPRACH can resolve, with corresponding maximum differential delay of 0.12 ms.

56.4 A GNSS-assisted method for LTE acquisition and synchronization over satellite An LTE-over-satellite acquisition mechanism must accommodate a delay between the satellite eNodeB and a terminal located anywhere in a beam. The delay can be decomposed into two parts: (1) the delay between the satellite eNodeB to the point Pnear in the beam that is closest to the satellite—we define this delay as the minimum delay of the beam, and (2) the differential delay to satellite eNodeB between the location of the terminal and Pnear. For a beam, its minimum delay is deterministic, being a function of a satellite position in space, beam radius, and beam center location on earth. On the other hand, the differential delay is nondeterministic prior to the terminal acquiring into the system and is dependent on the

GNSS-assisted acquisition technique for LTE over satellite

721

terminal location within the beam. An important goal of the acquisition process is to determine this differential delay. Figure 56.4 illustrates the geometry of Pnear and maximum value of differential delay. As an analogy, in terrestrial LTE illustrated in Figure 56.2, minimum delay is 0, and differential delay at a UE is the delay between the eNodeB and the UE. It is relatively easy to determine per-beam minimum delay and to accommodate this delay in PRACH procedure either on the eNodeB side [3] or on the UE side [7]. There remains the need to resolve the in-beam differential delay which is longer than standard LTE can resolve. It may look straightforward to define a new PRACH burst format for LTEover-satellite, by increasing both the durations of CP of LTE PRACH preambles and the durations of GT of LTE PRACH slots, such that min(CP, GT)  2max differential delay. But this method has two major drawbacks: 1.

2.

Existing terrestrial LTE implementation of PRACH preamble generation (at terminal) and PRACH preamble detection and estimation (at eNodeB) cannot be reused. The new preamble design would involve fundamental changes to the LTE PRACH design, leading to matters such as a CP being many times longer than the original Z-C sequence or a longer Z-C sequence which in turn further increases the duration of a PRACH slot. The PRACH slot duration would be very long, more than four times the max differential delay. The longer the PRACH slot, the more spectral resources (frequency by time) needs to be dedicated to acquisition, and thus the less spectral resources for user data. The PRACH slot duration may also exceed an LTE frame of 10 ms, leading to further complexities in spectral resource scheduling.

Now we will describe a method that avoids the above drawbacks and the need to determine, and the use of, per-beam minimum delay for an LTE-over-satellite system. Most new satellite communications terminals are equipped with a GPS receiver. A GPS receiver may act as a time source with accuracy better than 1 ms. In an LTE-over-satellite system, a GPS-equipped terminal may measure its time delay from the satellite eNodeB and then determine the timing advance for transmitting a PRACH preamble to acquire into the system. With accurate PRACH preamble transmission timing, accounting for satellite eNodeB to terminal delay, a PRACH preamble can arrive at the satellite eNodeB within a PRACH slot of the same duration that of the terrestrial LTE, shown in Table 56.1. If in a beam, communication service is to be provided only to GPS-equipped terminals, then the PRACH slot duration can be set to one of those of the terrestrial LTE as shown in Table 56.1. It may be straightforward to include the communications satellite ephemeris and satellite eNodeB position in downlink carrier broadcast, for the terminals to use to calculate their respective delays to the satellite eNodeB. But doing so increases signaling overhead and incurs additional processing at eNodeB.

722

Advances in communications satellite systems 2: ICSSC-2019

Terrestrial LTE reference system time is represented by the system frame number (SFN). The SFN is a 10-bit counter that repeats every 1,02410 ms ¼ 10.24 s. The LTE master information block (MIB) in the downlink carrier broadcast includes the upper 8 bits of SFN of the frame in which the MIB is transmitted from the eNodeB. The MIB is scrambled differently for each of the four frames sharing the same upper 8 bits of SFN. Thus, a UE can determine the whole SFN by successfully decoding (including descrambling) the MIB in any one frame. The GMR2-4G method consists of procedures at both the satellite eNodeB and a terminal. At satellite eNodeB: Align the start of the initial SFN cycle to a fixed epoch in GPS time known to the terminals. For example, GPS time 2020-Jan-01 00:00:00 may be defined as the epoch at which SFN ¼ 0. The satellite eNodeB thus must ensure that its SFN count is aligned to GPS time with the particular offset. As a result, for example, at 2020-Jan-01 12:00:15, which is 12 h 15 s after the epoch, the SFN should be (12  3,600 þ 15)  100 modulo 1,024 ¼ 220. At a terminal: 1. 2. 3. 4.

Measure the arrival GPS time tR of a downlink frame with SFN ¼ K; Derive, from K and tR, the transmission GPS time tT from eNodeB of the frame (with SFN ¼ K); Calculate PRACH preamble TA as 2  (tR– tT); and Transmit toward the satellite eNodeB a PRACH preamble with timing adjusted by TA.

In Step 2 above, it is recognized that a particular SFN value of K may correspond to infinitely many GPS time instances spaced at 10.24 s. With tR and the fact that the satellite eNodeB to terminal delay is less than 0.3 s, the correct value of tT can be uniquely resolved: tT is the instance of GPS time corresponding to K that immediately precedes tR and should be within 0.3 s before tR. For example, for SFN ¼ 450, suppose an approximate GPS time tR ¼ 2020-Jan-01 23:00:15.63697 is known to be within 0.3 s ( K elements and M RF chains such that K M