Ultra-Reliable and Low-Latency Communications (URLLC) Theory and Practice E-Book
98,99 €
Mehr erfahren.
- Herausgeber: John Wiley & Sons
- Kategorie: Wissenschaft und neue Technologien
- Sprache: Englisch
Ultra-Reliable and Low-Latency Communications (URLLC) Theory and Practice Comprehensive resource presenting important recent advances in wireless communications for URLLC services, including device-to-device communication, multi-connectivity, and more Ultra-Reliable and Low-Latency Communications (URLLC) Theory and Practice discusses the typical scenarios, possible solutions, and state-of-the-art techniques that enable URLLC in different perspectives from the physical layer to higher-level approaches, aiming to tackle URLLC's challenges with both theoretical and practical approaches, which bridges the lacuna between theory and practice. With long-term contributions to the development of future wireless networks, the text systematically presents a thorough study of the novel and innovative paradigm of URLLC; basic requirements are covered, along with essential definitions, state-of-the-art technologies, and promising research directions of URLLC. To aid in reader comprehension, tables, figures, design schematics, and examples are provided to illustrate abstract engineering concepts and make the text more accessible to a broader readership, and corresponding case studies are included in the last part of the book. Fundamental problems in URLLC, including designing building blocks for URLLC, radio resource management in URLLC, resource optimization, network availability guarantee, and coexisting with other future mobile networks, are also discussed. In Ultra-Reliable and Low-Latency Communications (URLLC) Theory and Practice, readers can expect to find detailed information on: * BCH and analog codes, stable matching, OFDM demodulation and turbo coding, and semi-blind receivers for URLLC * MIMO-NOMA with URLLC, PHY and MAC layer technologies for URLLC, and Network slicing or SDN for URLLC and eMBB * Integrating theoretical knowledge into deep learning for URLLC, Energy-Latency tradeoff in URLLC, and Downlink transmission for URLLC under physical layer aspects * Resource allocation for multi-user downlink URLLC, HARQ optimization for 5G URLLC, and Multi-Access edge computing with URLLC A unique resource with comprehensive yet accessible coverage of a complicated subject, Ultra-Reliable and Low-Latency Communications (URLLC) Theory and Practice is an ideal resource for a large and diverse population of researchers and practitioners in engineering, computer scientists, and senior undergraduate and graduate students in related programs of study.
Sie lesen das E-Book in den Legimi-Apps auf:
Seitenzahl: 512
Veröffentlichungsjahr: 2023
Ähnliche
Ultra-Reliable and Low-Latency Communications (URLLC) Theory and Practice
Advances in 5G and Beyond
Edited by Trung Q. Duong, Saeed R. Khosravirad, Changyang She, Petar Popovski, Mehdi Bennis and Tony Q.S. Quek
This edition first published 2023
© 2023 John Wiley & Sons Ltd
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.
The right of Trung Q. Duong, Saeed R. Khosravirad, Changyang She, Petar Popovski, Mehdi Bennis and Tony Q.S. Quek to be identified as the authors of the editorial material in this work has been asserted in accordance with law.
Registered Offices
John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK
For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com.
Wiley also publishes its books in a variety of electronic formats and by print-on-demand. Some content that appears in standard print versions of this book may not be available in other formats.
Trademarks: Wiley and the Wiley logo are trademarks or registered trademarks of John Wiley & Sons, Inc. and/or its affiliates in the United States and other countries and may not be used without written permission. All other trademarks are the property of their respective owners. John Wiley & Sons, Inc. is not associated with any product or vendor mentioned in this book.
Limit of Liability/Disclaimer of Warranty
While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials or promotional statements for this work.This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.
A catalogue record for this book is available from the Library of Congress
Hardback ISBN: 9781119818304; ePub ISBN: 9781119818335; ePDF ISBN: 9781119818311; oBook ISBN: 9781119818366
Cover image: © Krunja/Shutterstock
Cover design by Wiley
Set in 9.5/12.5pt STIXTwoText by Integra Software Services Pvt. Ltd, Pondicherry, India
Contents
Cover
Title page
Copyright
Preface
List of Contributors
1 URLLC: Faster, Higher, Stronger, and Together
2 Statistical Characterization of URLLC: Frequentist and Bayesian Approaches
3 Characterizing and Taming the Tail in URLLC
4 Unsupervised Deep Learning for Optimizing Wireless Systems with Instantaneous and Statistic Constraints
5 Channel Coding and Decoding Schemes for URLLC
6 Sparse Vector Coding for Ultra-reliable and Low-latency Communications
7 Network Slicing for URLLC
8 Beamforming Design for Multi-user Downlink OFDMA-URLLC Systems
9 A Full-Duplex Relay System for URLLC with Adaptive Self-Interference Processing
10 Mobility Prediction for Reducing End-to-End Delay in URLLC
11 Relay Robot-Aided URLLC in 5G Factory Automation with Industrial IoTs
Index
End User License Agreement
List of Tables
CHAPTER 01
Table 1.1 KPIs and research...
CHAPTER 02
Table 2.1 Estimated hyperparameters.
CHAPTER 04
Table 4.1 Simulation parameters.
Table 4.2 Number of time slots...
CHAPTER 05
Table 5.1 Table of notations.
CHAPTER 06
Table 6.1 Example of mapping...
Table 6.2 PDCCH versus SVC technique.
Table 6.3 The MMP-based SVC...
Table 6.4 The PL-SVC Decoding...
CHAPTER 08
Table 8.1 System parameters [21].
CHAPTER 09
Table 9.1 Conditions of Adaptive...
Table 9.2 Evaluation of latency...
CHAPTER 10
Table 10.1 Simulation Parameters [2].
Table 10.2 Prediction error probability...
Table 10.3 Prediction error probability...
CHAPTER 11
Table 11.1 The effectiveness of...
List of Illustrations
CHAPTER 01
Figure 1.1 Wireless AI for...
CHAPTER 02
Figure 2.1 Illustration of industrial...
Figure 2.2 Normalized throughput with...
Figure 2.3 Probability density for...
Figure 2.4 Map with 5 observed...
Figure 2.5 Realization of GPs...
Figure 2.6 Regression for the...
Figure 2.7 Estimated parameters and...
Figure 2.8 Estimation error for...
Figure 2.9 Statistical radio map...
Figure 2.10 500 transmitter locations...
Figure 2.11 Predictive means for...
Figure 2.12 Histogram for outage...
CHAPTER 03
Figure 3.1 Analysis of the...
Figure 3.2 Effectiveness of characterizing...
Figure 3.3 Effectiveness of characterizing...
Figure 3.4 Learning accuracy versus...
Figure 3.5 Time overheads and...
Figure 3.6 Improvement of the...
Figure 3.7 Trade-offs between...
Figure 3.8 AoI versus decoding...
CHAPTER 04
Figure 4.1 Large-scale channels...
Figure 4.2 Complementary cumulative distributions...
Figure 4.3 Total bandwidth required...
CHAPTER 05
Figure 5.1 Comparison of error...
Figure 5.2 Comparison of error...
Figure 5.3 Comparison of different...
Figure 5.4 Algorithmic complexity versus...
Figure 5.5 The BLER performance...
Figure 5.6 The approximation of...
Figure 5.7 The distributions of...
Figure 5.8 ...
Figure 5.9 Decoding (64,30,14)...
Figure 5.10 Decoding (64,30,14)...
Figure 5.11 The comparisons of...
Figure 5.12 The average time...
CHAPTER 06
Figure 6.1 Metaphoric illustration of...
Figure 6.2 SVC-based packet...
Figure 6.3 Snapshot of the...
Figure 6.4 BLER performance of...
Figure 6.5 Decoding failure as...
Figure 6.6 BLER performance as...
Figure 6.7 BLER performance for...
Figure 6.8 Required SNR for...
Figure 6.9 Probability of transmission...
Figure 6.10 BLER performance as...
Figure 6.11 PER performance of...
Figure 6.12 PER performance of...
Figure 6.13 PER performance of...
Figure 6.14 Transceiver structure of...
Figure 6.15 Empirical simulation and...
Figure 6.16 BLER performance as...
Figure 6.17 BLER performance as...
Figure 6.18 Probability of transmission...
CHAPTER 07
Figure 7.1 A RAN slicing...
Figure 7.2 A convergence curve...
Figure 7.3 Trend of the...
Figure 7.4 Trends of transmit...
Figure 7.5 Trend of the...
Figure 7.6 Trends of URLLC...
CHAPTER 08
Figure 8.1 Multi-user downlink...
Figure 8.2 ASST [bits/s/Hz] versus...
Figure 8.3 ASST [bits/s/Hz] versus...
CHAPTER 09
Figure 9.1 Emerging devices and...
Figure 9.2 Latency diagram for...
Figure 9.3 Diagram of residual...
Figure 9.4 For traditional DF...
Figure 9.5 An AF FD...
Figure 9.6 An AF FD...
Figure 9.7 BER performance of...
Figure 9.8 BER performance of...
Figure 9.9 BER performance of...
Figure 9.10 The probability of...
CHAPTER 10
Figure 10.1 Illustration of network...
Figure 10.2 Illustration of prediction...
Figure 10.3 Joint optimization of...
Figure 10.4 Comparison of reliability...
Figure 10.5 ...
Figure 10.6 Required total bandwidth...
Figure 10.7 Experiment to obtain...
CHAPTER 11
Figure 11.1 The system model...
Figure 11.2 Learning-based optimization...
Figure 11.3 Examples of industrial...
Figure 11.4 The worst-case...
Figure 11.5 The worst-case...
Figure 11.6 The comparison of...
Figure 11.7 The comparison of...
Figure 11.8 The comparison of...
Guide
Cover
Title page
Copyright
Table of Contents
Preface
List of Contributors
Begin Reading
Index
End User License Agreement
Pages
i
ii
iii
iv
v
vi
vii
viii
ix
x
xi
xii
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
Preface
Pursuing ever higher data rates has been the central design goal in all the previous generations of mobile communications. This has been changed in the 5th generation (5G) mobile communications that aims to support various new emerging services with diverse and stringent quality-of-service requirements. The most formidable challenge in 5G is to achieve Ultra-Reliable Low-Latency Communications (URLLC) for many mission-critical services including autonomous vehicles, industry automation, and tele-robotic surgery, i.e. the roundtrip delay of 1 millisecond and less than 1 out of a million in packet loss. In the 4th generation (4G) systems, the average latency is usually a few hundred milliseconds, and the packet loss probability is around 1%. 5G systems need to significantly improve the latency and reliability by several orders of magnitude compared to 4G systems. This presents unprecedented challenges.
This book covers a range of topics from fundamental theories to practical solutions in URLLC.
In Chapters 2 and 3, the authors analyze the statistical features and tail distributions of wireless channels and provide useful insights on the performance of URLLC. Chapter 2 presents the statistical aspects of URLLC in both frequentist and Bayesian approaches. The authors have analyzed the statistical features and guarantees for outage probability in a narrowband wireless channel. Chapter 3 considers various metrics of URLLC including tail distribution, higher-order statistics, extreme events with very low occurrence probabilities, worst-case metrics, and reliability/latency. The authors have introduced readers to the entropic risk measure in financial mathematics, generalized extreme value distribution, and generalized Pareto distribution to investigate these metrics.
In Chapters 4–7, the authors have introduced several techniques to guarantee the reliability and latency of URLLC, including machine learning, candidate channel codes, sparse vector coding, and network slicing. Two problems of resource allocation in URLLC are addressed in Chapter 4 with an unsupervised learning approach. The results have shown that bandwidth utilization efficiency of URLLC can be improved more significantly by exploiting frequency diversity than by multi-user diversity. Chapter 5 discusses the channel coding and decoding schemes for URLLC. This chapter reviews state-of-the-art channel codes for URLLC and analyzes them in terms of performance and complexity. Furthermore, the Ordered Statistics Decoding (OSD) is promoted as one of the potential universal decoding algorithms for URLLC. In Chapter 6, a new approach to support short packet transmissions, referred as Sparse Vector Coding (SVC) is introduced. The numerical evaluations and performance analysis which validate the proposed SVC technique is highly effective in URLLC transmissions. Chapter 7 studies a CoMP-enabled RAN slicing system simultaneously supporting URLLC and eMBB services. The authors address a joint bandwidth and CoMP beamforming optimization problem to maximize the long-term total slice utility.
In Chapters 8 and 9, downlink Orthogonal Frequency Division Multiple Access systems (OFDMA) and full-duplex relay system are optimized for URLLC, respectively. Chapter 8 investigates the beamforming design for downlink ODFMA to enable the stringent delay requirement. In particular, the authors address a non-convex optimization problem to maximize the weighted system sum throughput subject to Quality-of-Service (QoS) of URLLC users. Chapter 9 presents an up-to-date overview of the end-to-end latency for a Full-Duplex (FD) relay system in the context of URLLC. The authors not only provide an insightful investigation of reliability and latency together for FD relay assisted URLLC but also discuss possible relaying latency reduction solutions in the chapter.
In Chapters 10 and 11, the authors investigate URLLC in vertical industries: Tactile Internet and Industrial Internet-of-Things. More specifically, Chapter 10 addresses an optimization problem that maximizes the number of URLLC services by jointly optimizing time and frequency resources and the prediction horizon. The numerical results clearly demonstrate the effectiveness of the proposed solution. In addition, a proof-of-concept experiment with the remote control in a virtual factory is also provided to illustrate a typical application of Tactile Internet. Finally, Chapter 11 considers relay robots-aided URLLC in 5G factory automation, which consists of multiple relay robot deployment and decoding error probability minimization problems. There are two different approaches introduced for relay robot deployment, including Deep Neural Networks (DNN) and the K-means clustering algorithm. A low-complexity iterative algorithm is also provided to deal with the joint blocklength and power allocation problem to minimize the decoding error probability.
List of Contributors
Mehdi BennisCentre for Wireless Communications University of Oulu Oulu, Finland
Xianbin CaoSchool of Electronics and Information Engineering Beihang University Beijing, China
Jingxuan ChenSchool of Electronics and Information Engineering Beihang University Beijing, China
Hanjun DuanThe School of Electronic and Information Engineering Harbin Institute of Technology Shenzhen, China
Trung Q. DuongSchool of Electronics Electrical Engineering and Computer Science Queen’s University Belfast, UK
Walid R. GhanemFriedrich-Alexander-University Erlangen-Nuremberg (FAU) Erlangen, Germany
Zhanwei HouThe School of Electrical and Information Engineering The University of Sydney Sydney, Australia
Yung-Lin HsuGraduate Institute of Communication Engineering National Taiwan University Taipei, Taiwan
Dang Van HuynhSchool of Electronics Electrical Engineering and Computer Science Queen’s University Belfast, UK
Vahid JamaliTechnical University Darmstadt Darmstadt, Germany
Yufei JiangThe School of Electronic and Information Engineering Harbin Institute of Technology Shenzhen, China
Tobias KallehaugeConnectivity Section at the Department of Electronic Systems Aalborg University Aalborg, Denmark
Anders E. KalørConnectivity Section at the Department of Electronic Systems Aalborg University Aalborg, Denmark
Saeed R. KhosraviradNokia Bell Labs Murray Hill New Jersey, US
Yonghui LiSchool of Electrical and Information Engineering University of Sydney Sydney, Australia
Chen-Feng LiuTechnology Innovation Institute Masdar City Abu Dhabi, UAE
Antonino MasaracchiaSchool of Electronics Electrical Engineering and Computer Science Queen’s University Belfast, UK
Yuexing PengBeijing University of Posts and Telecommunications Beijing, China
Petar PopovskiConnectivity Section at the Department of Electronic Systems Aalborg University Aalborg, Denmark
Tony Q.S. QuekInformation Systems Technology and Design Singapore University of Technology and Design Singapore
Pablo Ramirez-EspinosaConnectivity Section at the Department of Electronic Systems Aalborg University Aalborg, Denmark
Robert SchoberFriedrich-Alexander-University Erlangen-Nuremberg (FAU) Erlangen, Germany
Changyang SheThe School of Electrical and Information Engineering The University of Sydney Sydney, Australia
Byonghyo ShimInstitute of New Media and Communications and Department of Electrical and Computer Engineering Seoul National University Seoul, Korea
Mahyar ShirvanimoghaddamThe School of Electrical and Information Engineering The University of Sydney Sydney, Australia
Chengjian SunSchool of Electronics and Information Engineering Beihang University Beijing, China
Branka VuceticThe School of Electrical and Information Engineering The University of Sydney Sydney, Australia
Hung-Yu WeiDepartment of Electrical Engineering National Taiwan University Taipei, Taiwan
Dapeng WuDepartment of Electrical and Computer Engineering University of Florida Gainesville, USA
Xing XiSchool of Electronics and Information Engineering Beihang University Beijing, China
Chenyang YangSchool of Electronics and Information Engineering Beihang University Beijing, China
Peng YangInformation Systems Technology and Design Singapore University of Technology and Design Singapore
Chentao YueThe School of Electrical and Information Engineering The University of Sydney Sydney, Australia
Fu-Chun ZhengThe School of Electronic and Information Engineering Harbin Institute of Technology Shenzhen, China
Xu ZhuThe School of Electronic and Information Engineering Harbin Institute of Technology Shenzhen, China
1 URLLC: Faster, Higher, Stronger, and Together
Changyang She1,*, Trung Q. Duong2, Saeed R. Khosravirad3, Petar Popovski4, Mehdi Bennis5, and Tony Q.S. Quek6
1School of Electrical and Information Engineering, University of Sydney, 2006, NSW, Australia2School of Electronics Electrical Engineering and Computer Science, Queen’s University Belfast, BT7 1NN, Belfast, UK3Nokia Bell Labs, NJ 07974-0636, Murray Hill, USA4Connectivity Section at the Department of Electronic Systems, Aalborg University, 9220, Aalborg, Fredrik Bajers Vej 7A, Denmark5Centre for Wireless Communications, University of Oulu, Oulu, Finland6Information Systems Technology and Design, Singapore University of Technology and Design, 487372, Singapore*Corresponding Author
As one of the new communication scenarios in the 5th Generation (5G) mobile communications, Ultra-Reliable and Low-Latency Communications (URLLC) are crucial for enabling a wide range of emerging applications, including industry automation, intelligent transportation, telemedicine, Tactile Internet, and Virtual/Augmented Reality (VR/AR). According to the requirements in 5G standards, to support emerging mission-critical applications, the End-to-End (E2E) delay cannot exceed 1 ms and the packet loss probability should be –. Compared with the existing cellular networks, the delay and reliability require significant improvements by at least two orders of magnitude for 5G networks. This capability gap cannot be fully resolved by the 5G New Radio (NR), i.e. the physical-layer technology for 5G, even though the transmission delay in Radio Access Networks (RANs) achieves the ms target. Transmission delay contributes only a small fraction of the E2E delay, as the stochastic delays in upper networking layers, such as queuing delay, processing delay, and access delay, are key bottlenecks for achieving URLLC. Beyond 5G systems or so-called 6th Generation (6G) systems should guarantee the E2E delay bound with high reliability.
In addition to the latency and reliability requirements, some other Key Performance Indicators (KPIs) should also be taken into account, including Spectrum Efficiency (SE), throughput, Energy Efficiency (EE), Age of Information (AoI), jitter of latency, round-trip delay, network availability, and security (shown in Table 1.1). These requirements will pose unprecedented challenges in terms of design methodologies and enabling technologies in the 6th Generation (6G) mobile communications. To fill the gap between 5G URLLC and the diverse KPI requirements, we shall investigate novel methodologies and innovative technologies for the next generation URLLC (xURLLC), also known as eXtreme URLLC, [23]. This book will cover various methods and technologies to achieve URLLC from the physical layer, link layer, and network layer, to diverse applications in vertical industries of 5G/6G communications.
Table 1.1 KPIs and research challenges. © IEEE 2021. Reprinted with permission from [28].
Indoor large-scale scenarios
Applications
Factory automation
KPIs
SE, EE, and AoI
Research Challenges
Scalability and network congestions
VR/AR applications
SE and throughput
Processing/transmission 3D videos
Indoor wide-area scenarios
Applications
KPIs
Research Challenges
Tele-surgery
Round-trip delay, throughput, and jitter
Propagation delay and high data rate
eHealth monitoring
EE and network availability
Propagation delay and localization
Outdoor large-scale scenarios
Applications
KPIs
Research Challenges
Vehicle safety
AoI, SE, security, and network availability
High mobility and scalability
Outdoor wide-area scenarios
Applications
KPIs
Research Challenges
Smart grid
SE
Propagation delay and scalability
Tele-robotic control
SE, security, network availability, and jitter
Propagation delay and high data rate
UAV control
EE, security, network availability, and AoI
Propagation delay and high mobility
1.1 Requirements of URLLC: Faster, Higher, Stronger, and Together
The next generation URLLC is expected to be “faster, higher, stronger - together”. The specific requirements and research challenges are discussed in the sequel.
1.1.1 Faster Responses and Movement
In factory automation and autonomous vehicles, mobile devices need to make decisions according to their local observation in a real-time manner. Given the fact that the energy budget and the computing capability of each device are limited, it may need the help of Mobile Edge Computing system (MEC). Unlike centralized mobile cloud computing with routing delay and propagation delay in backhauls and core networks, the E2E delay in MEC systems consists of Uplink (UL) and Downlink (DL) transmission delays, queuing delays in the buffers of users and Base Stations (BSs), and the processing delay in the MEC [26]. Although MEC helps reduce latency in communication systems, there are two bottlenecks for providing fast responses to mobile devices. First, optimization problems in MEC systems are generally non-convex. To find the optimal solution, such as user association and task offloading. The computing complexity for executing searching algorithms is too high to be implemented in real-time. Second, exchanging information among different edge servers will lead to high overheads and latency. To avoid this issue, edge servers need to make decisions in a distributed manner. As a result, a local decision may not be optimal for all the devices.
Supporting high mobility URLLC is critical for some outdoor applications, e.g. Unmanned Aerial Vehicle (UAV) control. Since the mobile devices are moving fast, the Doppler frequency shift is large. Thus, the inter-symbol interference is strong and the receiver needs to adjust the carrier frequency according to the Doppler shift. Besides, frequent handovers in high mobility URLLC will result in service interruption. The BS with good channel quality may not have sufficient radio resources due to the dynamic traffic load in high mobility scenarios. Therefore, how to serve high mobility URLLC remains an open issue in 6G.
1.1.2 Higher Throughput and Density
As one of the killer applications in 5G networks, VR/AR applications require ultra-reliable and low-latency tactile feedback and high data rate videos [8]. Meanwhile, as the sizes of devices shrinks, battery lifetime will become a bottleneck for enabling high data rate URLLC [24]. To implement VR/AR applications in future wireless networks, we need to investigate the fundamental trade-offs among throughput, energy efficiency, reliability, and latency in communications, caching, and computing systems [34], as well as enabling technologies such as touch user interface and haptic codecs [4, 31].
Due to the explosive growth of the numbers of autonomous vehicles and mission-critical IoT devices [9], future wireless networks are expected to support massive URLLC. To support massive URLLC, novel communication and learning techniques are needed. With orthogonal multiple access technologies, the required bandwidth increases linearly with the number of devices. To achieve better trade-offs among delay, reliability, and scalability, other multiple access technologies should be used, such as non-orthogonal multiple access and contention-based multiple access technologies [29, 30]. Meanwhile, we may need to exploit the above GHz spectrum including mmWave [16] and the Terahertz band [36].
1.1.3 Stronger Connectivity and Security
Multi-connectivity is a promising approach to provide seamless services to URLLC users [17]. As illustrated in [22], one way to improve network availability without sacrificing spectrum efficiency is to serve each user with multiple BSs over the same subchannel (or subcarrier). The disadvantage of this intra-frequency multi-connectivity is that the failures of different links are highly correlated. For example, if there is a strong interference on a subchannel, then the signal to interference plus noise ratios of all the links are low. To alleviate cross-correlation among different links, different nodes can connect to one user with different subchannels or even with different communication interfaces [21]. Further considering that terrestrial networks may not cover rural areas and marine areas, we need to use non-terrestrial networks to provide global connectivity for long-distance URLLC services, e.g. Tactile Internet.
Future URLLC systems will suffer from different kinds of attacks that result in inefficient communications [19]. The widely used cryptography algorithms require high-complexity signal processing, and may not be suitable for URLLC, especially for IoT devices with low computing capacities. To defend against eavesdropping attacks in URLLC, physical layer security is a viable solution [3]. The maximal secret communication rate in the short blocklength regime over a wiretap channel was derived in [37]. The results show that there are trade-offs among delay, reliability, and security. Based on this fundamental result, we can further investigate the technologies for improving physical-layer security.
1.1.4 Human Intelligence Together with Artificial Intelligence in URLLC
As illustrated in Figure 1.1, a new trend of developing communication networks is to integrating human intelligence (expert knowledge in wireless communications) into artificial intelligence (deep learning) for optimizing communication systems.
Figure 1.1 Wireless AI for developing URLLC systems. © IEEE 2021. Reprinted with permission from [28].
Cross-layer Design Existing design methods divide communication networks into multiple layers according to the Open Systems Interconnection model [18]. Communication technologies in each layer are often developed without considering the impacts on other layers, despite the fact that the interactions across different layers are known to significantly impact on the E2E delay and reliability. Most existing approaches do not reflect such interactions; this leads to suboptimal solutions and thus we are yet to be able to meet the stringent requirements of URLLC. To guarantee the E2E delay and the reliability of the communication system, we need accurate and analytically tractable cross-layer models to reflect the interactions across different layers.
Deep Learning With 5G NR, the radio resources are allocated in each Transmission Time Interval (TTI) with a duration of 0.125 1 ms [1]. To implement optimization algorithms in 5G systems, the processing delay should be less than the duration of one TTI. Since the crosslayer models are complex, related optimization problems are non-convex in general. Most of the existing optimization algorithms incur high computing overheads, and hence can hardly be implemented in real-world systems. Deep learning has significant potential to address the above issue in beyond 5G/6G networks. The basic idea is to approximate the optimal policy with a Deep Neural Network (DNN). After the training phase, a near-optimal solution of an optimization problem can be obtained from the output of the DNN in each TTI. Essentially, by using deep learning, we are trading off the online processing time with the computing resource for off-line training.
Integrating Knowledge into Learning Algorithms Although deep learning algorithms have shown significant potential, the application of deep learning in URLLC is not straightforward. As shown in [27], deep learning algorithms converge slowly in the training phase and need a large number of training samples to evaluate or improve the E2E delay and reliability. If some knowledge of the environment is available, such as the estimated packet loss probability of a certain decision, the system can exploit this knowledge to improve the learning efficiency [13]. Domain knowledge of communications and networking including models, analytical tools, and optimization frameworks have been extensively studied in the existing literature [12, 14]. How to exploit them to improve deep learning algorithms for URLLC has drawn significant attention as well, including in [32, 15].
Fine-tuning in Real-world Systems Communication environments in wireless networks are non-stationary in general. Theoretical models used in off-line training may not match this non-stationary nature of practical networks. As a result, a DNN trained off-line cannot guarantee the Quality-of-Service (QoS) constraints of URLLC. Such an issue is referred to as the model mismatch problem in [5]. To handle the model mismatch, wireless networks should be intelligent to adjust themselves in dynamic environments, explore unknown optimal policies, and transfer knowledge to practical networks.
1.2 Scope of This Book
We invited leading researchers in both academia and industry from diverse backgrounds to share their recent studies in URLLC.
In Chapter 2, T. Kallehauge et al. focus on the physical layer and present the statistical aspects of URLLC, detailing both frequentist and Bayesian approaches. Specifically, the authors analyze the statistical features and guarantees for outage probability in a narrowband wireless channel. As a motivating example, they treat the practical case in which a Base Station (BS) collects channel statistics for users at different locations and attempts to predict the performance of a user at a new location. Their results show that the BS can obtain high-quality predictions of the reliability performance even for locations that are not in proximity.
In Chapter 3, instead of analyzing the average latency and the delay outage probability, C.-F. Liu et al. investigate the statistical information and/or metrics, rooted in the tail behavior of probability distributions to gain insights in URLLC systems. Specifically, the authors analyzed the tail distribution of the delay, channel fading, or packet inter-arrival time; variance and higher-order statistics; threshold deviation with a very low occurrence probability; worst-case metrics; age of information. Useful methodologies and extensive numerical results are provided in this chapter.
In Chapter 4, C. Sun et al. establish a unified framework of using unsupervised deep learning to solve both kinds of problem with both instantaneous and statistic constraints. For a constrained variable optimization, the authors first convert it into an equivalent functional optimization problem with instantaneous constraints. Then, to ensure the instantaneous constraints in the functional optimization problems, the authors use DNN to approximate the Lagrange multiplier functions, which is trained together with a DNN to approximate the policy. By taking resource allocation problems in URLLC as examples, the authors show that unsupervised learning outperforms supervised learning in terms of quality-of-service violation probability and approximation accuracy of the optimal policy.
In Chapter 5, C. Yue et al. overview candidate channel codes for URLLC, and compare them in terms of performance and complexity. Their respective strengths and weaknesses are investigated in terms of the performance gap to theoretical limits and the computational complexity of practical decoding algorithms. Furthermore, Ordered Statistics Decoding (OSD) is introduced as one of the potential universal decoding algorithms for URLLC, which can achieve near-optimal performance for any block code. The error performance and computational complexity of OSD are investigated in this chapter. Finally, recent improvements on OSD, including decoding rules and complete decoder design, are studied.
In Chapter 6, B. Shim et al. introduce a new type of short packet transmission framework named the Sparse Vector Coding (SVC) technique. The key idea behind SVC is to transform an information vector into the sparse vector in the transmitter and to exploit the sparse recovery algorithm in the receiver. Metaphorically, SVC can be thought as marking dots on the empty table. As long as the number of dots is small enough and the measurements contains enough information to figure out the marked cell positions, accurate decoding of the SVC packet can be guaranteed. The numerical results demonstrate that SVC is very effective in the short packet transmission for URLLC scenarios.
In Chapter 7, P. Yang et al. consider a CoMP-enabled RAN slicing system simultaneously supporting URLLC and eMBB traffic transmission. In the presence of eMBB traffic, the authors orchestrate the shared network resources of the system to guarantee a more reliable bursty URLLC service provision from the perspectives of lowering both URLLC packet blocking probability and codeword decoding error probability. The authors formulate the problem of RAN slicing for bursty URLLC and eMBB service multiplexing as a resource optimization problem and develop a joint bandwidth and CoMP beamforming optimization algorithm to maximize the long-term total slice utility. Several bandwidth allocation and beamforming algorithms are evaluated in the RAN slicing system.
In Chapter 8, W. R. Ghanem et al. investigate the beamforming design for downlink Orthogonal Frequency Division Multiple Access (OFDMA) URLLC systems. To enable the stringent URLLC delay requirements, finite blocklength transmission is adopted for the beamforming algorithm design. The authors formulate the beamforming algorithm design as a non-convex optimization problem for maximization of the weighted system sum throughput subject to constraints on the Quality of Service (QoS) of the URLLC users. A sub-optimal algorithm is proposed based on Sequential Convex Approximation (SCA). Numerical results reveal that the proposed design can achieve a considerable gain compared to several baseline schemes.
In Chapter 9, H. Duan et al. first present an up-to-date overview of the end-to-end latency for a Full-Duplex (FD) relay system. The authors investigate the possible solutions in the literature to achieve the goal of URLLC. The efficient solution is to allow a simple Amplify-and-Forward (AF) FD relay mode with low-complexity SI radio frequency and analog cancellations, and process the residual SI alongside the desired signal at base station in an adaptive manner, rather than being canceled at relay in digital domain. Their results show that the FD relay assisted system with adaptive SI utilization or cancellation enables extended network coverage, enhanced reliability, and reduced latency, compared to the existing overview work.
In Chapter 10, Z. Hou et al. aim to reduce the user experienced delay through prediction and communication co-design, where each mobile device predicts its future states and sends them to a data center in advance. Since predictions are not error-free, the authors consider prediction errors and packet losses in communications when evaluating the reliability of the system. Then, the authors formulate an optimization problem that maximizes the number of URLLC services supported by the system by optimizing time and frequency resources and the prediction horizon. Simulation and experiment results verify the effectiveness of the proposed method, and show that the trade-off between user experienced delay and reliability can be improved significantly via prediction and communication co-design.
In Chapter 11, D. V. Huynh et al. investigate the URLLC supported Industrial Internet-of-Things (IIoT) devices in industry automation. To enhance the URLLC performance, the authors propose two approaches to optimize the deployment of multiple relay robots in assisting URLLC system, namely DNN-based deployment, and the K-means clustering algorithm. An optimal resource allocation scheme is proposed to minimize the error probability at the IIoT devices. To solve the highly non-covex optimization problems of URLLC, the authors propose an effective iterative algorithm for solving the reliability maximization. Representative numerical results demonstrate the proposed scheme can significantly improve the reliability over various conventional approaches.
1.3 Future Directions
Considering the diverse application scenarios and KPIs of URLLC in 6G, the design of future communication systems requires considerable additional research efforts beyond what the community has done so far. We discuss some promising research directions in this section.
1.3.1 Constrained Deep Learning for URLLC
When applying deep learning for URLLC applications, the reward is usually defined as a weighted sum of different KPIs. With different weighting coefficients, the final achieved KPIs are different. Thus, we need to select these weighting coefficients manually to achieve satisfactory KPIs. A potential way to overcome this difficulty is to formulate the problem as a constrained optimization problem and use constrained unsupervised deep learning to find the optimal policy [7]. If the problem turns out to be a sequential-decision making problem with constraints, constrained deep reinforcement learning algorithms can be applied [20]. Nevertheless, the reliability of URLLC is extremely high and the closed-form results may not be available. Therefore, the required number of training samples is extremely large. How to achieve the target KPIs with a reasonable amount of training samples remains an open issue.
1.3.2 Distributed Learning for URLLC
In IIoT, collecting the status of all the devices in the central server will bring considerable overheads to wireless networks, and hence the devices should be controlled in a distributed manner. To achieve this goal, distributed learning with partial observation is a promising framework. With this framework, edge computing servers, VR glasses, and IoT devices can take actions based on local observations. In this way, the overheads for exchanging control information and updating global status can be reduced remarkably [25]. The major issues of distributed learning include long convergence time, poor performance with partial observation, and limited computing and storage resources of mobile devices.
1.3.3 Graph Neural Networks for Network Management of URLLC
Since the dimensions of the input and the output of a Fully-connected Neural Network (FNN) grows with the number of devices, we need to adjust the hyper-parameters of the FNN and retrain the parameters whenever the number of devices varies. Thus, FNNs are not flexible in managing dynamic networks. To address this issue, a promising approach is to use GNNs to represent the topology of wireless networks [11]. As indicated in [35], GNN is a very general structure that can be applied to solve large-scale problems with non-Euclid data structure. Since the number of parameters of a GNN does not increase with the dimension of the input, GNNs are suitable for resource management in dynamic networks [6]. In most of the existing literature, the network status is assumed to be perfectly known at the central control plane. How to find optimal policy with inaccurate/outdated/partial network status remain open problems.
1.3.4 Few-shot Learning for URLLC
The real-world data samples from practical systems are limited and could be non-stationary in wireless networks. A promising learning framework for fast adaptation is known as few-shot learning [33]. The basic idea is to use meta learning to optimize hyper-parameters, including initial parameters, learning rates, and the structures of neural networks [10, 2]. After off-line training in existing tasks, the neural networks can be transferred to new tasks with limited data samples. Most of the existing few-short learning algorithms for image classification cannot achieve high reliability. Whether it is possible to meet the reliability requirement of URLLC with few-shot learning remains open.
Bibliography
[1]
3GPP. Study on new radio (NR) access technology; Physical layer aspects (release 14). Stockholm, Sweden, Jun.2017. TR 38.802 V2.0.0.
[2]
Marcin Andrychowicz, Misha Denil, Sergio Gomez, Matthew W Hoffman, David Pfau, Tom Schaul, Brendan Shillingford, and Nando De Freitas. Learning to learn by gradient descent by gradient descent. In
Proc. Neural information processing systems (NIPS)
, pages 3981–3989, 2016.
[3]
Riqing Chen, Chunhui Li, Shihao Yan, Robert Malaney, and Jinhong Yuan. Physical layer security for ultra-reliable and low-latency communications.
IEEE Wireless Commun.
, 26 (5): 6–11, 2019.
[4]
Victor Adriel de Jesus Oliveira, Luciana Nedel, and Anderson Maciel. Assessment of an articulatory interface for tactile intercommunication in immersive virtual environments.
Computers & Graphics
, 76: 18–28, 2018.
[5]
Rui Dong, Changyang She, Wibowo Hardjawana, Yonghui Li, and Branka Vucetic. Deep learning for radio resource allocation with diverse quality-of-service requirements in 5G.
IEEE Trans. Wireless Commun.
, 20 (4): 2309–2324, 2020.
[6]
Mark Eisen and Alejandro Ribeiro. Optimal wireless resource allocation with random edge graph neural networks.
IEEE Trans. Signal Process.
, 68: 2977–2991, 2020.
[7]
Mark Eisen, Clark Zhang, Luiz FO Chamon, Daniel D Lee, and Alejandro Ribeiro. Learning optimal resource allocations in wireless systems.
IEEE Trans. Signal Process.
, 67 (10): 2775–2790, 2019.
[8]
Mohammed S Elbamby, Cristina Perfecto, Mehdi Bennis, and Klaus Doppler. Toward low-latency and ultra-reliable virtual reality.
IEEE Netw.
, 32 (2): 78–84, 2018.
[9]
Ericsson. Internet of things forecast.
Ericsson Mobility Report
, 2019. URL
https://www.ericsson.com/en/mobility-report/internet-of-things-forecast
.
[10]
Chelsea Finn, Pieter Abbeel, and Sergey Levine. Model-agnostic meta-learning for fast adaptation of deep networks. In
Proc. International Conference on Machine Learning (ICML)
, pages 1126–1135, 2017.
[11]
Fernando Gama, Joan Bruna, and Alejandro Ribeiro. Stability properties of graph neural networks.
IEEE Trans. Signal Process.
, 68: 5680–5695, 2020.
[12]
Andrea Goldsmith.
Wireless Communications
. Cambridge university press, 2005.
[13]
Shixiang Gu, Timothy Lillicrap, Ilya Sutskever, and Sergey Levine. Continuous deep Q-learning with model-based acceleration. In
Proc. International conference on machine learning (ICML)
, pages 2829–2838. PMLR, 2016.
[14]
Mor Harchol-Balter.
Performance modeling and design of computer systems: Queueing Theory in Action
. Cambridge University Press, 2013.
[15]
Hengtao He, Shi Jin, Chao-Kai Wen, Feifei Gao, Geoffrey Ye Li, and Zongben Xu. Model-driven deep learning for physical layer communications.
IEEE Wireless Commun.
, 26 (5): 77–83, 2019.
[16]
Ibrahim A Hemadeh, Katla Satyanarayana, Mohammed El-Hajjar, and Lajos Hanzo. Millimeter-wave communications: Physical channel models, design considerations, antenna constructions, and link-budget.
IEEE Commun. Surveys & Tuts.
, 20 (2): 870–913, 2017.
[17]
Tom Hobler, Lucas Scheuvens, Norman Franchi, Meryem Simsek, and Gerhard P. Fettweis. Applying reliability theory for future wireless communication networks. In
Proc. IEEE Personal, Indoor and Mobile Radio Communications (PIMRC)
, 2017.
[18]
Xiaolin Jiang, Hossein Shokri-Ghadikolaei, Gabor Fodor, Eytan Modiano, Zhibo Pang, Michele Zorzi, and Carlo Fischione. Low-latency networking: Where latency lurks and how to tame it.
Proc. IEEE
, 107 (2): 280–306, 2018.
[19]
Chong Li, Chih-Ping Li, Kianoush Hosseini, Soo Bum Lee, Jing Jiang, Wanshi Chen, Gavin Horn, Tingfang Ji, John E Smee, and Junyi Li. 5G-based systems design for tactile internet.
Proc. IEEE
, 107 (2): 307–324, 2018.
[20]
Qingkai Liang, Fanyu Que, and Eytan Modiano. Accelerated primal-dual policy optimization for safe reinforcement learning.
Proc. Neural Information Processing Systems (NIPS)
, 2019.
[21]
Jimmy J. Nielsen, Rongkuan Liu, and Petar Popovski. Ultra-reliable low latency communication (URLLC) using interface diversity.
IEEE Trans. Commun.
, 66 (3): 1322–1334, Mar. 2018.
[22]
David Ohmann, Ahmad Awada, Ingo Viering, Meryem Simsek, and Gerhard P. Fettweis. Modeling and analysis of intra-frequency multi-connectivity for high availability in 5G. In
Proc. IEEE Vehicular Technology Conference (VTC) Spring
, 2018.
[23]
Jihong Park, Sumudu Samarakoon, Hamid Shiri, Mohamed K Abdel-Aziz, Takayuki Nishio, Anis Elgabli, and Mehdi Bennis. Extreme URLLC: Vision, challenges, and key enablers.
arXiv preprint arXiv:2001.09683
, 2020.
[24]
Walid Saad, Mehdi Bennis, and Mingzhe Chen. A vision of 6G wireless systems: Applications, trends, technologies, and open research problems.
IEEE Netw.
, 34 (3): 134–142, 2019.
[25]
Mohit K Sharma, Alessio Zappone, Mohamad Assaad, Mérouane Debbah, and Spyridon Vassilaras. Distributed power control for large energy harvesting networks: A multi-agent deep reinforcement learning approach.
IEEE Trans. Cog. Commun. and Netw
., 5 (4): 1140–1154, 2019.
[26]
Changyang She, Yifan Duan, Guodong Zhao, Tony Q. S. Quek, Yonghui Li, and Branka Vucetic. Cross-layer design for mission-critical IoT in mobile edge computing systems.
IEEE Internet of Things J.
, 6 (6): 9360–9374, 2019.
[27]
Changyang She, Rui Dong, Zhouyou Gu, Zhanwei Hou, Yonghui Li, Wibowo Hardjawana, Chenyang Yang, Lingyang Song, and Branka Vucetic. Deep learning for ultra-reliable and low-latency communications in 6G networks.
IEEE Netw.
, 34 (5): 219–225, 2020.
[28]
Changyang She, Chengjian Sun, Zhouyou Gu, Yonghui Li, Chenyang Yang, H Vincent Poor, and Branka Vucetic. A tutorial on ultra-reliable and low-latency communications in 6G: Integrating domain knowledge into deep learning.
Proc. IEEE
, 109 (3): 204–246, Mar. 2021.
[29]
Mahyar Shirvanimoghaddam, Mischa Dohler, and Sarah J Johnson. Massive non-orthogonal multiple access for cellular IoT: Potentials and limitations.
IEEE Commun. Mag.
, 55 (9): 55–61, 2017.
[30]
Bikramjit Singh, Olav Tirkkonen, Zexian Li, and Mikko A Uusitalo. Contention-based access for ultra-reliable low latency uplink transmissions.
IEEE Wireless Commun. Lett.
, 7 (2): 182–185, 2017.
[31]
Eckehard Steinbach, Matti Strese, Mohamad Eid, Xun Liu, Amit Bhardwaj, Qian Liu, Mohammad Al-Ja’afreh, Toktam Mahmoodi, Rania Hassen, Abdulmotaleb El Saddik, et al. Haptic codecs for the tactile internet.
Proc. IEEE
, 107 (2): 447–470, 2018.
[32]
Chengjian Sun and Chenyang Yang. Learning to optimize with unsupervised learning: Training deep neural networks for URLLC. In
Proc. IEEE Personal, Indoor and Mobile Radio Communications (PIMRC)
, pages 1–7. IEEE, 2019.
[33]
Qianru Sun, Yaoyao Liu, Tat-Seng Chua, and Bernt Schiele. Meta-transfer learning for few-shot learning. In
Proc. IEEE Computer Vision and Pattern Recognition (CVPR)
, pages 403–412, 2019.
[34]
Yaping Sun, Zhiyong Chen, Meixia Tao, and Hui Liu. Communications, caching, and computing for mobile virtual reality: Modeling and tradeoff.
IEEE Trans.Commun
., 67 (11): 7573–7586, 2019.
[35]
Zonghan Wu, Shirui Pan, Fengwen Chen, Guodong Long, Chengqi Zhang, and S Yu Philip. A comprehensive survey on graph neural networks.
IEEE Trans. Neural Netw. learning Syst.
, 32 (1): 4–24, 2020.
[36]
Yunchou Xing and Theodore S Rappaport. Propagation measurement system and approach at 140 GHz-moving to 6G and above 100 GHz. In
Proc. IEEE Global Communications Conference (GLOBECOM)
, pages 1–6. IEEE, 2018.
[37]
Wei Yang, Rafael F Schaefer, and H Vincent Poor. Wiretap channels: Nonasymptotic fundamental limits.
IEEE Trans. Inf. Theory
, 65 (7): 4069–4093, 2019.
[38]
Alessio Zappone, Marco Di Renzo, and Mérouane Debbah. Wireless networks design in the era of deep learning: Model-based, AI-based, or both?
IEEE Trans. Commun
., 67 (10): 7331–7376, 2019.
2 Statistical Characterization of URLLC: Frequentist and Bayesian Approaches
Tobias Kallehauge, Pablo Ramirez-Espinosa, Anders E. Kalør and Petar Popovski*
Connectivity Section at the Department of Electronic Systems, Aalborg University, 9220, Aalborg, Fredrik Bajers Vej 7A, Denmark*Corresponding Author
2.1 Introduction
URLLC is one of the most significant novelties brought by 5G, aiming to support wireless connections with very stringent requirements in terms of latency and reliability, as specified, for example, in 3rd Generation Partnership Project (3GPP) [1]. Meeting these requirements is highly non-trivial, as the strict latency constraint limits the degrees of freedom that can be used to ensure reliable communication, and thus high reliability comes at a very high cost in terms of spectral efficiency. Low latency and high reliability are contradicting requirements [9,42], but both can be boosted by investing in more bandwidth [34], such that the problem of choosing the necessary resources (e.g., bandwidth) is of central importance. From a physical layer viewpoint, special attention must be paid to channel modeling, since the required amount of resources and the actual reliability are closely related to the wireless propagation characteristics [7,44]. In this context, a bad channel model may force the operators to use an overly conservative provisioning scheme, which will decimate the performance of the system. Of course, investing more in the estimation of channel statistics will ensure better models, hence, more suitable provisioning for the channel, but the excessive number of samples required to characterize the ultra-rare events that URLLC concerns, may leave very little to no time for data transmission [7]. This is particularly detrimental for non-stationary systems since the time required to estimate, say, the probability that the signal-to-noise ratio (SNR) is below a certain threshold with high confidence may exceed the time that channel statistics remain constant.
To tackle the difficulties of assuring and ensuring that the strict requirements for URLLC are met, this chapter applies a rigorous statistical approach to the problem. Hence, we define statistical measures that evaluate if a URLLC system fulfills its requirements (i.e., assurance) and then uses the insight gained from the statistical analysis to choose an appropriate transmission scheme to fulfill them (i.e., ensurance). Central to the statistical analysis of URLLC is the characterization of ultra-rare events which necessitates special statistical measures and careful analysis. For example, the average SNR is not very informative about the rare event when the SNR is outside its lower bound, therefore another statistic is required to characterize this event.
The amount of literature on statistical guarantees for URLLC and adjacent topics since the introduction of URLLC has been modest although some relevant articles have been published on the issue. One of the early works in the area is [7], which discusses the fundamental concepts of assuring communication reliability, defines statistical measures for this purpose, proposes different resource allocation schemes tailored to the statistical measures and highlights the issues of model mismatch and the number of samples required to estimate rare-event statistics. Other promising directions include the conditional value at risk (CVaR) as a statistical measure to characterize worst-case events [6,26] and extreme value theory (EVT) which offers a more direct way of characterizing rare events (see Section 2.3.5 for a brief introduction to EVT).
This chapter will introduce a fundamental approach to fulfilling service requirements for URLLC from a statistical perspective. We focus on reliability at the physical communication layer with narrowband transmission to limit the scope and simplify system models. As the chapter title suggests, we will explore both frequentist and Bayesian approaches. For this, consider the following motivating example. An industrial port services large container ships where cranes lift shipping containers onto autonomous guided vehicles known as shuttle carriers that place the containers in the port for temporary storage — see Figure 2.1 for an illustration. On the route between the cranes and storage locations, the shuttle carriers are in constant communication with a central control unit that schedules tasks for each shuttle, determines container locations and manages traffic in the port to optimize flow and avoid collisions. The wireless communication link between the control unit and each shuttle is therefore critical to performance and safety. In 3GPP, this type of communication can be classified under mobile robots with strict service requirements, such as a mean time between communication outages of the order of years [2, p. 15]. The quality of the wireless channel varies in both space and time, so channel state information (CSI) must continuously be updated to meet these requirements. A frequentist approach in this context would acquire CSI by estimating the channel every so often using pilot signals and then possibly assigning confidence intervals to the estimated CSI values. A Bayesian approach on the other hand would also rely on prior information to obtain a posterior belief about the likely CSI given the information at hand. The distinction between the frequentist and Bayesian approaches is somewhat subtle1, but for the purposes here, the Bayesian approach is especially attractive since it directly allows prior information to be incorporated into CSI acquisition. Prior information is readily available in this scenario, e.g., by using previous CSI estimates from other shuttle carriers in proximity to the location where a new transmission takes place. By relying on prior information, the CSI can be estimated with higher precision, or less time can be dedicated to pilot transmission. For URLLC, a Bayesian approach, therefore, has the potential to solve the difficulties of estimating rare-event statistics under strict latency constraints assuming that accurate prior information is available. Despite this, there has been a rather limited utilization of Bayesian statistics for URLLC in the recent literature, which is therefore explored here along with the frequentist approach.
Figure 2.1 Illustration of industrial port with automatic guided shuttle carriers controlled wirelessly.
The remainder of the chapter is organized as follows. Preliminaries for statistical channel modeling and reliability at the physical layer are given in Section 2.2. Section 2.3 introduces different statistical guarantees and shows how to fulfill them in a frequentist context. We then formally introduce the concept of Bayesian statistics and how it can be used in resource allocation for URLLC. Two illustrative examples are given. The first example (Section 2.4.2) shows the benefits of using Bayesian statistic in the ideal case where prior information about channel parameters are perfectly known. The second example (Section 2.5) shows how such prior information could be obtained in practice through statistical radio maps.
2.2 Preliminaries
2.2.1 Channel Models
The reliability of a communication system is inherently determined by the random wireless channel that alters the signal from the transmitter to the destination. Consequently, modeling the channel is central to describing the reliability of a wireless system. However, characterizing the channel in a wireless communication link is by no means a trivial task. The electromagnetic waves propagating from the transmitter to the receiver antenna are subjected to multiple physical phenomena, such as scattering and diffraction, giving rise to signal changes and fluctuations. Considering the propagation environment as a linear time-invariant (LTI) system — the reader is referred to standard textbooks, e.g., [20, 43, 46] for a more detailed description — these propagation effects are encapsulated in the baseband equivalent channel impulse response
where is the number of resolvable paths, is the channel coefficient associated with the -th path, is the Dirac delta function and is the corresponding delay. The ability of a system to resolve the different paths is related to the bandwidth and the delays . If , then only one path can be resolved with aggregated channel coefficient
We say therefore that the transmission is narrowband, and we will pay attention to this particular case in the rest of the chapter.
As stated before, the channel coefficient (or simply, the channel), captures the variations in the received signal, usually referred to as fading. Depending on the spatial scale of these variations, propagation effects are usually classified as [46]:
Large-scale fading, associated with the attenuation due to shadowing and distance (pathloss). It occurs at a scale of tens of meters.
Small-scale fading, due to constructive and destructive interference of the scattered waves arriving at the receiver, noticeable at a wavelength scale.
Pathloss is a consequence of the propagation of the electromagnetic waves throughout the medium, and is characterized by Friis’ widely-used transmission formula [20, Eq. (2.7)]
where and are the averaged received and transmitted powers, respectively, is the line-of-sight (LoS) distance, is the wavelength and is a generic term accounting for the antenna gains, polarization losses, etc.
Whilst pathloss can be seen as the averaged loss in power due to the distance, slow variations in space are produced by large obstacles like trees or buildings, giving rise to the so-called shadowing. This slow fluctuation, here represented by a random variable , is usually assumed to follow a lognormal distribution [43]. Due to the mathematical complexity of the lognormal formulation, other distributions have been used to characterize shadowing, such as Gamma [4,5] or inverse Gamma [36].
Finally, the small-scale fading (or simply, fading) represents the interference of multiple paths that cannot be resolved at the receiver. In its simplest form, it gives rise to a single complex coefficient representing the sum of multiple homogeneous planar waves [12,13]:
where is the amplitude of the -th planar wave and its phase. Since small differences in distance render noticeable changes in the phase of the incoming waves, the different ’s are usually assumed to be independent and, in most cases, uniformly distributed. Moreover, if , then can be approximated as Gaussian by the central limit theorem, i.e., , leading to the most common fading distribution:
Rayleigh
fading, in which , used to characterize non-line-of-sight (NLoS) environments with probability density function (PDF)
which is also known as the
exponential
distribution for with scale .
Rician
fading, in which represents some dominant path, used to characterize LoS environments with PDF
and parameters and where is the modified Bessel function of the first kind of order .
Aiming to generalize both Rayleigh and Rician distributions, several fading models have been proposed over the years [40, 49], although for the purpose of this chapter we stick to the simpler aforementioned models.
Combining pathloss, shadowing, and fading gives rise to the widely-used signal model
where is the transmitted complex symbol and the noise is white Gaussian with power spectral density , i.e., . In this simplified model, , since the average power is captured by .
2.2.2 Outage Probability
From the last section, and specifically from the random variable model in (2.7), we have seen that the transmitted symbol is directly affected by the channel , and then corrupted by noise. Naturally, more terms may be added to this model representing, e.g., interference coming from other users in the same system or even other systems operating at the same frequency band. However, for the sake of simplicity, we here stick to the simpler form in (2.7). It should be noted that the resulting received symbol is also a random variable, allowing a statistical analysis of the communication link. Along this line, one of the most extended metrics to characterize the performance of a system is the SNR, defined as the ratio between the received signal power and the noise power. Since the noise is assumed to be normalized, the SNR reads
with as the bandwidth.
A common assumption in wireless communications analyses is block-fading; that is, we consider the transmission of long blocks composed of several symbols , and for all of them the channel coefficient remains constant (albeit random) while the noise takes independent realizations. Under block-fading, the SNR can be rewritten as
The last equality in (2.9) is useful to characterize the system in a local area where shadowing and pathloss remains constant, and thus we can rewrite , with being the SNR in the case of no fading. Consequently, we observe that the SNR is also a random variable, whose distribution is a scaled version of that of (or directly in a local area).
Once the SNR is presented in terms of random variables, we can introduce an important metric in performance analysis: the outage probability. It is defined as the probability of the SNR to fall below a given threshold [40], i.e.,
where and denote, respectively, the PDF and cumulative distribution function (CDF) of . If is the minimum SNR required to successfully decode the received symbol, this outage event can be used to characterize the impact of the channel in the communication reliability.
