Comprehensive overview of the principles, theories, and techniques needed to build end-to-end semantic communication systems, with case studies included.
In this rapidly evolving landscape, the integration of connected intelligence applications highlights the pressing need for networks to gain intelligence in a non-siloed and ad hoc manner. The traditional incremental approach to network design is no longer sufficient to support the diverse and dynamic requirements of these emerging applications. This necessitates a paradigm shift towards more intelligent and adaptive network architectures.
From theory to application, Foundations of Semantic Communication Networks describes and provides a comprehensive understanding of everything needed to build end-to-end semantic communication systems. This book covers various interdisciplinary topics such as the mathematical foundations of semantic communications, information theoretical perspectives, joint-source channel coding, semantic-aware resource management strategies, interoperability under heterogeneous semantic communication users, advanced artificial intelligence (AI) and machine reasoning techniques for enabling connected intelligent applications, secure and privacy-preserving semantic communication systems, and the coexistence and interoperability of semantic, goal-oriented, and legacy systems.
The book examines unique features of end-to-end networking with semantic communications, including instilling reasoning behaviors in communication nodes, the role of the semantic plane in information filtering, control of communication and computing resources, transmit and receive signaling schemes, and connected intelligence device control. It emphasizes the importance of data semantics and age of information metrics. The book also discusses the profound impact of semantic communications on the telecom industry, highlighting changes in network performance, resource management, traffic, as well as spectral and energy efficiency.
Furthermore, the book provides insights into the mathematical constructs and AI theories for formulating semantic information, such as topology and category theory. It explores real-world applications, case studies, and future research directions as wireless technologies transition to 6G and beyond.
Written by four recognized experts in the field with a wealth of expertise from academia, industry, and research institutions, Foundations of Semantic Communication Networks addresses sample topics, including:
- Proposing new formulations using rigorous mathematical frameworks such as category theory and algebraic topology
- Focusing on real-world scenarios, addressing multiple access and networking challenges through collaborative frameworks for multi-modal transmissions, examining multiple access schemes to enhance transmission efficiency, and ensuring coexistence with legacy systems
- Enabling efficient large-scale systems for 6G and beyond wireless systems through AI-native air interfaces and semantic-aware resource allocation strategies
- Utilizing causality and neuro-symbolic artificial intelligence for minimalistic transmissions, and achieving generalizability and transferability across contexts and data distributions to develop high-fidelity semantic communication systems
- Examining security vulnerabilities associated with deep neural networks in semantic communications, and proposing encrypted, privacy-preserving semantic communication systems (ESCS) as a solution
Foundations of Semantic Communication Networks is an excellent forward-thinking resource on the subject for readers with a strong background in the subject matter, including graduate-level students, academics, practitioners, and industry researchers.
Table of Contents
About the Editors xvii
List of Contributors xxi
Preface xxvii
Acknowledgment xxix
Acronyms xxxi
1 Introduction to Semantic Communications 1
Christina Chaccour, Christo Kurisummoottil Thomas, Walid Saad, and Merouane Debbah
1.1 From Information Streams to Streams of Understanding: The Rise of Semantic Communication Networks 1
1.1.1 How Does It Work? 3
1.1.2 Why Now? What Factors Contribute to Our Ongoing Reliance on Traditional Communications? 6
1.1.3 What Is NOT Semantic Communications? 7
1.1.3.1 Semantic Communications Is Not Data Compression 7
1.1.3.2 Semantic Communications Is Not Only an “AI for Wireless” Concept 9
1.1.3.3 Semantic Communications Is Not Only Goal-Oriented Communications 9
1.1.3.4 Semantic Communications Is Not Only Application-Aware Communications 10
1.2 Reimagining Future G Applications with Semantic Communications 11
1.2.1 Semantic Communication for Next-Generation XR 11
1.2.2 Digital Reality and Massive Twinning: Speaking the Same Language 13
1.2.3 Semantic Communication and Sustainable Networks: A Convergence for Efficiency 14
1.3 Structure and Path of the Book 16
Bibliography 17
Part I Fundamentals of Semantic Communications 19
2 Semantic Compression and Communication: Fundamentals and Methodologies 21
Emrecan Kutay and Aylin Yener
2.1 Introduction 21
2.1.1 Notation 23
2.2 Semantic Index Assignment 23
2.2.1 System Model 24
2.2.1.1 Minimization of Semantic Distortion 25
2.2.1.2 Graph Coloring Problem 26
2.2.1.3 Joint Graph Coloring and Index Assignment 27
2.3 The Rise of Machine Intelligence in Perception 29
2.4 Semantic Compression for Multimodal Sources 32
2.4.1 System Model 32
2.4.1.1 Semantic Quantization 33
2.4.1.2 Semantic Compression 34
2.4.1.3 Semantic Vector Quantized Autoencoder 36
2.4.2 Results 39
2.5 Conclusion 44
Bibliography 44
3 Toward a Theory of Semantic Information 49
Jean-Claude Belfiore and Daniel Bennequin
3.1 Introduction 49
3.2 Cohomological Nature of Information 50
3.3 Axioms for Information Spaces 52
3.4 Comparison with Other Propositions of Semantic Information Measures 54
3.5 Carnap and Bar-Hillel Languages 54
3.6 Shepard’s Experiment 57
Bibliography 59
4 Deep Joint Source and Channel Coding 61
Haotian Wu, Chenghong Bian, Yulin, Shao, and Deniz Gündüz
4.1 Introduction 61
4.2 DeepJSCC for MIMO Channels 62
4.2.1 System Model 62
4.2.1.1 Open-loop MIMO with CSIR 63
4.2.1.2 Closed-loop MIMO with CSIT 64
4.2.2 A DeepJSCC-MIMO Solution 65
4.2.2.1 Image-to-Sequence Transformation 66
4.2.2.2 Channel Heatmap Construction 66
4.2.2.3 ViT Encoding 67
4.2.2.4 ViT Decoding 67
4.2.2.5 Loss Function 68
4.2.3 Training and Evaluation 68
4.2.3.1 Open-loop MIMO System with CSIR 68
4.2.3.2 Closed-loop MIMO System with CSIT 71
4.3 DeepJSCC for Relay Channels 75
4.3.1 Cooperative Relay 75
4.3.1.1 System Model 75
4.3.1.2 DeepJSCC for Cooperative Relay 76
4.3.1.3 Numerical Experiments 79
4.3.2 Multihop Relay 81
4.3.2.1 System Model 82
4.3.2.2 Existing Methods 83
4.3.2.3 A Hybrid JSCC Solution 84
4.3.2.4 Numerical Experiments 88
4.4 DeepJSCC for Feedback Channels 91
4.4.1 System Model 91
4.4.2 A JSCCFormer-f Solution 93
4.4.2.1 ViT Encoder 93
4.4.2.2 ViT Decoder 95
4.4.3 Training and Evaluation 96
4.4.3.1 Transmission Performance 96
4.4.3.2 Impacts of Bandwidth Ratio and Block Number 97
4.4.3.3 Noisy Feedback Channel 100
4.4.3.4 Adaptability 100
4.4.3.5 High Resolution Dataset and Visualization 102
4.4.3.6 Variable Rate Transmission 102
4.5 Concluding Remarks 106
Bibliography 107
5 When Information Is a Function of Data - Some Information Theoretic Perspectives on Semantic Communications 111
Alexander Mariona, Homa Esfahanizadeh, Rafael Gregorio Lucas D’Oliveira, and Muriel Médard
5.1 The Central Limit Theorem 112
5.2 Quantitative Bounds 113
5.3 General Polynomials 115
5.4 Examples and Applications 119
5.4.1 The Computational Wiretap Channel 120
5.5 Further Generalizations 122
Bibliography 123
6 Interoperability and Coexistence of 6G Semantic, Goal-Oriented, and Legacy Systems 125
Emilio Calvanese Strinati, Mohamed Sana, Mattia Merluzzi, and Tomás Huttebraucker
6.1 Introduction 125
6.2 Interoperability Issue in Goal-oriented and Semantic Systems 126
6.2.1 Language in Multiuser Communication 128
6.2.2 A Measure of Semantic Mismatch 129
6.2.3 Semantic Channel Equalization 130
6.3 Coexistence of Semantic, Goal-Oriented, and Legacy Services in 6G 134
6.3.1 Goal-Oriented Resource Allocation 135
6.3.2 Goal-Driven Measures for Edge Inference 135
6.4 Conclusion 137
Acknowledgment 138
Bibliography 138
Part II Semantic Communications Networking 141
7 Optimization of Image Transmission in a Cooperative Semantic Communication Networks 143
Ye Hu and Mingzhe Chen
7.1 Introduction 143
7.1.1 Related Works 144
7.2 Representative Work 145
7.2.1 System Model 145
7.2.2 Semantic Information Extraction 146
7.2.3 Transmission Model 149
7.2.4 Image Semantic Similarity Model 150
7.2.5 Problem Formulation 151
7.3 Value-Decomposition-based Entropy-Maximized Multi-Agent RL Method 152
7.3.1 Components of VD-ERL Method 152
7.3.2 VD-ERL Algorithm for Semantic-Oriented Resource Allocation 155
7.3.3 Complexity and Convergence of the Introduced Algorithm 157
7.4 Simulation Results and Analysis 158
7.5 Conclusion 161
Bibliography 162
8 Multiple Access Design for Joint Semantic and Classical Communications 165
Xidong Mu and Yuanwei Liu
8.1 Introduction 165
8.2 Heterogeneous Semantic and Bit Multiuser Network 167
8.2.1 Multiple Access for the Heterogeneous Semantic and Bit Multiuser Network 169
8.2.2 Interplay Between Semantic Communications and NOMA 169
8.3 NOMA-Enabled Heterogeneous Semantic and Bit Multiuser Communications 170
8.3.1 Semantic Rate: A New Performance Metric 170
8.3.2 Semi-NOMA: A Unified Multiple Access Scheme 171
8.3.3 Fundamental Limit: Semantic-Versus-Bit Rate Region 173
8.4 Semantic Communications-Enhanced NOMA 175
8.4.1 Early-Late Rate Disparity Issue in NOMA 175
8.4.2 An Opportunistic Semantic and Bit Communication Approach for Noma 177
8.4.3 Numerical Case Studies 177
8.5 Concluding Remarks and Future Research 179
Bibliography 179
9 Contextual Reasoning-based Semantics-Native Communication 181
Hyowoon Seo, Yoon Huh, Heekang Song, Wan Choi, and Mehdi Bennis
9.1 Semantics-Native Communication 181
9.1.1 System Model 182
9.1.1.1 Information-Theoretic Model Description 183
9.1.1.2 Motivation from Triangle of Meaning Model 183
9.2 Contextual Reasoning for Semantics-Native Communication 184
9.2.1 Motivation from Referential Game 185
9.2.2 Single-Sided Contextual Reasoning 185
9.2.3 Double-Sided Contextual Reasoning 188
9.2.4 Multi-round Contextual Reasoning 189
9.3 Context Synchronization for Semantics-Native Communication 192
9.3.1 Bayesian Inverse Contextual Reasoning 193
9.3.2 Inverse Linearized Contextual Reasoning 194
9.3.2.1 Linearizing Contextual Reasoning 195
9.3.2.2 Invertible Linearized Contextual Reasoning 196
9.4 Information Bottleneck Contextual Reasoning 197
9.4.1 Information Bottleneck Method 197
9.4.2 Implementing Information Bottleneck with Contextual Reasoning 198
9.5 Conclusion 198
Bibliography 199
10 Interoperable Semantic Communication 201
Jinhyuk Choi, Hyelin Nam, Jihong Park, Seung-Woo Ko, Jinho Choi, Mehdi Bennis, and Seong-Lyun Kim
10.1 Pitfalls of Federated Learning for Semantic Alignment 201
10.2 Split Learning for Semantic Alignment 203
10.3 In-Context Learning for Semantic Alignment 207
10.4 Conclusion and Future Directions 211
Bibliography 212
Part III Machine Reasoning for Ai-native Semantic Communication Networks 215
11 Causal Reasoning Foundations of Semantic Communication Systems 217
Christo Kurisummoottil Thomas, Christina Chaccour, Walid Saad, and Merouane Debbah
11.1 Introduction 217
11.2 Causality Primer 219
11.3 Causal Semantic Communications 222
11.3.1 System Model 222
11.3.1.1 How to Pose the Proper Interventions and Counterfactuals via Queries? 224
11.3.2 Emergent Language Model 226
11.3.3 Semantic Information Measure 227
11.3.4 Signaling Game Model and Generalized Nash Equilibrium Problem 230
11.3.5 Characterization of the Generalized Local NE 232
11.3.6 Analysis of the Signaling Game Equilibria for Emergent Language 233
11.3.7 Average Semantic Representation Length for Classical and Emergent Language Based ESC 235
11.4 Numerical Results 236
11.4.1 Illustrative Example for NeSy AI’s Potential in Wireless Versus Classical AI Based Wireless 236
11.5 Conclusion 239
Bibliography 240
12 Reinforcement Learning-Based Unicast and Broadcast Wireless Semantic Communications 241
Zhilin Lu, Rongpeng Li, Ekram Hossain, Zhifeng Zhao, and Honggang Zhang
12.1 Introduction 241
12.2 System Model And Problem Formulation 245
12.2.1 Unicast Model 245
12.2.2 Broadcast Model 246
12.2.3 Problem Formulation 248
12.3 SemanticBC-SCAL Schemes with Alternating Learning Mechanism 250
12.3.1 The Markov Decision Process (MDP) Framework 250
12.3.2 Self-Critical Optimization Under Alternate Learning Mechanism 252
12.4 Performance Evaluation 258
12.4.1 Simulation Settings 258
12.4.2 Numerical Results And Analysis 260
12.4.2.1 Performance In Point-to-Point SC 260
12.4.2.2 Performance In Semantic BC 263
12.4.2.3 Convergence Analyses 264
12.4.2.4 Ablation Study 265
12.5 Conclusions 267
Bibliography 267
13 Imitation Learning-based Implicit Semantic-aware Communication Networks 273
Yiwei Liao, Zijian Sun, Yong Xiao, Guangming Shi, Yingyu Li, H. Vincent Poor, Walid Saad, Merouane Debbah, and Mehdi Bennis
13.1 Introduction 273
13.1.1 Framework of Implicit Semantic Communications 275
13.1.1.1 Representation of Semantics 275
13.1.1.2 Explicit Semantics 275
13.1.1.3 Implicit Semantics U V 275
13.1.1.4 Reasoning Mechanism Π 276
13.1.2 Knowledge Base 276
13.1.3 Reasoning Mechanism Modeling and Learning 277
13.2 System Model and Problem Formulation 277
13.2.1 System Model 277
13.3 iSAC Architecture 280
13.3.1 Source User Side 280
13.3.1.1 Semantic Encoding 280
13.3.1.2 Semantic Distance 283
13.3.2 Destination User Side 284
13.3.2.1 Semantic Interpreter 284
13.3.3 Algorithm and Theoretical Analysis 284
13.4 Extension to Collaborative Reasoning 286
13.4.1 Collaborative Reasoning Computing Network 286
13.4.2 Algorithm for Collaborative Reasoning 288
13.5 Conclusion 288
Bibliography 289
14 Semantic and Goal-Oriented Communication: A Data Valuation Perspective 291
Shashi Raj Pandey, Van Phuc Bui, and Petar Popovski
14.1 Introduction 291
14.2 Data Valuation Principles 292
14.2.1 Semantic Approaches In Satellite Communications For Earth Observation 293
14.2.2 Goal-Oriented Problems in FL 293
14.3 Semantic Communication For Earth Observation with LEO Satellites 294
14.3.1 Energy Model 296
14.3.2 Energy Efficient Data Downloading With Change Detection Constraint 296
14.3.3 Preprocessing and Semantic Encoding 297
14.3.3.1 Preprocessing 297
14.3.3.2 Cloud Removing 297
14.3.3.3 Change Scoring and Semantic Encoding 298
14.3.4 Numerical Results 298
14.4 Goal-Oriented Communications In FL 300
14.4.1 Contribution-Based Aggregation 301
14.4.2 Contribution-Based Participation 301
14.4.3 Performance Evaluations 302
14.5 Conclusion 303
Bibliography 304
Part IV Security of Semantic Networks 307
15 Securing Semantic Communications Against Adversarial Attacks 309
Yalin Evren Sagduyu, Aylin Yener, and Sennur Ulukus
15.1 Introduction 309
15.2 Semantic Communications 311
15.3 Multitask Learning For Semantic Communications 313
15.4 Adversarial Attacks 318
15.4.1 Untargeted Adversarial Attack 319
15.4.2 Targeted Adversarial Attack 320
15.4.3 Fast Gradient Sign Method (FGSM) Attack 320
15.4.4 Projected Gradient Descent (PGD) Attack 321
15.4.5 Basic Iterative Method (BIM) Attack 322
15.4.6 Momentum Iterative Method (MIM) Attack 322
15.4.7 DeepFool Attack 323
15.5 Adversarial Attacks On Semantic Communications 324
15.6 Defense Against Adversarial Attacks 330
15.7 Adversarial Training as Defense Against Adversarial Attacks on Semantic Communications 331
15.8 Future Research Directions 334
15.9 Conclusion 335
Bibliography 335
16 Encrypted Semantic Communications for Privacy Preserving 339
Zhiyong Chen, Meixia Tao, Zhongwei Wang, and Xinlai Luo
16.1 Introduction 339
16.2 Basics Of Semantic Communication Systems 340
16.3 Security Issues Of Semantic Communication 343
16.3.1 Security Risk Of Effective Transmission 343
16.3.2 Security Risk Of Privacy Protection 345
16.4 Encrypted Semantic Communications 346
16.4.1 Overall System Architecture 346
16.4.2 Physical-Layer Encryptor And Decryptor Structure In EnSC 348
16.4.3 Semantic Encoder And Decoder Structure In EnSC 349
16.5 Adversarial Encryption Training 349
16.5.1 Loss Functions 349
16.5.2 Training 351
16.5.3 Performance Evaluation 352
16.6 Conclusion 356
Bibliography 356
Appendix A 361
Index 367
