Machine Learning for Future Wireless Communications

FA-LONG LUO, Ph.D, Silicon Valley, California, USADr. Fa-Long Luo is an IEEE Fellow and an Affiliate Full Professor of Electrical & Computer Engineering Department at the University of Washington in Seattle. Having gained international high recognition, Dr. Luo has 36 years of research and industry experience in wireless communication, neural networks, signal processing, machine learning and broadcasting with real-time implementation, applications and standardization. Including his well-received book: Signal Processing for 5G: Algorithms and Implementations (2016, Wiley-IEEE), Dr. Luo has published 6 books and more than 100 technical papers in the related fields. Dr. Luo has also contributed 61 patents/inventions which have successfully resulted in a number of new or improved commercial products in mass production. He has served as the Chairman of IEEE Industry DSP Standing Committee and the Technical Board Member of Signal Processing Society. Dr. Luo was awarded the Fellowship by the Alexander von Humboldt Foundation of Germany.

• List of Contributors xvPreface xxiPart I Spectrum Intelligence and Adaptive Resource Management 11 Machine Learning for Spectrum Access and Sharing 3Kobi Cohen1.1 Introduction 31.2 Online Learning Algorithms for Opportunistic Spectrum Access 41.3 Learning Algorithms for Channel Allocation 91.4 Conclusions 19Acknowledgments 20Bibliography 202 Reinforcement Learning for Resource Allocation in Cognitive Radio Networks 27Andres Kwasinski, Wenbo Wang, and Fatemeh Shah Mohammadi2.1 Use of Q-Learning for Cross-layer Resource Allocation 292.2 Deep Q-Learning and Resource Allocation 332.3 Cooperative Learning and Resource Allocation 362.4 Conclusions 42Bibliography 433 Machine Learning for Spectrum Sharing in Millimeter-Wave Cellular Networks 45Hadi Ghauch, Hossein Shokri-Ghadikolaei, Gabor Fodor, Carlo Fischione, and Mikael Skoglund3.1 Background and Motivation 453.2 System Model and Problem Formulation 493.3 Hybrid Solution Approach 543.4 Conclusions and Discussions 59Appendix A Appendix for Chapter 3 61A.1 Overview of Reinforcement Learning 61Bibliography 614 Deep Learning–Based Coverage and Capacity Optimization 63Andrei Marinescu, Zhiyuan Jiang, Sheng Zhou, Luiz A. DaSilva, and Zhisheng Niu4.1 Introduction 634.2 Related Machine Learning Techniques for Autonomous Network Management 644.3 Data-Driven Base-Station Sleeping Operations by Deep Reinforcement Learning 674.4 Dynamic Frequency Reuse through a Multi-Agent Neural Network Approach 724.5 Conclusions 81Bibliography 825 Machine Learning for Optimal Resource Allocation 85Marius Pesavento and Florian Bahlke5.1 Introduction and Motivation 855.2 System Model 885.3 Resource Minimization Approaches 905.4 Numerical Results 965.5 Concluding Remarks 99Bibliography 1006 Machine Learning in Energy Efficiency Optimization 105Muhammad Ali Imran, Ana Flávia dos Reis, Glauber Brante, Paulo Valente Klaine, and Richard Demo Souza6.1 Self-Organizing Wireless Networks 1066.2 Traffic Prediction and Machine Learning 1106.3 Cognitive Radio and Machine Learning 1116.4 Future Trends and Challenges 1126.5 Conclusions 114Bibliography 1147 Deep Learning Based Traffic and Mobility Prediction 119Honggang Zhang, Yuxiu Hua, Chujie Wang, Rongpeng Li, and Zhifeng Zhao7.1 Introduction 1197.2 Related Work 1207.3 Mathematical Background 1227.4 ANN-Based Models for Traffic and Mobility Prediction 1247.5 Conclusion 133Bibliography 1348 Machine Learning for Resource-Efficient Data Transfer in Mobile Crowdsensing 137Benjamin Sliwa, Robert Falkenberg, and Christian Wietfeld8.1 Mobile Crowdsensing 1378.2 ML-Based Context-Aware Data Transmission 1408.3 Methodology for Real-World Performance Evaluation 1488.4 Results of the Real-World Performance Evaluation 1498.5 Conclusion 152Acknowledgments 154Bibliography 154Part II Transmission Intelligence and Adaptive Baseband Processing 1579 Machine Learning–Based Adaptive Modulation and Coding Design 159Lin Zhang and Zhiqiang Wu9.1 Introduction and Motivation 1599.2 SL-Assisted AMC 1629.3 RL-Assisted AMC 1729.4 Further Discussion and Conclusions 178Bibliography 17810 Machine Learning–Based Nonlinear MIMO Detector 181Song-Nam Hong and Seonho Kim10.1 Introduction 18110.2 A Multihop MIMO Channel Model 18210.3 Supervised-Learning-based MIMO Detector 18410.4 Low-Complexity SL (LCSL) Detector 18810.5 Numerical Results 19110.6 Conclusions 193Bibliography 19311 Adaptive Learning for Symbol Detection: A Reproducing Kernel Hilbert Space Approach 197Daniyal Amir Awan, Renato Luis Garrido Cavalcante, Masahario Yukawa, and Slawomir Stanczak11.1 Introduction 19711.2 Preliminaries 19811.3 System Model 20011.4 The Proposed Learning Algorithm 20311.5 Simulation 20711.6 Conclusion 208Appendix A Derivation of the Sparsification Metric and the Projections onto the Subspace Spanned by the Nonlinear Dictionary 210Bibliography 21112 Machine Learning for Joint Channel Equalization and Signal Detection 213Lin Zhang and Lie-Liang Yang12.1 Introduction 21312.2 Overview of Neural Network-Based Channel Equalization 21412.3 Principles of Equalization and Detection 21912.5 Performance of OFDM Systems With Neural Network-Based Equalization 23212.6 Conclusions and Discussion 236Bibliography 23713 Neural Networks for Signal Intelligence: Theory and Practice 243Jithin Jagannath, Nicholas Polosky, Anu Jagannath, Francesco Restuccia, and Tommaso Melodia13.1 Introduction 24313.2 Overview of Artificial Neural Networks 24413.3 Neural Networks for Signal Intelligence 24813.4 Neural Networks for Spectrum Sensing 25513.5 Open Problems 25913.6 Conclusion 260Bibliography 26014 Channel Coding with Deep Learning: An Overview 265Shugong Xu14.1 Overview of Channel Coding and Deep Learning 26514.2 DNNs for Channel Coding 26814.3 CNNs for Decoding 27714.4 RNNs for Decoding 27914.5 Conclusions 283Bibliography 28315 Deep Learning Techniques for Decoding Polar Codes 287Warren J. Gross, Nghia Doan, Elie Ngomseu Mambou, and Seyyed Ali Hashemi15.1 Motivation and Background 28715.2 Decoding of Polar Codes: An Overview 28915.3 DL-Based Decoding for Polar Codes 29215.4 Conclusions 299Bibliography 29916 Neural Network–Based Wireless Channel Prediction 303Wei Jiang, Hans Dieter Schotten, and Ji-ying Xiang16.1 Introduction 30316.2 Adaptive Transmission Systems 30516.3 The Impact of Outdated CSI 30716.4 Classical Channel Prediction 30916.5 NN-Based Prediction Schemes 31316.6 Summary 323Bibliography 323Part III Network Intelligence and Adaptive System Optimization 32717 Machine Learning for Digital Front-End: a Comprehensive Overview 329Pere L. Gilabert, David López-Bueno, Thi Quynh Anh Pham, and Gabriel Montoro17.1 Motivation and Background 32917.2 Overview of CFR and DPD 33117.3 Dimensionality Reduction and ML 34117.4 Nonlinear Neural Network Approaches 35017.5 Support Vector Regression Approaches 36817.6 Further Discussion and Conclusions 373Bibliography 37418 Neural Networks for Full-Duplex Radios: Self-Interference Cancellation 383Alexios Balatsoukas-Stimming18.1 Nonlinear Self-Interference Models 38418.2 Digital Self-Interference Cancellation 38618.3 Experimental Results 39118.4 Conclusions 393Bibliography 39519 Machine Learning for Context-Aware Cross-Layer Optimization 397Yang Yang, Zening Liu, Shuang Zhao, Ziyu Shao, and Kunlun Wang19.1 Introduction 39719.2 System Model 39919.3 Problem Formulation and Analytical Framework 40219.4 Predictive Multi-tier Operations Scheduling (PMOS) Algorithm 40919.5 A Multi-tier Cost Model for User Scheduling in Fog Computing Networks 41319.6 Conclusion 420Bibliography 42120 Physical-Layer Location Verification by Machine Learning 425Stefano Tomasin, Alessandro Brighente, Francesco Formaggio, and Gabriele Ruvoletto20.1 IRLV by Wireless Channel Features 42720.2 ML Classification for IRLV 42820.3 Learning Phase Convergence 43120.4 Experimental Results 43320.5 Conclusions 437Bibliography 43721 Deep Multi-Agent Reinforcement Learning for Cooperative Edge Caching 439M. Cenk Gursoy, Chen Zhong, and Senem Velipasalar21.1 Introduction 43921.2 System Model 44121.3 Problem Formulation 44321.4 Deep Actor-Critic Framework for Content Caching 44621.5 Application to the Multi-Cell Network 44821.6 Application to the Single-Cell Network with D2D Communications 45221.7 Conclusion 454Bibliography 455Index 459