Federated Learning for Future Intelligent Wireless Networks

• Record Nr.: UNINA9911019509903321
• Autore: Sun Yao
• Pubbl/distr/stampa: Newark : , : John Wiley & Sons, Incorporated, , 2023; ©2024
• ISBN: 9781119913900; 111991390X; 9781119913924; 1119913926
• Edizione: [1st ed.]
• Descrizione fisica: 1 online resource (317 pages)
• Altri autori (Persone): YouChaoqun; FengGang; ZhangLei
• Disciplina: 621.384
• Soggetti: Wireless communication systems; Edge computing
• Lingua di pubblicazione: Inglese
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Nota di contenuto: Cover -- Title Page -- Copyright -- Contents -- About the Editors -- Preface -- Chapter 1 Federated Learning with Unreliable Transmission in Mobile Edge Computing Systems -- 1.1 System Model -- 1.1.1 Local Model Training -- 1.1.2 Update Result Feedback via the Wireless Channels -- 1.1.3 Global Model Averaging -- 1.2 Problem Formulation -- 1.2.1 Model Accuracy Loss -- 1.2.2 Communication Loss ϵt,mC -- 1.2.3 Sample Selection Loss ϵt,mS -- 1.2.4 Model Training Loss ϵt,mM -- 1.2.5 Problem Formulation -- 1.2.5.1 Objective Function -- 1.2.5.2 Energy Consumption Constraint -- 1.2.5.3 User Selection Constraint -- 1.2.5.4 Data Volume Constraint of Local Training Datasets -- 1.3 A Joint Optimization Algorithm -- 1.3.1 Compression Optimization -- 1.3.1.1 Optimization of At,m -- 1.3.1.2 Optimization of Dt,m -- 1.3.2 Joint Optimization of At,m and Dt,m -- 1.3.3 Optimization of Sample Selection -- 1.3.4 Optimization of User Selection -- 1.3.5 A Joint Optimization Algorithm -- 1.4 Simulation and Experiment Results -- Bibliography -- Chapter 2 Federated Learning with non‐IID data in Mobile Edge Computing Systems -- 2.1 System Model -- 2.1.1 Local ; Model Training -- 2.1.2 Federated Averaging -- 2.2 Performance Analysis and Averaging Design -- 2.2.1 The Analysis of Expected Weight Divergence -- 2.2.1.1 The Analysis of Expected Data Distribution Divergence E{ℒm} -- 2.2.1.2 An Upper Bound of δ(tK) -- 2.2.2 Rethinking the Settings of Federated Averaging Weights -- 2.3 Data Sharing Scheme -- 2.3.1 Data Sharing -- 2.3.2 Problem Formation -- 2.3.2.1 Objective Function -- 2.3.3 Optimization Constraints -- 2.3.4 A Joint Optimization Algorithm -- 2.3.4.1 CPU Cycle Frequency Optimization Subproblem -- 2.3.4.2 Transmit Power Allocation Subproblem -- 2.3.4.3 Sharing Dataset Optimization Subproblem -- 2.3.4.4 User Selection Optimization Subproblem.; 2.3.4.5 A Joint Optimization Algorithm -- 2.4 Simulation Results -- Bibliography -- Chapter 3 How Many Resources Are Needed to Support Wireless Edge Networks -- 3.1 Introduction -- 3.2 System Model -- 3.2.1 FL Model -- 3.2.1.1 Loss Function -- 3.2.1.2 Updating Model -- 3.2.2 Computing Resource Consumption Model -- 3.2.3 Communication Resource Consumption Model -- 3.2.3.1 Uplink -- 3.2.3.2 DownLink -- 3.3 Wireless Bandwidth and Computing Resources Consumed for Supporting FL‐Enabled Wireless Edge Networks -- 3.3.1 SINR Analysis (Uplink Direction) -- 3.3.1.1 Probability Density Function (PDF) of SINR -- 3.3.1.2 Transmission Success Probability of Local Models -- 3.3.2 SNR Analysis (Downlink Direction) -- 3.3.3 Wireless Bandwidth Needed for Transmitting Local/Global Models -- 3.3.4 Computing Resources Needed for Training Local Models -- 3.4 The Relationship between FL Performance and Consumed Resources -- 3.4.1 Local Model Accuracy -- 3.4.2 Global Model Accuracy -- 3.5 Discussions of Three Cases -- 3.5.1 Case 1: Sufficient Communication Resources and Computing Resources -- 3.5.2 Case 2: Sufficient Computing Resources and Insufficient Communication Resources -- 3.5.3 Case 3: Sufficient Communication Resources and Insufficient Computing Resources -- 3.6 Numerical Results and Discussion -- 3.6.1 Simulation Setting -- 3.6.2 Simulation Results -- 3.6.2.1 Verifying Analytical Results -- 3.6.2.2 Measuring the Performance of FL Settings -- 3.6.2.3 Examining the Trade‐Off between the Computing and Communication Resources under FL Framework -- 3.7 Conclusion -- 3.8 Proof of Corollary 3.2 -- 3.9 Proof of Corollary 3.3 -- References -- Chapter 4 Device Association Based on Federated Deep Reinforcement Learning for Radio Access Network Slicing -- 4.1 Introduction -- 4.2 System Model -- 4.2.1 Network Model -- 4.2.2 RAN Slicing -- 4.2.3 Service Requirements.; 4.2.4 Handoff Cost -- 4.3 Problem Formulation -- 4.3.1 Problem Statement -- 4.3.2 Markov Decision Process Modeling for Device Association -- 4.3.2.1 State -- 4.3.2.2 Action -- 4.3.2.3 Transition Probability -- 4.3.2.4 Reward -- 4.4 Hybrid Federated Deep Reinforcement Learning for Device Association -- 4.4.1 Framework of HDRL -- 4.4.1.1 DRL on Smart Devices -- 4.4.1.2 Horizontal Model Aggregation (hDRL) Level -- 4.4.1.3 Vertical Model Aggregation (vDRL) Level -- 4.4.2 Algorithm of Horizontal Model Aggregation -- 4.4.2.1 DDQN for Training Local Model -- 4.4.2.2 Update Models -- 4.4.3 Algorithm of Vertical Model Aggregation -- 4.4.4 HDRL Algorithm for Device Association -- 4.4.5 Convergence Analysis -- 4.5 Numerical Results -- 4.5.1 Simulation Settings -- 4.5.2 Numerical Results and Discussions -- 4.6 Conclusion -- Acknowledgment -- References -- Chapter 5 Deep Federated Learning Based on Knowledge Distillation and Differential Privacy -- 5.1 Introduction -- 5.2 Related Work -- 5.3 System Model -- 5.3.1 Security Model -- 5.4 The Implementation Details of the Proposed Strategy -- 5.4.1 Security Analysis -- 5.5 Performance Evaluation -- 5.5.1 Experimental Environment -- 5.5.2 ; Experimental Results -- 5.6 Conclusions -- Bibliography -- Chapter 6 Federated Learning‐Based Beam Management in Dense Millimeter Wave Communication Systems -- 6.1 Introduction -- 6.1.1 Prior Work -- 6.1.2 Contributions -- 6.2 System Model -- 6.3 Problem Formulation and Analysis -- 6.4 FL‐Based Beam Management in UDmmN -- 6.4.1 Markov Decision Process Model -- 6.4.2 FL‐Based Beam Management -- 6.4.2.1 Data Cleaning -- 6.4.2.2 Model Training -- 6.4.3 User Association in UDmmN -- 6.4.3.1 Downlink Measurements -- 6.4.3.2 User Perception -- 6.4.3.3 Multiple Association -- 6.5 Performance Evaluation -- 6.5.1 Comparison with BFS and EDB -- 6.5.2 Comparison with BMDL and BMCL.; 6.6 Conclusions -- Bibliography -- Chapter 7 Blockchain‐Empowered Federated Learning Approach for An Intelligent and Reliable D2D Caching Scheme -- 7.1 Introduction -- 7.2 Related Work -- 7.2.1 Learning‐Based D2D Caching Schemes -- 7.2.2 Blockchain‐Enabled D2D Caching Schemes -- 7.3 System Model -- 7.3.1 D2D Network -- 7.3.2 Content Caching Schemes -- 7.3.3 Transmission Latency -- 7.4 Problem Formulation and DRL‐Based Model Training -- 7.4.1 Problem Formulation -- 7.4.1.1 Action -- 7.4.1.2 State -- 7.4.1.3 Reward and Return -- 7.4.2 DRL‐Based Local Model Training -- 7.5 Privacy‐Preserved and Secure BDRFL Caching Scheme Design -- 7.5.1 Task and Requirements Publication -- 7.5.2 Appropriate UE Selection -- 7.5.3 Local Model Training -- 7.5.4 Area Model Update and Recording -- 7.5.5 Global Model Update and Recording -- 7.6 Consensus Mechanism and Federated Learning Model Update -- 7.6.1 Double‐Layer Blockchain Consensus Mechanism -- 7.6.2 FL Area Model Update in Subchain Layer -- 7.6.3 FL Global Model Update in Mainchain Layer -- 7.7 Simulation Results and Discussions -- 7.7.1 Simulation Setting -- 7.7.2 Numerical Results -- 7.8 Conclusion -- References -- Chapter 8 Heterogeneity‐Aware Dynamic Scheduling for Federated Edge Learning -- 8.1 Introduction -- 8.2 Related Works -- 8.3 System Model for FEEL -- 8.3.1 Flow of FEEL with Scheduling -- 8.3.2 Delay and Energy Model in FEEL -- 8.3.2.1 Delay Model -- 8.3.2.2 Energy Model -- 8.4 Heterogeneity‐Aware Dynamic Scheduling Problem Formulation -- 8.4.1 Convergence of FEEL with Scheduling -- 8.4.2 Scheduling Policy with Sequential Transmission -- 8.4.3 Problem Formulation -- 8.5 Dynamic Scheduling Algorithm Design and Analysis -- 8.5.1 Benchmark: R‐Round Lookahead Algorithm -- 8.5.2 DISCO: Dynamic Scheduling Algorithm -- 8.5.3 Algorithm Analysis, Complexity Reduction, and Implementation Discussion.; 8.5.3.1 Algorithm Analysis -- 8.5.3.2 Complexity Reduction -- 8.5.3.3 Implementation Discussion -- 8.6 Evaluation Results -- 8.6.1 Parameter Settings -- 8.6.2 Numerical Results -- 8.6.3 Experimental Results -- 8.7 Conclusions -- 8.A.1 Proof of Theorem 8.2 -- 8.A.2 Proof of Theorem 8.3 -- 8.A.2.1 Feasibility Proof -- 8.A.2.2 Optimality Proof -- References -- Chapter 9 Robust Federated Learning with Real‐World Noisy Data -- 9.1 Introduction -- 9.1.1 Work Prior to FedCorr -- 9.2 Related Work -- 9.2.1 Federated Methods -- 9.2.2 Local Intrinsic Dimension (LID) -- 9.2.2.1 Estimation of LID -- 9.3 FedCorr -- 9.3.1 Preliminaries -- 9.3.1.1 Data Partition -- 9.3.1.2 Noise Model -- 9.3.1.3 LID Scores for Local Models -- 9.3.2 Federated Preprocessing Stage -- 9.3.2.1 Client Iteration and Fraction Scheduling -- 9.3.2.2 Mixup and Local Proximal Regularization -- 9.3.2.3 Identification of Noisy Clients and Noisy Samples -- 9.3.3 Federated Fine‐Tuning Stage -- 9.3.4 Federated Usual Training Stage -- 9.4 Experiments -- 9.4.1 Experimental Setup -- 9.4.1.1 Baselines -- 9.4.1.2 Implementation Details -- 9.4.2 Comparison with State‐of‐the‐Art Methods -- 9.4.2.1 IID Settings -- 9.4.2.2 Non‐IID Settings -- 9.4.2.3 Combination with ; Other FL Methods -- 9.4.2.4 Comparison of Communication Efficiency -- 9.4.3 Ablation Study -- 9.5 Further Remarks -- Bibliography -- Chapter 10 Analog Over‐the‐Air Federated Learning: Design and Analysis -- 10.1 Introduction -- 10.2 System Model -- 10.3 Analog Over‐the‐Air Model Training -- 10.3.1 Salient Features -- 10.3.2 Heavy‐Tailed Interference -- 10.4 Convergence Analysis -- 10.4.1 Preliminaries -- 10.4.2 Convergence Rate of AirFL -- 10.4.3 Key Observations -- 10.5 Numerical Results -- 10.6 Conclusion -- Bibliography -- Chapter 11 Federated Edge Learning for Massive MIMO CSI Feedback -- 11.1 Introduction -- 11.2 System Model.; 11.2.1 Channel Model and Signal Model.
• Sommario/riassunto: This book, 'Federated Learning for Future Intelligent Wireless Networks,' explores the application of federated learning in mobile edge computing systems, emphasizing its potential to enhance communication efficiency and privacy in wireless networks. Edited by Yao Sun, Chaoqun You, Gang Feng, and Lei Zhang, the book presents various models and techniques for implementing federated learning in challenging environments, including unreliable transmissions, non-IID data, and noisy data conditions. Key topics include resource allocation, device association, and privacy-preserving methods such as differential privacy and knowledge distillation. Aimed at researchers and professionals in wireless communications and machine learning, this work provides insights into the integration of AI technologies into future 6G networks.