Federated Learning for Wireless Networks

Preface
Acknowledgement
Contents
Part I Fundamentals and Background
1 Introduction
	1.1 Machine Learning for Wireless Networks
		1.1.1 Current Challenges
		1.1.2 Distributed Machine Learning
		1.1.3 Federated Learning Briefing
	1.2 Organization of the Book
2 Fundamentals of Federated Learning
	2.1 Introduction and History
	2.2 Federated Learning Key Challenges
		2.2.1 Statistical Heterogeneity
		2.2.2 System Heterogeneity
	2.3 Key Design Aspects
		2.3.1 Resource Allocation
		2.3.2 Incentive Mechanism
		2.3.3 Security and Privacy
	2.4 Federated Learning Algorithms
		2.4.1 FedAvg
		2.4.2 FedProx
		2.4.3 q-Federated Learning
		2.4.4 Federated Multi-Task Learning
	2.5 Summary
Part II Wireless Federated Learning: Design and Analysis
3 Resource Optimization for Wireless Federated Learning
	3.1 Introduction
	3.2 Wireless Federated Learning: Convergence Analysis and Resource Allocation
		3.2.1 System Model
			Federated Learning Over Wireless Networks
			Computation Model
			Communication Model
		3.2.2 Problem Formulation
		3.2.3 Decomposition-Based Solution
			SUB1 Solution
			SUB2 Solution
			SUB3 Solution
			FEDL Solution
		3.2.4 Numerical Results
			Impact of UE Heterogeneity
			Pareto Optimal Trade-off
			Impact of η
	3.3 Wireless Federated Learning: Resource Allocation and Transmit Power Allocation
		3.3.1 Motivation
		3.3.2 System Model
			Machine Learning Model
			Transmission Model
			Packet Error Rates
			Energy Consumption Model
			Problem Formulation
		3.3.3 Convergence Analysis
		3.3.4 Optimization of RB Allocation and Transmit Power for FL Training Loss Minimization
			Optimal Transmit Power
			Optimal Uplink Resource Block Allocation
		3.3.5 Numerical Results
	3.4 Collaborative Federated Learning
		3.4.1 Motivation
		3.4.2 Preliminaries and Overview
			Original Federated Learning
			Collaborative Federated Learning
		3.4.3 Communication Techniques for Collaborative Federated Learning
			Network Formation
			Device Scheduling
			Coding
	3.5 Summary
4 Incentive Mechanisms for Federated Learning
	4.1 Introduction
	4.2 Game Theory-Enabled Incentive Mechanism
		4.2.1 System Model
			Federated Learning Background
			Cost Model
		4.2.2 Stackelberg Game-Based Solution
			Incentive Mechanism: A Two-Stage Stackelberg Game Approach
			Stackelberg Equilibrium: Algorithm and Solution Approach
		4.2.3 Simulations
	4.3 Auction Theory-Enabled Incentive Mechanism
		4.3.1 System Model
			Preliminary of Federated Learning
			Computation and Communication Models for Federated Learning
			Auction Model
			Deciding Mobile Users's Bid
			Iterative Algorithm
			Optimization of Uplink Transmission Power
			Optimization of CPU Cycle Frequency and Number of Antennas
			Convergence Analysis
			Complexity Analysis
		4.3.2 Auction Mechanism Between BS and Mobile Users
			Problem Formulation
			Approximation Algorithm Design
			Approximation Ratio Analysis
			Payment
			Properties
		4.3.3 Simulations
	4.4 Summary
	Appendix
		A.1 KKT Solution
5 Security and Privacy
	5.1 Introduction
	5.2 Functional Encryption Enabled Federated Learning
		5.2.1 Federated Learning
		5.2.2 All or Nothing Transform (AONT)
		5.2.3 Multi-Input Functional Encryption for Inner Product
		5.2.4 Threat Model
	5.3 Secure Aggregation for Wireless Federated Learning
		5.3.1 Participant Pre-processing Mode Updates
		5.3.2 Secure Aggregation at Aggregator
	5.4 Security Analysis
		5.4.1 Security for Encryption
		5.4.2 Privacy for Participant
	5.5 Implementation and Evaluation
		5.5.1 Implementation
		5.5.2 Evaluation
	5.6 Summary
6 Unsupervised Federated Learning
	6.1 Introduction
	6.2 Problem Formulation
	6.3 Dual Averaging Algorithm
		6.3.1 Algorithm Description
		6.3.2 Data Labeling Step
		6.3.3 DA-Based Centroid Computation Step
		6.3.4 Weight Computation via Bin Method
		6.3.5 Weight Computation via Self-Organizing Maps
	6.4 Simulations
	6.5 Summary
Part III Federated Learning Applications in Wireless Networks
7 Wireless Virtual Reality
	7.1 Motivation
	7.2 Existing Works
	7.3 Representative Work
		7.3.1 System Model
			Transmission Model
			Break in Presence Model
			Problem Formulation
		7.3.2 Federated Echo State Learning for Predictions of the Users' Location and Orientation
			Components of Federated ESN Learning Algorithm
			ESN Based Federated Learning Algorithm for Users' Location and Orientation Predictions
		7.3.3 Memory Capacity Analysis
		7.3.4 User Association for VR Users
		7.3.5 Simulation Results and Analysis
	7.4 Summary
8 Vehicular Networks and Autonomous Driving Cars
	8.1 Introduction and State of Art
	8.2 Vehicular Networks
		8.2.1 Selective Model Aggregation
		8.2.2 System Model
			Image Quality
			Computation Capability
			Utility Function and Type of Vehicular Client
			Utility Function of Central Server
			Global Loss Decay
			End-to-end Latency
		8.2.3 Contract Formulation
		8.2.4 Problem Relaxation and Transformation
			Relaxing Constraint
			Simplifying Complicated Constraint
		8.2.5 Solution to Optimal Contracts
		8.2.6 Numerical Results
			Simulation Settings
	8.3 Autonomous Driving Cars
		8.3.1 System Model and Problem Formulation
			Federated Learning Model
			Communication Model
			Problem Formulation
		8.3.2 Joint Association and Resource Allocation Algorithm for DFL
			Matching Game-Based Resource Allocation
			Autonomous Car-RSU Association Algorithm
		8.3.3 Numerical Results
	8.4 Summary
9 Smart Industries and Intelligent Reflecting Surfaces
	9.1 Smart Industry
		9.1.1 System Model and Problem Formulation
		9.1.2 Block Successive Upper-Bound Minimization-Based Solution
		9.1.3 Simulations
	9.2 Intelligent Reflecting Surfaces
		9.2.1 Introduction
		9.2.2 Problem Formulation
		9.2.3 FL Assisted Optimal Beam Reflection
		9.2.4 Simulation
	9.3 Summary
References