Federated Learning Over Wireless Edge Networks (Wireless Networks)

✍ Scribed by Wei Yang Bryan Lim, Jer Shyuan Ng, Zehui Xiong, Dusit Niyato, Chunyan Miao

• Publisher: Springer
• Year: 2022
• Tongue: English
• Leaves: 175
• Category: Library

✦ Synopsis

This book first presents a tutorial on Federated Learning (FL) and its role in enabling Edge Intelligence over wireless edge networks. This provides readers with a concise introduction to the challenges and state-of-the-art approaches towards implementing FL over the wireless edge network. Then, in consideration of resource heterogeneity at the network edge, the authors provide multifaceted solutions at the intersection of network economics, game theory, and machine learning towards improving the efficiency of resource allocation for FL over the wireless edge networks. A clear understanding of such issues and the presented theoretical studies will serve to guide practitioners and researchers in implementing resource-efficient FL systems and solving the open issues in FL respectively.

✦ Table of Contents

Preface
Contents
List of Figures
List of Tables
1 Federated Learning at Mobile Edge Networks: A Tutorial
1.1 Introduction
1.2 Background and Fundamentals of Federated Learning
1.2.1 Federated Learning
1.2.2 Statistical Challenges of FL
1.2.3 FL Protocols and Frameworks
1.2.4 Unique Characteristics and Issues of FL
1.3 Communication Cost
1.3.1 Edge and End Computation
1.3.2 Model Compression
1.3.3 Importance-Based Updating
1.4 Resource Allocation
1.4.1 Worker Selection
1.4.2 Joint Radio and Computation Resource Management
1.4.3 Adaptive Aggregation
1.4.4 Incentive Mechanism
1.5 Privacy and Security Issues
1.5.1 Privacy Issues
1.5.1.1 Information Exploiting Attacks in Machine Learning: A Brief Overview
1.5.1.2 Differential Privacy-Based Protection Solutions for FL Workers
1.5.1.3 Collaborative Training Solutions
1.5.1.4 Encryption-Based Solutions
1.5.2 Security Issues
1.5.2.1 Data Poisoning Attacks
1.5.2.2 Model Poisoning Attacks
1.5.2.3 Free-Riding Attacks
1.6 Applications of Federated Learning for Mobile Edge Computing
1.6.1 Cyberattack Detection
1.6.2 Edge Caching and Computation Offloading
1.6.3 Base Station Association
1.6.4 Vehicular Networks
1.7 Conclusion and Chapter Discussion
2 Multi-dimensional Contract Matching Design for Federated Learning in UAV Networks
2.1 Introduction
2.2 System Model and Problem Formulation
2.2.1 UAV Sensing Model
2.2.2 UAV Computation Model
2.2.3 UAV Transmission Model
2.2.4 UAV and Model Owner Utility Modeling
2.3 Multi-dimensional Contract Design
2.3.1 Contract Condition Analysis
2.3.2 Conversion into a Single-Dimensional Contract
2.3.3 Conditions for Contract Feasibility
2.3.4 Contract Optimality
2.4 UAV-Subregion Assignment
2.4.1 Matching Rules
2.4.2 Matching Implementation and Algorithm
2.5 Performance Evaluation
2.5.1 Contract Optimality
2.5.2 UAV-Subregion Preference Analysis
2.5.3 Matching-Based UAV-Subregion Assignment
2.6 Conclusion and Chapter Discussion
3 Joint Auction–Coalition Formation Framework for UAV-Assisted Communication-Efficient Federated Learning
3.1 Introduction
3.2 System Model
3.2.1 Worker Selection
3.2.2 UAV Energy Model
3.2.2.1 Flying Energy
3.2.2.2 Computational Energy
3.2.2.3 Communication Energy
3.2.2.4 Hovering Energy
3.2.2.5 Circuit Energy
3.3 Coalitions of UAVs
3.3.1 Coalition Game Formulation
3.3.2 Coalition Formation Algorithm
3.4 Auction Design
3.4.1 Buyers' Bids
3.4.2 Sellers' Problem
3.4.3 Analysis of the Auction
3.4.4 Complexity of the Joint Auction–Coalition Algorithm
3.5 Simulation Results and Analysis
3.5.1 Communication Efficiency in FL Network
3.5.2 Preference of Cells of Workers
3.5.3 Profit-Maximizing Behavior of UAVs
3.5.4 Allocation of UAVs to Cells of Workers
3.5.5 Comparison with Existing Schemes
3.6 Conclusion and Chapter Discussion
4 Evolutionary Edge Association and Auction in Hierarchical Federated Learning
4.1 Introduction
4.2 System Model and Problem Formulation
4.2.1 System Model
4.2.2 Lower-Level Evolutionary Game
4.2.3 Upper-Level Deep Learning Based Auction
4.3 Lower-Level Evolutionary Game
4.3.1 Evolutionary Game Formulation
4.3.2 Worker Utility and Replicator Dynamics
4.3.3 Existence, Uniqueness, and Stability of the Evolutionary Equilibrium
4.4 Deep Learning Based Auction for Valuation of Cluster Head
4.4.1 Auction Formulation
4.4.2 Deep Learning Based Auction for Valuation of Cluster Heads
4.4.3 Monotone Transform Functions
4.4.4 Allocation Rule
4.4.5 Conditional Payment Rule
4.4.6 Neural Network Training
4.5 Performance Evaluation
4.5.1 Lower-Level Evolutionary Game
4.5.1.1 Stability and Uniqueness of the Evolutionary Equilibrium
4.5.1.2 Evolutionary Equilibrium Under Varying Parameters and Conditions
4.5.2 Upper-Level Deep Learning Based Auction
4.5.2.1 Evaluation of the Deep Learning Based Auction
4.6 Conclusion and Chapter Discussion
5 Conclusion and Future Works
References
Index