Mobility-Aware Decentralized Federated Learning with Joint Optimization of Local Iteration and Leader Selection for Vehicular Networks

2025-03-09

AI Summary

To address slow convergence and poor communication robustness in decentralized federated learning (DFL) over high-mobility, resource-constrained vehicular networks, this paper proposes Mobility-Aware Decentralized Federated Learning (MDFL). We formulate the joint optimization of local iteration counts and dynamic leader selection as a decentralized partially observable Markov decision process (Dec-POMDP) for the first time, and solve it via multi-agent proximal policy optimization (MAPPO) to enable adaptive, cooperative training. MDFL explicitly accounts for network dynamics, balancing model convergence speed and communication stability under time-varying topologies. Experiments demonstrate that MDFL achieves 1.8× faster convergence and improves test accuracy by 4.2 percentage points over baseline DFL methods, while significantly reducing communication overhead and leader-switching frequency.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Multiagent Systems: Multiagent Learning
• Planning, Routing, and Scheduling: Planning with Markov Models (MDPs, POMDPs)

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Federated learning (FL) emerges as a promising approach to empower vehicular networks, composed by intelligent connected vehicles equipped with advanced sensing, computing, and communication capabilities. While previous studies have explored the application of FL in vehicular networks, they have largely overlooked the intricate challenges arising from the mobility of vehicles and resource constraints.In this paper, we propose a framework of mobility-aware decentralized federated learning (MDFL) for vehicular networks. In this framework, nearby vehicles train an FL model collaboratively, yet in a decentralized manner. We formulate a local iteration and leader selection joint optimization problem (LSOP) to improve the training efficiency of MDFL. For problem solving, we first reformulate LSOP as a decentralized partially observable Markov decision process (Dec-POMDP), and then develop an effective optimization algorithm based on multi-agent proximal policy optimization (MAPPO) to solve Dec-POMDP. Finally, we verify the performance of the proposed algorithm by comparing it with other algorithms.

Problem

• Optimizes local iteration and leader selection in vehicular networks.
• Addresses mobility and resource constraints in decentralized federated learning.
• Proposes a mobility-aware decentralized FL framework for intelligent vehicles.

Innovation

• Decentralized Federated Learning for vehicular networks
• Joint optimization of local iteration and leader selection
• Multi-agent proximal policy optimization algorithm