Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Mobility-Aware Federated Learning: Multi-Armed Bandit Based Selection in Vehicular Network

Haoyu Tu, Lin Chen, Zuguang Li, Xiaopei Chen, Wen Wu

TL;DR

A mobility-aware vehicular federated learning (MAVFL) scheme in which vehicles drive through a road segment to perform FL, and a multi-armed bandit-based vehicle selection algorithm to minimize the utility function considering training loss and delay is proposed.

Abstract

In this paper, we study a vehicle selection problem for federated learning (FL) over vehicular networks. Specifically, we design a mobility-aware vehicular federated learning (MAVFL) scheme in which vehicles drive through a road segment to perform FL. Some vehicles may drive out of the segment which leads to unsuccessful training. In the proposed scheme, the real-time successful training participation ratio is utilized to implement vehicle selection. We conduct the convergence analysis to indicate the influence of vehicle mobility on training loss. Furthermore, we propose a multi-armed bandit-based vehicle selection algorithm to minimize the utility function considering training loss and delay. The simulation results show that compared with baselines, the proposed algorithm can achieve better training performance with approximately 28\% faster convergence.

Mobility-Aware Federated Learning: Multi-Armed Bandit Based Selection in Vehicular Network

TL;DR

A mobility-aware vehicular federated learning (MAVFL) scheme in which vehicles drive through a road segment to perform FL, and a multi-armed bandit-based vehicle selection algorithm to minimize the utility function considering training loss and delay is proposed.

Abstract

In this paper, we study a vehicle selection problem for federated learning (FL) over vehicular networks. Specifically, we design a mobility-aware vehicular federated learning (MAVFL) scheme in which vehicles drive through a road segment to perform FL. Some vehicles may drive out of the segment which leads to unsuccessful training. In the proposed scheme, the real-time successful training participation ratio is utilized to implement vehicle selection. We conduct the convergence analysis to indicate the influence of vehicle mobility on training loss. Furthermore, we propose a multi-armed bandit-based vehicle selection algorithm to minimize the utility function considering training loss and delay. The simulation results show that compared with baselines, the proposed algorithm can achieve better training performance with approximately 28\% faster convergence.

Paper Structure

This paper contains 19 sections, 2 theorems, 18 equations, 3 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Considered Scenario and System Model
  3. Considered Scenario
  4. Vehicular Federated Learning Scheme
  5. Vehicle Selection and Model Distribution
  6. Local Updating
  7. Model Uploading
  8. Model Aggregation
  9. Training Delay model
  10. Convergence Analysis
  11. Problem Formulation
  12. Proposed Algorithm
  13. Exploitation
  14. Exploration
  15. Simulation Results
  16. ...and 4 more sections

Key Result

Lemma 1

The divergence of the local model and virtual global model is bounded: $\mathbb{E}\left[\sum_{k=1}^K\left\|\overline{\mathbf{w}}_t-\mathbf{w}_k^t\right\|^2\right] \leq 4K\eta_t^2(E-~1)^2G^2$.

Figures (3)

  • Figure 1: System Model
  • Figure 2: Training performance of different vehicle selection algorithms for CIFAR-10
  • Figure 3: Training performance of different vehicle selection algorithms for GTSRB

Theorems & Definitions (3)

  • Lemma 1
  • Theorem 1
  • proof