HEART: Achieving Timely Multi-Model Training for Vehicle-Edge-Cloud-Integrated Hierarchical Federated Learning

TL;DR

The paper tackles the challenge of timely multi-model training in vehicle-edge-cloud hierarchical federated learning (VEC-HFL) under high vehicular mobility. It proposes a hybrid synchronous-asynchronous aggregation rule to balance model quality and latency, and introduces HEART, a two-stage optimization framework that combines a hybrid PSO-GA for cross-task scheduling with a greedy per-vehicle task-ordering approach. Through simulations on real-world datasets with multiple tasks and diverse neural architectures, HEART demonstrates superior time efficiency and balanced task distribution compared to baseline methods. This work provides a scalable, practical framework for concurrent, multi-task learning in dynamic IoV environments, with potential impact on training efficiency and privacy-preserving distributed learning in vehicular networks.

Abstract

The rapid growth of AI-enabled Internet of Vehicles (IoV) calls for efficient machine learning (ML) solutions that can handle high vehicular mobility and decentralized data. This has motivated the emergence of Hierarchical Federated Learning over vehicle-edge-cloud architectures (VEC-HFL). Nevertheless, one aspect which is underexplored in the literature on VEC-HFL is that vehicles often need to execute multiple ML tasks simultaneously, where this multi-model training environment introduces crucial challenges. First, improper aggregation rules can lead to model obsolescence and prolonged training times. Second, vehicular mobility may result in inefficient data utilization by preventing the vehicles from returning their models to the network edge. Third, achieving a balanced resource allocation across diverse tasks becomes of paramount importance as it majorly affects the effectiveness of collaborative training. We take one of the first steps towards addressing these challenges via proposing a framework for multi-model training in dynamic VEC-HFL with the goal of minimizing global training latency while ensuring balanced training across various tasks-a problem that turns out to be NP-hard. To facilitate timely model training, we introduce a hybrid synchronous-asynchronous aggregation rule. Building on this, we present a novel method called Hybrid Evolutionary And gReedy allocaTion (HEART). The framework operates in two stages: first, it achieves balanced task scheduling through a hybrid heuristic approach that combines improved Particle Swarm Optimization (PSO) and Genetic Algorithms (GA); second, it employs a low-complexity greedy algorithm to determine the training priority of assigned tasks on vehicles. Experiments on real-world datasets demonstrate the superiority of HEART over existing methods.

Figures (6)

• Figure 1: A schematic of multi-model training over VEC-HFL.
• Figure 2: A schematic of multi-task scheduling and training VEC-HFL architecture of our interest.
• Figure 3: The average number of times that a task has been executed on vehicles for training across different methods. Left subplot: 4 tasks where C1= CIFAR task; C2= MNIST task; C3= Driver Yawning task; C4= 20 Newsgroups task; Right subplot: 9 tasks that combine different amounts of data and labels for tasks C1-C4.
• Figure 4: Test accuracy of the global models for different tasks: (a) CIFAR; (b) MNIST; (c) Driver Yawning; (d) 20 Newsgroups.
• Figure 5: The time that it takes for the global model of all tasks to achieve the fixed accuracy under different vehicle numbers, data samples per vehicles, and different methods: (a) 25 vehicles, and (b) 50 vehicles. Note that #1-#4 represent the number of 4 different data samples that the vehicle has for each task, which are 200, 400, 600, 800.