HiFlash: Communication-Efficient Hierarchical Federated Learning with Adaptive Staleness Control and Heterogeneity-aware Client-Edge Association

TL;DR

HiFlash is a hierarchical federated learning system that reduces communication costs and improves efficiency by using adaptive staleness control and heterogeneity-aware client-edge association, validated through extensive experiments.

Contribution

The paper introduces HiFlash, an enhanced hierarchical federated learning framework that integrates deep reinforcement learning for adaptive staleness control and heterogeneity-aware client-edge association.

Findings

Significant reduction in communication overhead.

Improved model accuracy and system efficiency.

Validated performance gains through extensive experiments.

Abstract

Federated learning (FL) is a promising paradigm that enables collaboratively learning a shared model across massive clients while keeping the training data locally. However, for many existing FL systems, clients need to frequently exchange model parameters of large data size with the remote cloud server directly via wide-area networks (WAN), leading to significant communication overhead and long transmission time. To mitigate the communication bottleneck, we resort to the hierarchical federated learning paradigm of HiFL, which reaps the benefits of mobile edge computing and combines synchronous client-edge model aggregation and asynchronous edge-cloud model aggregation together to greatly reduce the traffic volumes of WAN transmissions. Specifically, we first analyze the convergence bound of HiFL theoretically and identify the key controllable factors for model performance improvement. We then advocate an enhanced design of HiFlash by innovatively integrating deep reinforcement learning based adaptive staleness control and heterogeneity-aware client-edge association strategy to boost the system efficiency and mitigate the staleness effect without compromising model accuracy. Extensive experiments corroborate the superior performance of HiFlash in model accuracy, communication reduction, and system efficiency.