Asynchronous Semi-Decentralized Federated Edge Learning for Heterogeneous Clients

TL;DR

This paper introduces an asynchronous semi-decentralized federated edge learning framework that improves training efficiency and convergence speed by addressing device heterogeneity through a novel aggregation scheme.

Contribution

It proposes a new asynchronous training algorithm for SD-FEEL with staleness-aware aggregation, enhancing convergence and performance in heterogeneous edge networks.

Findings

Faster convergence compared to synchronous methods

Improved learning performance in heterogeneous environments

Effective handling of staleness in model updates

Abstract

Federated edge learning (FEEL) has drawn much attention as a privacy-preserving distributed learning framework for mobile edge networks. In this work, we investigate a novel semi-decentralized FEEL (SD-FEEL) architecture where multiple edge servers collaborate to incorporate more data from edge devices in training. Despite the low training latency enabled by fast edge aggregation, the device heterogeneity in computational resources deteriorates the efficiency. This paper proposes an asynchronous training algorithm for SD-FEEL to overcome this issue, where edge servers can independently set deadlines for the associated client nodes and trigger the model aggregation. To deal with different levels of staleness, we design a staleness-aware aggregation scheme and analyze its convergence performance. Simulation results demonstrate the effectiveness of our proposed algorithm in achieving faster convergence and better learning performance.