Federated Learning with Communication Delay in Edge Networks

TL;DR

This paper introduces FedDelAvg, a federated learning algorithm that accounts for communication delays in edge networks, improving convergence speed by optimizing model weighting and learning rates.

Contribution

It proposes FedDelAvg, a novel algorithm that incorporates delay-aware weighting in federated averaging, with theoretical analysis and experimental validation.

Findings

Significant convergence speed improvements with delay-aware weighting.

Theoretical upper bound on global loss depending on weighting and learning rate.

Delay-aware scheme outperforms standard federated averaging in experiments.

Abstract

Federated learning has received significant attention as a potential solution for distributing machine learning (ML) model training through edge networks. This work addresses an important consideration of federated learning at the network edge: communication delays between the edge nodes and the aggregator. A technique called FedDelAvg (federated delayed averaging) is developed, which generalizes the standard federated averaging algorithm to incorporate a weighting between the current local model and the delayed global model received at each device during the synchronization step. Through theoretical analysis, an upper bound is derived on the global model loss achieved by FedDelAvg, which reveals a strong dependency of learning performance on the values of the weighting and learning rate. Experimental results on a popular ML task indicate significant improvements in terms of convergence speed when optimizing the weighting scheme to account for delays.