FedFm: Towards a Robust Federated Learning Approach For Fault Mitigation at the Edge Nodes

TL;DR

This paper proposes a robust federated learning method for edge devices that mitigates the impact of device failures and network issues, ensuring reliable model training in heterogeneous edge ecosystems.

Contribution

It introduces a strategy for selecting optimal edge devices and a mitigation approach to handle device failures, enhancing federated learning robustness at the edge.

Findings

Optimal device selection improves model performance.

Failure mitigation maintains model accuracy despite device dropouts.

Edge ecosystem behavior analyzed under various failure scenarios.

Abstract

Federated Learning deviates from the norm of "send data to model" to "send model to data". When used in an edge ecosystem, numerous heterogeneous edge devices collecting data through different means and connected through different network channels get involved in the training process. Failure of edge devices in such an ecosystem due to device fault or network issues is highly likely. In this paper, we first analyse the impact of the number of edge devices on an FL model and provide a strategy to select an optimal number of devices that would contribute to the model. We observe how the edge ecosystem behaves when the selected devices fail and provide a mitigation strategy to ensure a robust Federated Learning technique.