Federated Learning within Global Energy Budget over Heterogeneous Edge Accelerators

TL;DR

This paper introduces FedJoule, a novel federated learning client selection framework that optimizes model accuracy within a global energy budget, balancing energy efficiency and performance across heterogeneous edge devices.

Contribution

It formulates a bi-level ILP optimization problem for energy-aware client selection in federated learning, incorporating approximate Shapley values and energy-time prediction models.

Findings

FedJoule outperforms SOTA and baselines in accuracy and training time.

Achieves 15% higher accuracy and 48% faster training under diverse energy constraints.

Effectively balances energy consumption and model performance in heterogeneous FL environments.

Abstract

Federated Learning (FL) enables collaborative model training across distributed clients while preserving data privacy. However, optimizing both energy efficiency and model accuracy remains a challenge, given device and data heterogeneity. Further, sustainable AI through a global energy budget for FL has not been explored. We propose a novel optimization problem for client selection in FL that maximizes the model accuracy within an overall energy limit and reduces training time. We solve this with a unique bi-level ILP formulation that leverages approximate Shapley values and energy-time prediction models to efficiently solve this. Our FedJoule framework achieves superior training accuracies compared to SOTA and simple baselines for diverse energy budgets, non-IID distributions, and realistic experiment configurations, performing 15% and 48% better on accuracy and time, respectively. The results highlight the effectiveness of our method in achieving a viable trade-off between energy usage and performance in FL environments.