Towards Benchmarking Power-Performance Characteristics of Federated Learning Clients

TL;DR

This paper investigates the real-world power-performance characteristics of federated learning clients, emphasizing the importance of accurate energy estimates for improving energy-efficient scheduling in diverse device conditions.

Contribution

It introduces a first analysis of how device power modes and concurrent workloads affect energy consumption in federated learning, highlighting the need for better power-performance modeling.

Findings

Real-world device power modes significantly impact energy consumption.

Current models oversimplify energy use by assuming constant per-sample energy.

Improved power-performance estimates can enhance energy-efficient FL scheduling.

Abstract

Federated Learning (FL) is a decentralized machine learning approach where local models are trained on distributed clients, allowing privacy-preserving collaboration by sharing model updates instead of raw data. However, the added communication overhead and increased training time caused by heterogenous data distributions results in higher energy consumption and carbon emissions for achieving similar model performance than traditional machine learning. At the same time, efficient usage of available energy is an important requirement for battery constrained devices. Because of this, many different approaches on energy-efficient and carbon-efficient FL scheduling and client selection have been published in recent years. However, most of this research oversimplifies power performance characteristics of clients by assuming that they always require the same amount of energy per processed sample throughout training. This overlooks real-world effects arising from operating devices under different power modes or the side effects of running other workloads in parallel. In this work, we take a first look on the impact of such factors and discuss how better power-performance estimates can improve energy-efficient and carbon-efficient FL scheduling.