A Comprehensive Survey on Energy Efficiency in Federated Learning: Strategies and Challenges

Abstract

Federated Learning (FL), a burgeoning approach in machine learning, facilitates collaborative model training across distributed devices while maintaining data privacy. Although gaining traction, FL faces a critical challenge in energy efficiency, which is vital for its scalability and practicality, especially in resource-limited settings like IoT networks and mobile devices. This paper provides a comprehensive survey of current methods and techniques aimed at enhancing energy efficiency in FL systems. We delve into various resource allocation techniques and algorithm optimization strategies. Additionally, we examine the role of cutting-edge technologies such as Blockchain and 6G networks, which play a crucial role in minimizing the energy footprint of FL systems. Our survey pinpoints the principal challenges and identifies prospective areas for future research, intending to spur further advancements in energy-efficient FL. We discuss the intricate interplay between energy efficiency, model accuracy, and system scalability in FL. Furthermore, the paper emphasizes the real-world implications of these strategies, highlighting their practical relevance in various technological applications.