Decentralized Machine Learning Paradigm in Edge Computing: A Systematic Review of Challenges and Research Trajectories

Abstract

Federated Learning (FL) is a decentralized machine learning approach that enables model training across multiple devices while preserving data privacy. When applied to edge computing environments, FL provides a range of benefits, including reduced latency, bandwidth efficiency, and enhanced data privacy. This paper explores the current state of FL in edge computing, examines the unique challenges posed by these environments, and identifies future research directions to further develop this emerging field.