Joint Channel and Learning Quality-Aware Client Scheduling in Wireless Federated Learning

Abstract

Federated learning (FL) has emerged as a promising distributed learning paradigm for Internet of Things (IoT) applications, enabling participating clients to train local models using their private data and periodically upload model updates for global aggregation. By keeping raw data at the client side, FL enhances data privacy while reducing the need for centralized data collection. However, the limited and heterogeneous nature of wireless communication resources poses significant challenges to efficient client participation and model training. In this letter, we propose a client scheduling strategy for wireless federated learning that jointly considers channel quality and learning quality to identify clients that can contribute more effectively to the global training process. Unlike conventional scheduling approaches that primarily prioritize clients according to their wireless channel conditions, the proposed strategy incorporates both communication and learning-related characteristics in the client selection process. We evaluate the proposed approach against conventional channel-quality-based scheduling methods. Experimental results demonstrate that the proposed strategy substantially improves the FL training performance, achieving higher model accuracy and faster convergence while effectively utilizing the available wireless resources.