Towards Energy-Aware Federated Traffic Prediction for Cellular Networks

Cellular traffic prediction is a crucial activity for optimizing networks in fifth-generation (5G) and beyond. Accurate forecasting is imperative for designing intelligent networks, allocating resources and mitigating anomalies. Although machine learning (ML) is a promising approach to effectively predict network traffic, the centralization of massive data in a single data center raises issues regarding confidentiality, privacy and data transfer demands. To address these challenges, federated learning (FL) emerges as a promising approach and offers enhanced privacy through parallel distributed computations. However, the environmental impact of these methods is often overlooked, which throws into question their sustainability. In this paper, we address the trade-off between accuracy and energy consumption in FL by proposing a novel sustainability indicator that allows assessing the feasibility of ML models. In our FL-based scenario, we comprehensively evaluate state-of-the-art deep learning (DL) architectures using real-world measurements from base station (BS) sites in the area of Barcelona, Spain. Our findings indicate that larger ML models achieve marginally improved performance but have a significant environmental impact, making them impractical for real-world applications.