p14 / Journal
Federated learning in mobile networks: A comprehensive case study on traffic forecasting
IEEE Transactions on Sustainable Computing, 10(3), 576-587 · 2025 · Online 2024
In plain language
A detailed real-world study of how federated learning behaves when forecasting traffic across mobile-network sites.
Research context
It gives network operators evidence about design choices that matter when moving federated learning beyond a simplified experiment.
Approach
Barcelona base-station data, aggregation comparisons, outlier handling, personalization, client contribution analysis, external data, and sustainability evaluation.
Findings
The study finds that federated forecasting can accommodate heterogeneous sites while providing useful predictions and favorable sustainability characteristics in the evaluated settings.
Summary based on abstract. Journal issue: May-June 2025. First publication online: November 2024.