Communication Dans Un Congrès Année : 2025

Energy Optimization for Multi-Band Cellular Networks: A Traffic Prediction-Based Strategy

Résumé

We propose in this work a Machine Learning-based carrier shutdown strategy for multi-carrier cellular networks, leveraging probabilistic traffic prediction. We introduce a metric that accounts for the sector-level total load and capacity. This metric empowers our carrier shutdown strategy with the ability to determine the sufficient number of active carriers required to manage the traffic load without service disruption. Moreover, the probabilistic nature of our predictions allows our solution to trade off between energy gains (turning off carriers) and traffic loss (service disruption). Evaluations with a recent live network dataset show that our strategy, even in its most conservative configuration, achieves an energy gain of over 9% without compromising service continuity, outperforming today's carrier shutdown solutions by nearly a factor of 10.

Fichier principal
Vignette du fichier
2025062350.pdf (582.9 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05040262 , version 1 (18-04-2025)

Licence

Identifiants

Citer

Anh-Khoa Dang, Hicham Khalifé, Stéphane Rovedakis, Stefano Secci, Mathias Sintorn. Energy Optimization for Multi-Band Cellular Networks: A Traffic Prediction-Based Strategy. 2025 IEEE International Conference on Machine Learning for Communication and Networking (ICMLCN), May 2025, Barcelona, Spain. pp.1-6, ⟨10.1109/ICMLCN64995.2025.11140086⟩. ⟨hal-05040262⟩
315 Consultations
122 Téléchargements

Altmetric

Partager

  • More