Mobile Network Control with a World Model
2026-07-20 • Networking and Internet Architecture
Networking and Internet ArchitectureArtificial Intelligence
AI summaryⓘ
The authors created a smart system to manage mobile networks more efficiently by predicting how changes will affect the network in the future. They trained a 'world model' using old data to guess what will happen if the network settings are changed. Their system also knows how uncertain its predictions are and uses this to make better decisions about the network setup. They tested their method in simulations and with real data, showing it can save energy while keeping good service, doing better than some existing methods. The system can also adapt its goals without needing to be retrained.
world modelmobile networksnetwork controlenergy efficiencyquality of serviceuncertainty estimationreinforcement learningclosed-loop controlcounterfactual actionsthroughput constraints
Authors
Maxime Bouton, Ioanna Mitsioni, Simon Lindståhl, Jaeseong Jeong
Abstract
The increasing complexity of mobile networks necessitates intelligent and dynamic control strategies for efficient, energy-conserving management. We propose a world model-based approach for network control that enables adaptive configuration of crucial parameters. The world model is trained from historical data and predicts the impact of its actions on future network states. Our controller leverages the model's uncertainty estimate to robustly find optimal network configuration changes. Furthermore, the optimization objective can be changed dynamically without model retraining. We demonstrate the effectiveness of the approach in simulated closed-loop control of a mobile network energy-saving feature. Our results show improved performance in balancing energy savings with quality of service, compared to traditional methods and reinforcement learning approaches. Finally, we show the world model performance on real network data from, and evaluate counterfactual actions proposed by the controller under various throughput constraints.