Temporally Consistent Graph Q-Networks for Intelligent Network Control

2026-06-11

AI Summary

This work addresses the challenge of dynamically optimizing antenna parameters in complex mobile networks to accommodate diverse services and evolving traffic demands. The authors propose TC-GQN, a multi-agent reinforcement learning algorithm that integrates a temporally consistent self-predictive graph neural network to learn task-agnostic representations of the entire network. By leveraging a global reward signal, TC-GQN coordinates local decisions at individual base stations, enabling cooperative control across sectors and carriers. Experimental results demonstrate that, compared to existing graph-based methods and rule-based controllers, TC-GQN significantly extends hardware sleep duration while maintaining stringent quality-of-service (QoS) requirements, thereby exhibiting superior adaptability and control efficiency.

Abstract

Mobile networks continue to grow in complexity and next generation networks are expected to support both increasing traffic loads and more diverse services. As network complexity rises, optimizing antenna parameters under dynamic or changing objectives becomes increasingly challenging. We propose a novel multi-agent reinforcement learning (MARL) algorithm for high-level control and orchestration of mobile networks. The Temporally Consistent Graph Q-Network (TC-GQN) algorithm learns a self-predicting representation of the whole network that is task-independent and aggregates information from all base-stations. A graph neural network is trained using a global reward function to assign coordinated local actions based on the learned encoding of the global network state. We evaluate the algorithm in a simulated environment to orchestrate an energy-saving feature across multiple sectors and multiple carriers under different quality of service (QoS) constraints. The proposed algorithm outperforms state-of-the-art graph-based baselines and a competitive rule-based controller by improving hardware sleep time while maintaining QoS. Moreover, the learned representation enables rapid adaptation to changing intents.

Problem

• mobile network optimization
• antenna parameter tuning
• quality of service (QoS)
• energy-saving orchestration
• dynamic network control

Innovation

• Temporally Consistent Graph Q-Network
• multi-agent reinforcement learning
• graph neural network
• network orchestration
• QoS-aware energy saving