Joint Optimization of Energy Efficiency and User Outage Using Multi-Agent Reinforcement Learning in Ultra-Dense Small Cell Networks

With the substantial increase in spatio-temporal mobile traffic, reducing the network-level energy consumption while satisfying various quality-of-service (QoS) requirements has become one of the most important challenges facing six-generation (6G) wireless networks. We herein propose a novel multi-agent distributed Q-learning based outage-aware cell breathing (MAQ-OCB) framework to optimize energy efficiency (EE) and user outage jointly. Through extensive simulations, we demonstrate that the proposed MAQ-OCB can achieve the EE-optimal solution obtained by the exhaustive search algorithm. In addition, MAQ-OCB significantly outperforms conventional algorithms such as no transmission-power-control (No TPC), On-Off, centralized Q-learning based outage-aware cell breathing (C-OCB), and random-action algorithms.