Reinforcement Learning-Based Adaptive Power and Resource Allocation in Wireless Communication Networks

Abstract

The densification of wireless networks and growing real-time service demands have intensified the need for intelligent, energy-efficient resource allocation. Traditional static and centralized methods fall short in adapting to the dynamic and interference-prone nature of 5G and emerging 6G environments. This study proposes a decentralized reinforcement learning (RL)-based framework for joint power and spectrum allocation in ultra-dense wireless systems. Each base station acts as an autonomous agent, making real-time decisions based on local traffic and interference conditions. Simulated using a custom Python-based environment with 50 base stations and 500 users, the RL approach is benchmarked against static and optimization-based methods. Results show the RL model achieves up to 91% energy efficiency, 94% spectrum utilization, and only 5% QoS degradation, outperforming baseline models. This work demonstrates the viability of RL for distributed resource management and provides a reproducible simulation toolkit to support further research in AI-driven wireless communication systems.