Load Balancing and Energy Efficiency in Cellular Networks with a Scenario-Aware Reinforcement Learning Agent

The vast proliferation of wireless data networks demands efficient resource allocation strategies to accommodate the increasing number of devices and the dynamic nature of cellular networks. As future networks face challenges like severe congestion and varying traffic demands, achieving satisfactory Quality of Service (QoS) and Quality of Experience (QoE) requires dynamic management. This paper introduces an improved self-optimization framework that adopts deep reinforcement learning (RL) to dynamically adjust key network parameters, such as handover settings, power levels, and MIMO technology. This approach significantly enhances network throughput by effectively balancing load distribution. The proposed framework explores the trade-off between system complexity and performance gains, demonstrating that an agent tailored to optimize a frequently recurring single scenario can outperform generalized agents under specific network conditions.