Efficient Cell Control for 5G Networks: A Multi-Agent Approach for Energy Savings, User Perception, and Network Optimization

This paper addresses the challenge of high energy consumption in 5G base stations, which has become a pressing issue with the large-scale deployment of 5G networks. To tackle this problem, we propose an efficient energy-saving cell control scheme based on multi-agent collaboration. The key contributions of this work include: 1) a state encoding and fusion mechanism that models the interactions between a cell and its neighboring cells, 2) a fusion reward function that adaptively combines energy-saving rates with user perception metrics to generate rewards, and 3) a denoising diffusion probabilistic model that quickly generates action combinations based on state. The proposed scheme aims to reduce energy consumption while considering the complex interactions between cells and user perception in real-world scenarios. Empirical evaluations in real-world scenario demonstrate the effectiveness of our method.