面向高铁无线网络覆盖与容量优化的多agent模糊强化学习算法

为了提升高铁沿线LTE无线网络服务质量,提供最理想的覆盖与容量性能,在传统单agent学习算法的基础上,提出了通过多agent联合调整相邻eNodeB的天线下倾角从而实现覆盖与容量优化的模糊强化学习算法。并在LTE网络下的高速场景中进行仿真,仿真结果表明多agent学习算法与传统学习算法相比在高速环境下达到全局最优解的速率更快,特别是在应对环境突变的情况时恢复到最优解的速率有所提升。

Multi-Agent Fuzzy Reinforcement Learning Algorithm for Wireless Network Coverage and Capacity Optimization in High-Speed Railway

In order to enhance the service quality of LTE wireless network along high-speed railway and provide the optimal coverage and capacity performance, based on the traditional single-agent learning algorithm, a fuzzy reinforcement learning algorithm that jointly adjusts the neighboring eNodeB’s downtilt angle for network coverage and capacity optimization by means of multi-agent is proposed. In addition,simulation of LTE network in high-speed scenario indicates that the multi-agent learning algorithm could fairly improve the convergence rate of global optimal configurations in high-speed environment as compared with traditional reinforcement learning algorithm, in particular the restoration rate when dealing with the sudden environmental change.