Distributed Intelligence: A Verification for Multi-Agent DRL-Based Multibeam Satellite Resource Allocation
Xianglai Liao, Xin Hu, Zhijun Liu, Shijun Ma, Lexi Xu, Xiuhua Li, Weidong Wang, Fadhel M. Ghannouchi
Abstract
Centralized radio resource management method puts all of the computational burdens in an agent, which is unbearable with the increasing of data dimensionality. This letter focuses on how to schedule limited satellite-based radio resources efficiently to enhance transmission efficiency and extend broadband coverage with low complexity. We propose a cooperative multi-agent deep reinforcement learning (CMDRL) framework to achieve the radio resources management strategy. The bandwidth allocation problem is taken as an example to analyze the proposed novel method in simulation. The experimental results show that this approach is capable of enhancing transmission efficiency and be flexible to achieve the desired goal with low complexity.