Multi-agent DQN and Geographical Clustering-Based Beam Hopping for Multi-beam LEO Satellite Systems

Low Earth Orbit (LEO) satellites offer global coverage and are crucial for satellite-terrestrial integrated networks. However, their high mobility and time-varying traffic demands pose challenges for efficient resource utilization to meet the unevenly distributed traffic demands. This paper proposes a multi-satellite cooperative beam-hopping (BH) strategy based on satellite clustering to address these challenges. Additionally, a multi-agent Deep Reinforcement Learning (DRL) algorithm is employed to optimize resource scheduling and reduce computational complexity. Simulation results demonstrate that the proposed scheme effectively enhances system capacity and achieves load balancing.