Dynamic Beam Hopping of Double LEO Multi-beam Satellite based on Determinant Point Process

Low Earth Orbit (LEO) satellite communication is a promising system for expanding the coverage of communication networks. However, its application is limited due to its high relative ground speed and limited power. Moreover, the non-uniform geographical distribution and time-varying characteristics of ground service put forward higher requirements for adopting beam hopping technology in LEO satellite system. Different from Geostationary Earth Orbit (GEO) satellites, LEO satellites have higher real-time requirements. Therefore, heuristic algorithms such as genetic algorithm cannot achieve real-time scheduling due to their slow convergence. Recently, a reinforcement learning based method is proposed to implement the real-time beam hopping, in which the action space is exponentially increase with the number of beams especially when the serving spaces are overlapped. Therefore, in this paper, the determinant point process (DPP) algorithm is used to solve the LEO dual-satellite dynamic beam hopping problem by using the exclusion provided by the difference between inter-cell interference and inter-cell demand traffic delay. The simulation results show that the DPP algorithm can well balance overall throughput and inter-cell delay fairness. Additionally, when different traffic service is required, DPP algorithm can achieve superior results without retraining process.