A Flexible Resource Allocation Algorithm in Dual-Timescale Beam Hopping Satellite Systems

作者：LI Shuang-yi, Haoyu Sun, Lei Jizhao, Ma Xiaobo, Hao Qin · 年份：2024 · DOI：10.1109/iceict61637.2024.10671170 · 研究领域：Satellite Communication Systems、Spacecraft Design and Technology、Space Satellite Systems and Control

How to make full use of limited satellite resources and the spatial flexibility of beams to manage these traffic demands is a critical challenge that must be urgently addressed in beam hopping (BH) satellite systems. In this paper, a dual-timescale BH resource allocation algorithm based on deep reinforcement learning (DRL) is proposed. On the large-timescale, a user clustering method that dynamically updates beam coverage cells is used to optimize resource allocation and reduce spatial interference. On the short-timescale, the proximal policy optimization (PPO) algorithm is used to coordinate BH, reduce inter-beam interference, maintain inter-cell delay fairness, and improve system throughput. Simulation results demonstrate that our proposed dual-timescale algorithm significantly outperforms the baseline scheme.