DQN-Based Scheduling Algorithm for Beam-Hopping LEO Satellite Communication Systems

This paper investigates a downlink scheduling method in beam-hopping (BH) low-Earth orbit (LEO) satellite communication systems, aiming to minimize power consumption while meeting traffic demands. In particular, we focus on optimizing time slot and transmit power allocation for BH scheduling, which remains challenging due to excessive power usage and high computational complexity on a satellite in conventional approaches. To address these limitations, deep learning-based algorithms have been extensively studied, but their achievable performance is limited. Thus, we propose a deep Q-network (DQN)-based time slot allocation algorithm to enhance the performance of existing deep learning methods. Additionally, we present an efficient power allocation method to avoid unnecessary power consumption. Simulation results demonstrate that the proposed scheme significantly reduces both power consumption and computational overhead.