Dynamic Resource Allocation for Multi-beam Satellite Communication Systems

Dynamic resource allocation is a crucial concern for achieving flexible and efficient data transmission in satellite communication systems. This paper examines the dynamic resource allocation challenge within a satellite communication system with multiple beams. Considering the difference between user requirement and service providing capability, we establish the concept of system cost and cast the integrated problem of beam illumination, sub-channel, and power allocation as a minimization task for the average system cost. To address the defined issue, we first propose two beam scheduling schemes, then we frame the issue of sub-channel and power allocation as a Markov decision process. We introduce two algorithms, namely, a Double Deep Q Learning (DDQN)-based algorithm and an Improved Priority Experience Replay DDQN (PER-DDQN)-based algorithm, for the joint power and sub-channel allocation. The simulation outcomes illustrate the efficacy and superiority of the proposed algorithms.