Channel and Power Allocation for Uplink Multibeam LEO Satellite System with IoT Services

Satellite Internet of Things (IoT) networks have the potential to play a crucial role in remote regions lacking terrestrial network facilities. However, it is tough to allocate resources dynamically because of the rapid movement of low earth orbit (LEO) satellites. In this work, we propose a joint channel and power allocation algorithm to dynamically maximize the sum of transmission capacity in the uplink multibeam LEO satellite IoT system. Specifically, we formulate the problem as a mixed-integer programming problem and model it as a stochastic game. Furthermore, we adopt a multi-agent reinforcement learning (MARL) approach to solve the optimization described above problem. Simulation results prove the convergence of the proposed algorithm and show that it can perform better than other methods considering different numbers of subchannels and increase the uplink transmission capacity by as much as 13% considering the time-varying positions of LEO satellites over a period.