Dynamic beam hopping (DBH) is a crucial technology for adapting to the flexibility of service configurations in multi-beam satellite (MBS) communications. To solve the problem that the traditional beam-hopping method… Click to show full abstract
Dynamic beam hopping (DBH) is a crucial technology for adapting to the flexibility of service configurations in multi-beam satellite (MBS) communications. To solve the problem that the traditional beam-hopping method is difficult to adapt to the dynamic satellite environment, Deep Reinforcement Learning (DRL) was proposed. DRL can use the previous information to make decisions at the current moment and thus has low time complexity. However, because the generalization ability of the existing DRL methods cannot fully meet dynamic changes in multi-beam satellite communications scenarios, it is difficult to obtain the optimal decision. To address this issue, this letter explores a new framework in which DRL-Powered genetic algorithm (GA) approach. The proposed method uses DRL to assist the decision of DBH. Comparing the proposed method with DRL and traditional GA, simulation results show that the proposed method has a better performance in throughput and fairness when adapting to changes in the satellite environment.
               
Click one of the above tabs to view related content.