Deep Reinforcement Learning-based Satellite Handover Scheme for Satellite Communications

• Published in: International Conference on Wireless Communications and Signal Processing pp. 1 - 6
• Main Authors: Wang, Jie, Mu, Weiqing, Liu, Yanan, Guo, Lantu, Zhang, Shaofeng, Gui, Guan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 20.10.2021
• Subjects: Benchmark testing; deep reinforcement learning; Handover; handover scheme; Low earth orbit (LEO) satellites; Low earth orbit satellites; Reinforcement learning; Satellites; Simulation; Wireless communication
• ISSN: 2472-7628
• DOI: 10.1109/WCSP52459.2021.9613411

Low Earth orbit (LEO) satellites have attracted increasing attention in recent years due to their wide coverage area and flexible connectivity. However, the high speed mobility of LEO satellites imposes huge challenges for satellite handover in LEO satellites communication networks. Firstly, frequent handovers cause unacceptable signaling overhead. Secondly, multiple handover factors including the remaining visible time, the received signal strength, the shortest distance and the load balance of the satellite are crucial for handover decision and should be taken into account. However, the conventional handover strategies only consider single handover criterion or weight several handover factors simply, which cannot guarantee the success rate of handover, the communication quality and the maximum utilization of the resource of the satellites simultaneously. To resolve the above problems, this paper propose a novel deep reinforcement learning (DRL)-based handover scheme by considering multiple handover factors simultaneously. Simulation results showed that the proposed DRL based handover scheme can decrease the number of handovers more than 21% compared with the comparison benchmarks under no handover failures.