空天预警研究学报2025，Vol.39Issue(5)：356-359,374,5.DOI:10.3969/j.issn.2097-180X.2025.05.009
基于深度强化学习的5G-LEO融合网络动态切换方法
Deep reinforcement learning-based dynamic handover method for 5G-LEO converged networks
摘要
Abstract
In order to address the degradation of system information rate and the decline of Quality of Service(QoS)caused by traditional handover mechanisms in 5G-LEO converged networks,this paper proposes a deep Q-learning network-based dynamic handover(DQDH)algorithm.First is established the objective optimization problem of maximizing the system information rate,which is then transformed into a Markov Decision Process(MDP).And then,by setting the state space,action space and reward function,the intelligent agent is enabled to make decisions that are conducive to optimizing the information rate of the system.Simulation results demon-strate that,compared with the benchmark methods,the proposed DQDH approach improves the information rate of the system and reduces the time delay between LEO satellites and ground base stations when the access point handover is performed.关键词
低轨卫星/切换方法/信息速率/深度Q网络/切换时延Key words
low earth orbit(LEO)/handover algorithm/information rate/deep Q-network/handover delay分类
信息技术与安全科学引用本文复制引用
孙士兵..基于深度强化学习的5G-LEO融合网络动态切换方法[J].空天预警研究学报,2025,39(5):356-359,374,5.基金项目
湖南省教育厅科学研究项目(23C1076) （23C1076）
长沙市科技局自然科学基金项目(kq2402035) （kq2402035）
