基于强化学习的无人机网络自适应QoS路由算法  

Adaptive QoS routing algorithm of UAV network based on reinforcement learning

作　　者：谭周正 范琅 李宇峰 章小宁 Tan Zhouzheng;Fan Lang;Li Yufeng;Zhang Xiaoning(School of Information&Communication Engineering,University of Electronic Science&Technology of China,Chengdu 611731,China;State Key Laboratory of Astronautic Dynamics,Xi’an 710043,China)

机构地区：[1]电子科技大学信息与通信工程学院,成都611731 [2]宇航动力国家重点实验室,西安710043

出　　处：《计算机应用研究》2025年第4期1177-1184,共8页Application Research of Computers

基　　金：国家自然科学基金资助项目(62171085,62001087,U20A20156,61871097)。

摘　　要：由于无人机网络的动态特性,要保证其具有可靠的通信保障仍存在一定的挑战,尤其是在军事领域,对无人机网络的QoS保障能力要求更高。针对上述需求,提出了一种基于Q学习的自适应QoS路由算法。在该算法中,每个节点通过HELLO消息和数据包相结合的方式来进行邻居信息感知,通过接收邻居节点反馈的ACK来获取链路时延和丢包率,然后根据链路时延和丢包率来更新维护Q表。考虑节点移动特性,算法还引入了节点的位置信息。在路由过程中,节点综合考虑Q表和邻居节点的位置信息来进行最优下一跳选择。通过仿真验证,对比参考的路由算法,提出的路由算法能够在较低的路由开销下提供更低的传输时延和更高的传输成功率。Due to the dynamic nature of UAV network,there are still certain challenges in ensuring reliable communication support,especially in the military field,which requires higher QoS guarantee capability of UAV network.Aiming at the above requirements,this paper proposed an adaptive QoS routing algorithm based on Q-learning.In this algorithm,each node detected neighbor information through the combination of HELLO messages and data packets,received ACK feedback from neighbor nodes to obtain the link delay and packet loss rate,and then updated and maintained the Q table based on the link delay and packet loss rate.Considering the mobility characteristics of nodes,this algorithm also introduced the location information of nodes.During the routing process,the node comprehensively considered the Q table and the location information of neighbor nodes to select the optimal next hop.Through simulation verification,the proposed routing algorithm can provide lower transmission delay and higher transmission success rate under lower routing cost compared with the reference routing algorithms.

关 键 词：无人机网络 强化学习 路由算法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]