基于图神经网络与深度Q学习的低轨卫星路由算法OA

To address the challenges of rapid topology changes,unstable link conditions,and the difficulty of simultaneously guaranteeing multi-service Quality of Service(QoS)in low-Earth-orbit satellite networks,this paper studies an intelligent routing optimization method for multi-service scenarios.First,link state awareness is achieved through the HELLO message mechanism,to acquire information on link delay,bandwidth utilization,and congestion.Second,a Graph Neural Network(GNN)is employed to extract state representations of network topology and link features.On this basis,a Deep Q-Network(DQN)is introduced for routing decisions,and multi-objective QoS constraints are integrated into the reward function to achieve differentiated optimization for diverse service requirements.Simulation results demonstrate that the proposed method outperforms existing approaches in terms of average end-to-end delay,throughput,and packet loss rate,and can effectively enhance transmission efficiency and service quality of Low-Earth-Orbit satellite networks in dynamic environments.