Multi-Objective Intelligent Routing Algorithm for LEO Satellite Networks Based on DQN

In recent years, LEO satellite networks have developed rapidly and are widely used in global communication and internet access. This paper first summarized the current development status of LEO networks and the main LEO network routing algorithms. At the same time, aimed at the challenges of single optimization objectives, slow convergence speed, and difficulty in optimizing indicators such as delay and packet loss, this paper proposed a multi-objective intelligent routing algorithm based on deep Q-network (DQN). With the designed DQN multi-objective reward function, it realized the optimization of delay, packet loss, load balancing. Regarding delay, the proposed algorithm is superior to the Shortest Path First (SPF) algorithm. In terms of traffic distribution index and packet loss rate, the performances of the proposed algorithm are improved by more than 14% and 10% compared with the SPF and Dijkstra algorithms respectively. In addition, this paper illustrated the application methods of the proposed intelligent routing algorithm in various scenarios.