With the in-depth advancement of 5G technology, Non-Terrestrial Networks (NTN) have become an important supplement to ground communication due to their advantages of wide coverage and Low latency. The high-speed movement and frequent handover of Low Earth Orbit (LEO) satellites bring challenges in access and resource allocation. This paper focuses on the intelligent satellite selection strategy for dynamic environments in low-earth orbit satellite networks and proposes a reinforcement learning method based on the Proximal Policy Optimization (PPO) algorithm to improve the success rate of user access and the utilization efficiency of communication resources. By constructing the state space, action space and reward function, and optimizing the algorithm hyperparameters, the system can dynamically learn the optimal satellite selection strategy. The simulation results show that the PPO algorithm is superior to the traditional satellite selection strategy in terms of handover success rate, access delay and resource utilization rate, verifying its application potential in the dynamic satellite network environment.
Wan et al. (Mon,) studied this question.