Low earth orbit satellite admission control scheme based on deep Q-learning

Aiming at the characteristics of large delay and fast movement of LEO satellite links, taking into account the user access signal strength and the allocation of on-board resources, an admission control scheme based on cache technology and deep Q-learning is proposed. The experimental results show that the algorithm has good performance in terms of call blocking rate, handover times and network revenue, and can effectively adapt to the high dynamic characteristics of LEO satellites.