Deep learning Based Topology Optimization of Marine Satellite-to-Ground Network

Network topology affects important network performance metrics including link utilization, reliability, high energy consumption and delay. In this paper, we propose a deep learning based and with the assistance of MEC to optimize the topology of marine satellite-to-ground network. By combining graph neural network (GAT) and multi-objective optimization strategies, the key issues in the design of marine network topology are addressed. Experimental results show that the proposed method outperforms traditional methods in terms of reduced network latency, reduced energy consumption and improved reliability.