空天地网络中基于异质图学习的协同覆盖方法OA

Space-Air-Ground networks(SAGNs)represent a key development direction for 6G mobile communications,enabling global wide coverage and ubiquitous communication services by integrating satellite,aerial,and terrestrial networks.When low-Earth orbit satellites and high-altitude platform networks collaborate for coverage,traditional access selection algorithms based on distance or elevation angle weighting often lead to network node overload,resulting in degraded user communication quality.A collaborative coverage method based on heterogeneous graph learning is proposed to address this issue.This method constructs a heterogeneous graph to represent network nodes,terminal devices,and the relationships between them.A heterogeneous graph deep reinforcement learning model is designed to extract topological and implicit feature information of the SAGN through message passing and aggregation mechanisms,transforming the collaborative coverage problem into solving the connection probabilities of edges between nodes in the heterogeneous graph.Simulation results,conducted in a SAGN environment under various user scales,demonstrate that the proposed collaborative coverage algorithm achieves fast convergence,reduces average network load,and improves user communication rates.