Design And Implementation Of Graph-based Intelligent Handover Method In Satellite-ground Convergence Network

In satellite-ground convergence networks,due to the high-speed movement of low orbit communication satellites,users frequently handover between different satellites or different beams of the same satellite during communication.When users handover between different satellites,due to the relatively concentrated distribution of users,they often make handover requests to same satellite at almost the same time,leading to group handover problem and increasing network congestion,handover failure rate,and access blocking rate.When users handover between different fixed beams of the same satellite,in addition to reducing the handover frequency,it is also necessary to reduce the handover failure rate and improve the user’s signal strength to improve communication quality;In addition,the small variance of average channel utilization indicates that the number of users connected to each beam is relatively uniform,which is conducive to the long-term operation of the system.Therefore,it is necessary to reduce the variance of average channel utilization.Based on this,this paper studies the inter-satellite handover and inter-beam handover methods of users,which is of great significance for improving the handover performance of satellite-ground convergence networks.Aiming at the problem of frequent handover and group handover between satellites,this paper proposes an inter-satellite handover method based on spectral clustering and knapsack problem,after judging the user type,we construct a user similarity graph,and use the spectral clustering method in graph theory to divide the user similarity graph to realize the grouping of users,obtain the user groups that may handover almost at the same time,and redivide the user groups to disperse the user groups to different time slots or satellites to alleviate network congestion.The allocation problem of satellites and time slots is abstracted as a multiknapsack problem,and when the number of idle channels is limited,the total value of channel quality,elevation angle,and remaining service time is considered,and a heuristic algorithm is used to obtain a handover decision.Furthermore,aiming at the problem of users handover between beams,an inter-beam handover method based on graph convolutional network and reinforcement learning is proposed,and the features of the beam are preliminarily extracted by constructing a handover topology graph,the features of the handover topology graph are extracted by the graph convolutional network,and the handover judgment is obtained by using the reinforcement learning method.In this paper,a satellite-ground convergence simulation environment is built based on STK and Python,and the performance of the inter-satellite handover method based on spectral clustering and knapsack problem and the inter-beam handover method based on graph convolutional network.and reinforcement learning are tested and analyzed.The simulation results show that the inter-satellite handover method based on spectral clustering and knapsack problem reduces the handover failure rate and access blocking rate,reduces the handover frequency,and improves the average signal power.The inter-beam handover method based on graph convolutional network and reinforcement learning reduces the handover failure rate,access blocking rate,and average channel utilization variance,improves the average received signal strength,and reduces the handover times.