Research On Multi-dimensional Resource Management Techniques For GEO-LEO Satellite Communication System

With the development of ground mobile communication and internet communication,most cities and towns have realized real-time communication.However,due to the terrain and geographical constraints of the ground communication system,it is still not covered in remote areas such as mountains,deserts and oceans.With the development of satellite communication technology,it is possible to use the advantages of wide-area coverage and flexible satellite-ground network.In recent years,with the design and development of several satellite internet systems,especially the deployment of the "Starlink" system and its rapid launch,the technology of internet communication of satellite constellation has made great strides in various technical fields,including on-board regeneration communication technology,laser communication technology,space routing technology,satellite multi-beam hopping technology and so on.As more and more tasks are taken on by satellite communication systems,the constraint of satellite resources has become one of the critical factors which limit the performance of the system.Therefore,the research on satellite mission deployment and resource management and allocation technology has attracted much attention.Due to the limitation of on-board resources,in the satellite communication system,it is necessary to pay attention to the scheduling of traditional communication resources and the management of on-board computing and processing resources,storage resources,etc.Based on different GEO-LEO satellite communication scenarios,multi-dimensional resource management technologies,especially the technologies of cooperative user association and resource allocation of hybrid satellite network,intelligent resource allocation of multi-beam satellites,and intelligent satellite handover strategy based on onboard caching are studied in the dissertation,which are supposed to improve the overall performance of the system.1)In offloading field of GEO-LEO hybrid satellite network,the algorithm in the past is lack of considering dynamic time-varying characteristics based on cooperative user task scheduling and resource allocation,as well as the long-term continuous performance optimization.Reseai-ch on dynamic time-varying cooperative offloading technology are canried out in the dissertation,in which and an algorithm of cooperative user association and resource allocation(CUARA)is proposed to minimize the system total task delay.In order to solve the dynamic mixed-integer nonlinear problem of system optimization,it is decomposed into two sub-problems:sequential decisions for user association and resource allocation with fixed user association conditions.It has been proved that the optimal resource allocation problem with selected users is a convex optimization problem,which can be solved by a convex optimization method.The problem of sequential decisions for user association with known resource allocation is an integer linear programming(ILP)problem,which can be solved by deep reinforcement learning(DRL).The result of the simulation show that CUARA algorithm presented in this dissertation performs better in terms of system delay.2)Aiming at the problem that the traditional algorithms of multi-beam satellite communication are mostly based on beam-level scheduling,the granularity is large and the flexibility of resource scheduling is limited,a joint user scheduling and resource allocation near-end policy-fu-st algorithm(PPO-JSRA)is proposed,in which the full frequency is used and in order to reduce the interference,the user task scheduling,fr-equency and power resource allocation are planned dynamically.Simulation results show that,the algorithm,which combines user task scheduling with power and frequency resource allocation,makes a continuous,dynamic,intelligent decision to optimize system delay.Through simulation,it has apparent improvement in system throughput and delays.3)Considering that the traditional handover strategies lack considering the buffer resource of the source satellite which also affects the handover success rate,an intelligent multi-attribute fusion algorithm(IMF)is proposed in this dissertation which studies the joint effect of residual service time,residual idle channel and residual buffer capacity on handover performance in dynamic network topology.In this dissertation,the handover process is defined as a Markov decision process,and an intelligent handover strategy based on caching awareness is proposed for LEO satellites.Through simulation,there are performance improvements in the handover failure rate and call blocking rate.