Research On Caching Optimization Strategies For Satellite-terrestrial Integrated Networks

Satellite-terrestrial integrated networks tightly combine terrestrial wireless communication networks with satellite networks,and can achieve global seamless coverage owing to its flexible networking,strong disaster resistance,high reliability,emergency communication and other advantages.It’s also a critical infrastructure that pertains to national security strategic needs,and has vast potential for development and application prospects.Meanwhile,as a widely used network performance optimization technology,cache can effectively alleviate network congestion caused by repeated request services,and shorten the delay in obtaining request content,which is of great significance for improving the performance of satellite-terrestrial integrated networks and the Quality of Experience(QoE).However,existing caching strategies have not taken into account factors such as differentiated demands for concurrent multi-service and dynamic coverage by multi-satellite,leading to a decline in network performance and QoE.Therefore,it is necessary to further investigate and optimize caching strategies in satellite-terrestrial integrated networks.In view of this,this thesis focuses on the cache optimization strategy for the typical scenarios of concurrent duplicate request multi-services and dynamic coverage in satellite-terrestrial integrated networks.The main works of this thesis are summarized as follows:Firstly,in response to the issue of excessive link load and difficulty in ensuring QoE when dealing with diverse repetitive requests,we propose a two-level differential caching strategy for concurrent multi-service based on deep reinforcement learning(DRL)in the thesis.Through analyzing and modeling the delay of obtaining request content and the time utility functions of three typical services in satellite-terrestrial integrated networks,an optimization problem with the objective of maximizing the total system utility is formulated,and then solved through the use of the Multi-agent Deep Deterministic Policy Gradient(MADDPG)algorithm.The strategy comprehensively considers multiple factors such as the differentiation of service utility,user request status,satellite-terrestrial caching status,and network topology,to determine cache update decisions.Experimental results demonstrate that the proposed algorithm can increase the total system utility by approximately 47%,when compared with traditional cache update strategies such as Most Popular Content(MPC)strategy and Random Replacement(RR)strategy.Secondly,to address the coupling issue between cache and access strategies in the multi-satellite dynamic coverage scenario,a joint optimization strategy for caching and access based on DRL is proposed in the thesis.With the objective of minimizing the total delay of obtaining request content,the proposed strategy considers multiple factors such as user request status,satellite-terrestrial caching status as well as network connectivity status.An optimized caching and access solution is obtained based on the MADDPG algorithm.Simulation results demonstrate that compared with traditional cache and access strategies based on MPC/RR strategy,such as maximum cache hit ratio,maximum service time and shortest distance,the total delay of the proposed algorithm can be reduced by about 35%.