Research Of Distribution Strategy Optimization Based Time-evolving Graph

In recent years,the hybrid network integrating LEO constellation and terrestrial have attracted extensive attention due to its wide coverage,strong inter-link robustness and no limitation of topography.Compared with GEO and MEO,LEO satellite constellation can effectively complete various tasks in the hybrid network with the advantages of low deployment cost and propagation delay time.However,the LEO satellites moving along fixed orbit at high speed results in the rapidly changing of the relative link state among satellite nodes,so that the data transmission performance in hybrid networks is largely limited by dynamic topology changes.In particular,when multiple users in the hybrid network subscribe to the same data from the source node of data publication through LEO constellation,due to the time-varying topology of constellation,the delay of waiting time and transmission cost will increase significantly with the increase of user scale.In addition,LEO constellation has a large number of repeated paths in the process of distributing the same data for multiple users,and these redundant transfers will lead to huge transmission delay and overhead.Therefore,in this paper we will study the data distribution strategy in the hybrid network integrating LEO constellation and terrestrial.In the research process of data distribution strategy,aiming at the time-varying problem of LEO constellation topology,this work will utilize time-varying graph model to model the dynamic topology in the network,and design the global optimal routing algorithm based on the proposed time-varying graph model.At the same time,this paper introduces an in-network cache mechanism to reduce redundant links in the process of data distribution in order to improve the performance of data distribution in the whole hybrid network.The main contribution of this paper is to propose two cache-based data distribution strategies based on two different application scenarios:On the one hand,In the case of one decided source node,this paper will propose a data distribution strategy based on back-tracing partition caching node selection algorithm.Firstly,considering the advantages and disadvantages of the existing time-varying graph model,we will propose a Cross-Slot Graph(CSG)to describe the time-varying topology structure in the network.By using Dijkstra algorithm,the shortest time-delay path set from the source node to any destination node can be found in the graph.On this basis,a back-tracing partition cache node selection algorithm is proposed to optimize the cache location.By comparing with the random cache strategy,the proposed data distribution strategy based on back-tracing partition node cache mechanism in this paper can gain more benefits when the number of cache nodes is small.On the other hand,in the case of any source nodes in the network,in this paper,the distribution strategy based on the Minimum Time-Evolving Connected Dominating Set(Minimum Time-Evolving Connected Dominating Set,MTCDS)MTCDS is proposed.Firstly,an improved Event-Driven Graph called Update Discrete Graph(UDG)is used here to model the time-varying topology of LEO constellation,and a time-dependent connected dominating set MTCDS was constructed as the cache node set in UDG.Compared with the data distribution strategy based on node degree cache and random cache strategy,the data distribution strategy based on MTCDS node cache proposed in this paper can effectively reduce the delay and overhead of data distribution,and the algorithm has good stability.