Research On End-to-end Routing Technology For Time-varying Satellite Networks

With the increasing communication needs of people and the constraints of the complex environment on the earth,existing networks on the ground can no longer meet people's communication needs.The satellite communication network has the characteristics of flexible networking,wide coverage,rapid network construction,and weak geographical limitations,making it possible for satellite networks to support services that cannot be handled by terrestrial networks,such as environmental and disaster monitoring,scientific exploration,meteorology,and broadcast communications,etc..However,due to the continuous changes in the topology of satellite networks,the small number of satellite nodes,the frequent switching of communication links,and the extension of transmission time,the satellite network is a typical DTN network.At the same time,unlike other dynamic networks with weak regularities,satellite networks are also cyclical and predictable.As a core of network communication,routing technology has a direct bearing on the quality of network services and the number of services.Therefore,it is very important to study the routing technology of satellite networks for satellite networks.The existing research on the routing of satellite networks is roughly divided into two types.The first one is based on the periodicity and predictability of satellites.It uses off-line methods to perform route calculations.The "snapshots" are the most prominent.Another way to calculate the real-time route on the satellite for different data arrival time is mainly represented by the CGR algorithm under the contact graph model.This paper will use these two methods as the starting point for research.One the one hand,this paper introduces the snapshot model and its corresponding Dijkstra algorithm in detail,and then analyzes the problems existing in the model.To solve this problem,this paper proposed a multi-topology search shortest path algorithm(SNMTSP)under the snapshot model.The SNMTSP algorithm searches through the route results of multiple topologies and searches for reachable paths for nodes that do not have an end-to-end continuous link.Simulation results showed that the algorithm improves the end-to-end delivery rate under the snapshot model.On the other hand,this paper studied and improved the deficiencies in the CGR algorithm under the contact graph model.The CGR algorithm does not consider the problem that the network load leads to non-optimal routing.In this paper,the load-aware mechanism based on ant colony algorithm is given.Based on this mechanism,the shortest route under load-aware strategy of ant colony algorithm is proposed.Algorithm(ACSPSN),which uses source node calculations to route routing information into data bundles.Then,for the case that the ACSPSN algorithm is not suitable under certain scenarios,a distributed multi-point decision routing algorithm(MDSP)based on the ant colony algorithm load-aware policy is proposed.The algorithm counters the recalculation of routes at intermediate nodes of the transmission.The load-sensing bias caused by large delay in network transmission,and the final simulation results show that the MDSP algorithm and the ACSPSN have smaller end-to-end transmission delay and higher delivery rate than the CGR algorithm.