Research On Task-oriented Terrestrial-satellite Network Architecture Design And Route Optimization

With the rapid development of 5G/6G,Io T,Io V,big data such as video and remote sensing have become the main objects of information transmission in terrestrial-satellite networks.However,the topology of the terrestrial-satellite network has obvious timevarying characteristics.The high speed of satellite nodes makes it difficult for the network to provide a stable end-to-end path.At the same time,the end-to-end communication mode has become a bottleneck with high latency of network request response,which poses a huge challenge to the efficient transmission of flows.At present,the application of IP-based network architecture to terrestrial-satellite network faces problems such as complex control process,difficult network management,and low routing efficiency.Aiming at these problems,the thesis starts from two aspects of the architecture design of the terrestrial-satellite network and the routing optimization problem.Aiming at the architecture design of the terrestrial-satellite network,the thesis takes network architecture implementation and controller placement as the research content,and proposes a terrestrial-satellite network based on SDN(Software-Defined Networking)and ICN(Information-Centric Networking).The architecture uses the controller to achieve centralized control of the network,At the same time,the network uses IP multiplexing to identify ICN packets and IP packets,and uses multi-level flow tables to realize ICN network functions such as in-network caching,request aggregation,and name routing.Then,for the controller placement problem,the thesis divides it into two categories: single-objective optimization and multi-objective optimization.Single-objective controller placement uses a PAM(Partition Around Medoids)algorithm to calculate the optimal location of controller placement to minimize the average propagation delay between the switches and the controller.For multi-objective controller placement,the problem modeling is carried out,and the objective function is established.After that,the improved genetic algorithm is used to calculate the optimal solutions of the controller placement.The final simulation results show that,compared with the traditional SDN architecture,the terrestrial-satellite network architecture based on SDN and ICN proposed in the thesis can effectively reduce the request delay and hops; The proposed controller placement solution can effectively improve network performance.Aiming at the routing optimization problem of terrestrial-satellite network,the thesis takes the design of network caching strategy as an entry point,and proposes a static routing algorithm for neighborhood requests based on cooperative caching mechanism.Firstly,the thesis designs a collaborative caching mechanism suitable for terrestrial-satellite networks.The caching mechanism divides the popularity of contents and adopts different caching strategies for popular content and unpopular content.When performing cache replacement,satellite nodes can cache discarded contents to satellites in nearby unpopular areas; when the network topology changes,satellites can cache copies of popular content in the next area in advance before entering the next area,thereby improving network caching performance.Then,in order to avoid service interruption due to the dynamic characteristics of the satellite network topology,a static routing algorithm is designed.The algorithm is based on the implementation of virtual nodes and virtual topology.The constellation operation cycle is divided into several time slices with constant network topology.K alternative routes of the node pair are calculated in each time slice.The algorithm selects the route that passes through the most popular areas as the static route of the current virtual topology node pair; at the same time,a neighborhood request routing strategy is designed.When the satellite node receives the Interest packet and the node does not have a copy cache of the content,the node forwards the Interest packet to the adjacent nodes for content data.The final simulation results show that,compared with the heuristic routing algorithm based on the traditional caching strategies,the algorithm proposed in the thesis can effectively alleviate the problems of increasing number of request hops and cache redundancy caused by dynamic changes in network topology.