| With the development of informatization,communication demands in remote areas including land and sea increase.Satellite communication with a wide range of coverage,stable service and other characteristics have been more widely used.Among them,LEO satellite networks,such as One Web and Starlink,are developing rapidly due to the advantages of shorter transmission delay,less path loss and lower launch cost.Due to the small coverage area of LEO satellites,the satellite network formed by them requires a large number of satellite nodes.Moreover,because of the low orbital altitude of LEO satellites and their fast motion speed,the network topology of LEO satellites is more dynamic and has higher research value.In this paper,the idea of finite state automata is used to model satellite networks,and a feasible link planning algorithm and topology handover algorithm are proposed.The main research of this paper focuses on the following aspects:(1)Establish a multi-objective optimization model that takes into account topology stability,power consumption,and transmission delay.From the point of view of LEO satellite network,this paper analyzes the necessity of research on inter-satellite link planning algorithm.And this paper uses the idea of finite state automata to model the LEO satellite network,analyzes the measurement indicators of the topology performance of the LEO satellite network,and establishes a multi-objective optimization model that takes into account the three aspects of topology stability,power consumption,and communication transmission delay.(2)Improve the NSGA-III algorithm.Due to the limitation of the number of connection links of the laser transmitter and receiver carried by each satellite in the LEO satellite network,the random generation algorithm adopted by the traditional NSGA-Ⅲ algorithm for population initialization leads to a large number of initialization populations that do not meet the restriction conditions,which seriously reduces the computational efficiency.So this paper improves the initialization of the NSGA-Ⅲ algorithm.And because the solutions in the Pareto solution set output by the NSGA-Ⅲ algorithm cannot be mutually Pareto dominated,this paper proposes a combination of the TOPSIS algorithm and the NSGA-III algorithm to select the final solution with better performance from the Pareto solution set.(3)Propose a multi-slot topology construction strategy based on dynamic programming.Since the frequent demolition and construction of links in the LEO satellite network consumes a certain amount of bandwidth,and link alignment also consumes a certain amount of energy,it is necessary to switch less links between different time slots.In addition,there is an inverse relationship between the packet loss rate of the satellite network and the number of links in the topology,and only considering the number of switching links may result in a series of selected topologies with fewer links and a higher network packet loss rate.Based on the Pareto solution set proposed by the NSGA-III algorithm,this paper selects a series of topologies with better cross-slot performance based on the dynamic programming algorithm,combined with the evaluation indicators of the number of switching links and topology throughput. |