Research On Service Placement Strategy In LEO Satellite Edge Computing

In the process of promoting the integration of space and sky in the new generation network,LEO satellite network technology has been developed rapidly.By placing services from cloud computing centers on LEO satellites,edge computing of LEO satellite networks is of great importance to improve user service quality,reduce resource overhead,increase satellite system stability and relieve ground network pressure.At present,most of the studies in LEO satellite edge computing use generic system models for service placement,ignoring the differences among satellite scenarios,resulting in limited applicability.Therefore,this paper focuses on how to use the scenario characteristics to select suitable locations on the edge nodes to deploy services from cloud computing centers in the edge computing scenario containing only LEO satellites and the edge computing scenario with mixed LEO satellites and ground nodes to realize edge computing in LEO satellite networks.The specific research points and the research results achieved in this paper are as follows:(1)For the service placement problem in LEO satellite networks,two service placement algorithms based on weighted directed graph and service duration window are proposed.Firstly,we consider the constraints of satellite resources and service migration cost in LEO satellite networks and construct a service placement model with the objective of minimizing service communication delay.After that,a weighted directed graph mapping original problem is used to design a time-slice by time-slice solution algorithm.The service placement algorithm based on the service duration window is also designed by combining the characteristics of LEO satellite constellation.The simulation environment of LEO48 satellite constellation is constructed for experimental verification,and the results prove the effectiveness of the proposed algorithm in reducing the time delay and controlling the cost.(2)The service placement algorithm based on multi-dimensional binary Drosophila optimization is proposed for the cooperative service placement problem.The research firstly considers the constraints of service placement cost,server resources and service response latency,and formalizes the star-ground collaborative service placement problem with the objective of maximizing the number of service placements.A multi-dimensional extended binary Drosophila optimization algorithm is used to solve the original problem,and a new repair factor is introduced to accelerate the search process.Experimental validation is carried out using a real user request dataset,and the results show that the proposed algorithm performs well in securing user services,controlling costs,and has fast convergence.