Research On Edge Computing Architecture Of Leo Constellation Communication Network

Due to the advantages of wide coverage,flexible networking and low cost,LEO constellation communication network is suitable for the interconnection of everything and global communication.However,due to the large transmission delay in satellite communication network,it can not effectively meet the low delay requirements of high real-time services.In order to meet the needs of real-time services in LEO constellation communication network,this thesis uses the research of edge computing technology in ground broadband network for reference,applies it to LEO constellation communication network,sinks the processing capacity of cloud computing center to the network edge close to the user,so as to reduce the response delay of tasks and the bandwidth consumption of the return network of LEO constellation communication network.The research work of this thesis is mainly from the following aspects:Firstly,according to the characteristics of LEO constellation communication network,the thesis introduces the concept of edge computing,and innovatively designs the edge computing architecture of LEO constellation communication network.Considering that LEO satellite is a resource constrained device,an efficient virtualization technology is needed to improve the utilization rate of LEO satellite resources.At the same time,due to the time-varying topology of LEO constellation network,the link between the satellite and the back-end gateway station will be temporarily interrupted,resulting in the "offline" situation of LEO satellite.In order to solve this problem,the thesis designs the platform management layer of satellite edge computing,which makes the satellite edge computing node have the ability of independent work,so that it can provide continuous application services for the satellite end users.Secondly,the simulation test platform for edge computing of LEO constellation communication network is designed,and the functional verification and performance test are carried out for two kinds of typical application cases,I/O intensive and computing intensive.The experimental results show that the edge computing architecture designed in this thesis has the functions of general task processing and offline work support.At the same time,compared with the traditional back-end cloud computing center mode,the low orbit constellation communication network edge computing mode can reduce the response delay by 57% and the network bandwidth consumption by 66%.Finally,this thesis proposes a new algorithm based on queuing theory and comprehensive weighting.For the computing unloading request submitted by the end user,this strategy first considers to complete the unloading request on the nearest edge computing node of the communication access satellite;if the edge computing node on the communication access satellite is not competent,then comprehensively considers the total energy consumption of data transmission and calculation generated by forwarding the unloading request,the load balance between the computing nodes and the users' response delay and other factors of the candidate computing nodes are scored respectively,and finally the computing node with the highest comprehensive score is selected as the unloading target.Simulation experiments show that the strategy can reduce the average response delay of users by 17% to 34% while processing the unloading requests of all users in the world.