| Aerospace information network is an information transmission network composed of various space-based satellites and space-based aircraft nodes and dynamically connected through various wireless links.It has very important application value in modern military and civilian fields.Aerospace information network has the characteristics of large spatial coverage,high dynamic change of network topology,large number of nodes and strong heterogeneity,which are significantly different from terrestrial networks.With the rapid development of cloud computing technology in recent years,the industry has begun to study the combination of cloud computing technology and aerospace information network to build an aerospace cloud information network.The cost of the test platform constructed based on the real aerospace cloud information network is very high.Therefore,the study of high-fidelity network emulation technology is of great significance to the research of aerospace cloud information network.Facing the emulation requirements of the aerospace cloud information network,the paper studies the key technologies of the aerospace cloud information network emulation,and designs and implements a high-fidelity aerospace cloud information network emulation platform based on resource virtualization,cloud computing and edge computing technologies.Firstly,the paper analyzes the architecture and emulation requirements of the aerospace cloud information network as the emulation target.Then,a layered architecture model for the aerospace cloud information network emulation platform is proposed,including: resource virtualization layer,cloud resource management layer,and emulation control layer.Next,according to the design of the architecture model,and based on the secondary development of commercial shelf hardware and open source software,the paper realizes the multi-level cloud information network emulation platform in space.Finally,in order to improve the fidelity of topological dynamic emulation by the aerospace cloud information network emulation platform,the paper analyzes the emulation optimization requirements of the dynamic reorganization scene of UAV formation,and designs and implements a multi-objective differential evolution algorithm based on load balancing.The experimental results of performance evaluation show that the performance evaluation of the multi-objective differential evolution algorithm based on load balancing,in which the accuracy of CPU load prediction is as high as 88%.In addition,the algorithm in this paper is compared with the LRR＿MMT and MAD＿RS algorithms.It is found that the network traffic consumption is about 20% less than the comparison algorithm,the average CPU utilization of the underlying physical machine is about 30% higher than that of the comparison algorithm,and the maximum energy consumption is reduced by about 35% year-on-year.Therefore,the algorithm in this paper can improve the resource scheduling efficiency of the network emulation platform by improving the virtual machine migration efficiency.In this paper,the function test and performance evaluation of the constructed aerospace cloud information network emulation platform are carried out.On the one hand,the experimental results of functional testing show that the emulation platform can realize the function of deploying multi-level cloud scenarios and the function of dynamic migration of multi-level cloud nodes;On the other hand,the experimental results of performance evaluation show that the error between the packet loss rate created by the simulated link and the preset parameters is 10%,the network topology creation delay deviation is 0.6%,and the packet loss rate error is 8%.Therefore,the network emulation platform designed and implemented in this paper can meet the high-fidelity emulation requirements of the aerospace cloud information network,and has important theoretical research and practical application value. |
PDF Full Text Request