Research On Fast Topology Construction Method Of Large-scale Digital Twin Networks

Digital twin network(DTN)is the application of digital twin concept in computer and communication network.It is referred as an emulation network with the same network topology as the target tested network and sharing higher-level characteristics and state information in real time.DTN can be used as a test platform for network security policy verification,traffic planning and performance evaluation.With the continuous increase of the scale of the target tested network,the number of underlying physical resources and time required to build the DTN are also increasing in proportion,which will lead to the serious reduction of the construction efficiency of the DTN.Therefore,the research on how to quickly build a large-scale DTN topology has important theoretical and practical significance.For the technical requirements of digital twin network construction,aiming at the typical computer communication network,the rapid construction method of large-scale digital twin network topology based on cloud resource virtualization technology,network emulation technology,topology segmentation and mapping algorithm are researched by this thesis.At the same time,the prototype system of twin network construction is designed and implemented to solve the key problem of rapid construction of large-scale digital twin network.Firstly,aiming at how to realize the efficient utilization and rapid scheduling of the underlying resources of network simulation,the concept of twin network cloud resource partition is put forward by this thesis,and then a mathematical model of it is made.At the same time,a rapid replication strategy of virtual network nodes based on resource buffer pool is put forward by this thesis,which improves the resource scheduling speed and resource utilization of cloud resource partition by instantiating the underlying resources in advance.Secondly,in the scenario of constructing large-scale twin network topology,a large-scale network topology graph segmentation algorithm based on the concept of graph division is designed by this thesis.The algorithm first compresses the topology scale of the target tested network through aggregation,then the network topology is divided into several topology subgraphs under the condition of restricting the number of link cuts,and finally the compressed topology is separated and restored.At the same time,for twin network cloud resource partition and large-scale network topology segmentation algorithm,a target network partition mapping algorithm is proposed by this thesis,which combines the two.Firstly,after the weight of network nodes and the weight of nodes' connected links are comprehensively considered,the target network nodes are sorted by the algorithm,and the consistency between the network nodes and the cloud resource partition are calculated based on the sorting results,so as to select the appropriate cloud resource partition to accommodate the network nodes.Finally,in order to verify the practical effect of the strategy and algorithm proposed in this paper in the rapid construction of large-scale DTN topology,the twin network construction system is designed and implemented by this thesis.The experimental results of 15000 network node scale twin network construction show that the twin network construction system can complete the construction of twin network topology.At the same time,after integrating the strategies and algorithm modules proposed in this thesis,the construction of this scale network takes about 45 minutes,which saves about 32 minutes compared with the initial state,and the speed is increased by about 49%.The strategy and algorithm designed in this thesis solve the problem of low utilization of underlying physical resources,and realize the rapid construction of large-scale DTN topology.