Research On Modeling Of Multiplex Complex Networks And Identification Of Important Nodes

With the widespread of Unmanne Aircraft Vehicle swarm(UVAS)in the military field,research on UVAS technology has raised great attention.However,subjected to the huge amount of data and path diversity of information interaction,the UVAS has become the weakness exposed to the enemy during the mission,thus requiring urgent solution.This thesis is settled on the background of UVAS information interaction.Based on the complex network communication theory,propagation model and key node identification method in complex network are utilized.The research mainly explore UVAS’s data transmission modeling process and central UVA recognition problem in the battlefield with high complexity,incomplete information and strong confrontation.The main research results of the thesis are as follows:(1)For the multiple layers characteristic of UAVS network,a multiplex complex network susceptible-infected(SI)propagation model and a multiplex complex networks Susceptible-Infected-Delay Removed(SIDR)model added with confrontation are designed based on the complex network communication theory.Under the background of large amount of information interaction and behavior within the UAVS,according to the internal information interaction characteristics within UAVS and transmission method,the UAVS network can be divided into command layer,situational awareness layer,decision planning layer and command layer.The differences in network structure of different propagation modes are also considered.The complex network communication theory is used as the theoretical means.On the basis of the existing single-layer complex network propagation model,a multiplex complex network susceptible-infected(SI)propagation model and a multiplex complex networks Susceptible-Infected-Delay Removed(SIDR)model are introduced to describe the characteristics of UAVS’s information interaction.Under the comparison experiment with single-layer complex network propagation model,the multiplex complex network communication model is analyzed for the large amount of information interaction and path diversity.(2)A single-layer complex network efficiency index is established using the path-based method.And then setting the index as the basis of multiplex complex network layer analysis,a multi-layer complex network layer weight(LW)calculation method is proposed.Under the status quo of high complexity and large dimensionality in the UAVS information interaction model,it is limited to use traditional method that analyzing multi-layer complex network propagation problems with current single-layer and multi-layer classical mathematical modeling methods.A novel method is proposed to transform the multi-dimensional problem into a two-dimensional problem.To achieve this,a single-layer complex network efficiency index is designed and a Layer Weight(LW)method is proposed based on the index.By designing the weight index K,the importance of nodes at different layers is settled to calculate the importance of nodes in the entire multi-layer complex network.Through two sets of experiments,the validity of the network LW method of multiplex complex network in solving the problem of important node identification under the background of multiplex complex network propagation is verified.(3)Considering the dual effects of the transmission path and the neighbors of the nodes in the information exchange of the UVAS,Layer Weight-K-shell-Eigenvector centrality(LW-LC)is proposed for important node identification based on LW method.Several classical important node identification methods in single-layer complex networks are introduced and their limitations in the propagation theory of multi-layer complex networks are analyzed.On the basis of the previous two points,a LW-KC method for important node identification is proposed after analyzing the real situation of UAVS information exchange.LW-KC method is based on UAVS multi-layer complex network propagation model,which can make the UAV selected by this method have the highest transmission efficiency per unit time or the least time for all nodes.Through the experiment of random network data and real network data,and compared to the dissemination rate of other classical important node identification methods in the specified time,the proposed method’s optimization effect on the evaluation index of dissemination rate is verified.