Effects Of Network Structure And Virus Spreading Based On Null Model

The spread of the virus has been the important direction in the study of complex networks, this research has very important practical significance. A good complex network null model can offer an accurate reference to the original network. This paper analyzes the statistical properties of dK null models, and uses it to study the influence of the characteristics of network structure of the spread of the virus, and the existing null model algorithms are optimized. This article contribution summarized below:First, we use four kinds of real networks: Internet autonomous system, aviation network, encryption communication network and PPI network, a large number of numerical simulation calculation of the real network and its d K models, the characteristics of the network parameters,such as degree distribution, average nearest neighbor degree, clustering, betweenness, shortest path distance and spectral properties. To observe the data deviation from the null model and real network, to understand greatly the the null model structure.Second,this paper investigates the effect of clustering coefficient on virus propagation Based on Null Model, generating 2K,2.25 K and 2.5K null models. Applying the classical virus spreading model-SIR model to analyze the impact on network clustering coefficient,cluster spectrum and modularity on the virus spreading. The simulation results show that more uniform distribution of clusters, greater modularity and larger clustering coefficient inhibit the spread of the virus.Finally, aiming at the problem of how to quickly and effectively generate 2.25 K and 2.5K null models, two optimization algorithm of generating 2.25 K and 2.5K null models is proposed: dK-SA algorithm and d K-SAPSO algorithm.Combining with the simulated annealing process and Metropolis criterion, the dK-SA algorithm is designed. To improve the convergence rate, the dK-SAPSO algorithm combines with simulated annealing algorithm and particle swarm optimization algorithm.The simulation results indicate that the proposed algorithm can effectively generate 2.25 K and 2.5K null models, In comparison with the current algorithm,the algorithm can effectively improve the convergence rate.