Robustness Research And Analysis Of Scale-free Networks

With the rapid development of network technology,network security and corresponding countermeasure technologies have received extensive attention.In our real life,the network can be abstracted into a scale-free network model.Simulation attacks on different scale-free network models and robust analysis of successive faults are important for human real life and network security.Based on the complex network theory,this paper studies the robustness of scale-free networks.Firstly,it introduces the basic statistical characteristics of complex networks and related network models.The BA model,the growth model,the dynamic evolution model and the local world model network belonging to the scale-free network are established by the construction method and algorithm description.Then,five robustness indicators are set for the maximum connected subgraph,network connectivity efficiency,availability metr:ic,anti-destructive metric and network cost metric.Robust evaluation strategies are proposed,including:analysis methods,analysis steps and analysis strategies.Random attacks and deliberate attack algorithms are used to attack the network model,causing successive failures and analyzing their robustness.According to the analysis strategy,the network characteristics and robustness of the four scale-free network models are measured.The best robust model is the local world evolution model.Taking the minimum node degree value of the model as the main factor affecting the robustness as the starting point,an optimization method for setting the node saturation value is proposed,and the appropriate saturation value of the model is analyzed.Experiments show that the robustness of the model before and after the saturation-based optimization is basically the same when subjected to random attacks.When it is deliberately attacked,the robustness is increased by 5%on average;In the value interval of the saturation value when deliberately attacked,the robustness of the network model increases monotonically as the set node saturation value decreases.In order to further strengthen the robustness of the network model,a community partitioning algorithm using Louvain algorithm is proposed to di'vide the network model into associations,and the optimization method of connecting edges is adopted by using random and deliberate edge-to-edge strategies among communities.Through experimental comparison,in the deliberate side-by-side strategy,as the importance of connected nodes decreases,the robustness of the network model increases monotonically.Finally,the model before and after optimization is compared to verify that the local world evolution model based on the decentralized edge algorithm of the center coefficient of the community is the best.The most improved robustness after optimization is that the maximum connected subgraph(G)is improved by 55%and the network.The cost metric(?b)has increased by 40%.