Research On Complex Network Characteristics Of Internet Topology

Today,the Internet has infiltrated all aspects of our lives.The Internet has brought us increasingly rich and powerful services,meanwhile,its scale has become increasingly large and its structure has become increasingly complex.Since the Internet is a large and complex network system,it requires us to have a sufficient understanding of the characteristics of the network topology to conduct effective deployment,optimization,application,and supervision of the Internet.Based on the research on the Internet topology,we can better understand the structural characteristics and evolution rules of the Internet's underlying topology,and it can provide guidance for the deployment,optimization,and application of the Internet.The Internet is a typical complex network.There are many advantages over traditional analysis methods by using the theory of complex networks to analyze the Internet topology.The identification of Internet topology and the study of network robustness are two important research topics in the study of Internet topology.This thesis applies the methodology of complex networks to study the structural characteristics of the Internet topology,and explores the characteristics of the relationships among nodes and the characteristics of the groups to which the nodes belong.These features are applied to the Internet topology identification to solve the problem of identifying the affiliation between ASes and routers,and applied to the study of the robustness of Internet topology to solve the problem of analyzing the important nodes of the AS-level topology.The main work can be summarized in the following two parts:1.Aiming at the problem of identifying the affiliation between ASes and routers,this thesis proposes a method to assign routers to their owner ASes based on community characteristics.According to the characteristics of incomplete and insufficient data information in the router's owner AS identification problem,a method of obtaining information from the node itself,the similar nodes of the node,and the community where the node is located is proposed.In this method,stricter heuristic decision rules are used to ensure high decision accuracy.Based on the RA index,an RA indicator(TriangleRA)based on the triangle structure in the router topology is proposed to calculate the structural similarity of router nodes.Similar nodes and communities on basis of port and structural similarity are used to obtain additional information.2.Directing at the study of robustness of Internet,this thesis proposes an adaptive fast algorithm for node importance based on Collective Influence(CI).In order to study the vulnerability of the Internet in the face of deliberate attacks,the importance of Internet topology nodes is analyzed.The CI algorithm is applied to the selection of important nodes in the Internet topology,and the application results are analyzed to study the characteristics of the Internet topology nodes selected by the CI algorithm.In view of the shortcomings of the CI algorithm in the important nodes selection of Internet topology,based on the features of the Internet topology and the preferences of CI algorithm parameters,an adaptive fast CI algorithm is proposed to improve the applicability of the algorithm.The thesis conducts experiments on the real Internet topology.The experimental results illustrate that the method of this thesis improves the accuracy rate and recall rate by 10.2% and 5.4% respectively compared with CAIDA's heuristic method by using the link data obtained from CAIDA to identify the attribution between AS and routers.This thesis uses the AS-level topology provided by Stanford University to analyze important nodes and compares the improved algorithm proposed in this thesis with other common methods.The experimental results show that the improved algorithm proposed in the thesis is stable,and its overall performance in accuracy and computational complexity is better than other algorithms.