Internet Macroscopic Topology Characteristics Evolvement And Spectrum Research

The Internet is full of complexity. So if trying to improve the performance of Internet and redesign it by researching its structure, we have to do researches on the macroscopic topology. Recently, researches in this field have achieved great developments and many characteristic laws behind of the internal network were fond out. Currently the related research neither has enough space or time span or newest for data computing and analyzing, nor combined with other areas, nor has much complex measurements. Therefore, more comprehensive research needs to be done.In this dissertation, the research based on the massive data authorized by CAIDA(The Cooperative Association for Internet Data Analysis) Ark project and the data’s time span is from September2007to February2011. Calculate various characteristics of Internet AS-level macroscopic topology and analysis of the relationship between the characteristics. Then through the rich-club and the cluster coefficient, power law degree distribution of the PDF and CCDF analysis of connectivity and power law of the network. Different from earlier studies, we found the connectivity of network is enhanced. Demonstrate the core of the topology hierarchy of significance, that the core can be used to more accurately describe the network topology level. Analysis of the evolution of the core of the network indicates that over time to enhance the network level; Analysis of the core distribution of nodes that in lower region conform to power law. Analysis of the relationship between node degree and core, then analysis of degree distribution of each shell nodes. The study found the highest core nodes that an important influence on the network. The connection between different set of core nodes in the network expression correlation between all levels, analysis of it found that the highest core nodes of the network are the real core of the network.Algebraic theory combined with the nature of network topology, research the spectral characteristics of the network. Discussed the theory of the atlas, then detailed analysis of the adjacency spectrum and laplace spectrum. Demonstrate the connection preferences lead to a connection between nodes in the network there is some correlation. Reconnect edge algorithm proposed by to reconnect the network, and then analysis of the degree correlation coefficient influence on adjacency spectrum and laplace spectrum in assortative and disassortative network.