Evolution Analysis And Modeling Of Internet Hierarchy Based On Cooperative Relationship Between ASes

With the rapid development of Internet economy, Internet has become a new engine which drive the development of national economy. How to providing the Internet service, more quickly and better, for more customers is a key problem what the ISP (Internet Service Provider) is confronting. So research on how to build link relationship between the ISPs driven by the commercial interests and needs is of great significance for understanding the Internet structure, performance, evolution and dynamic.Firstly, the cooperative relationship between ASes in the Internet topology is analysed. This thesis proposes a new algorithm based on the betweenness centrality of nodes which can infers the relationship between ASes by using the clique mesh, node’s betweenness and the triple of AS path. Compared with CAIDA’s algorithm, this new algorithm avoids misinferenc what the transit degree of P is less than that of C, and is more optimal in inferring process. According to validation with the real data about the AS relationship, the precision of new algorithm is 99.5% and 99.1% for P2C and P2P relationships iferred, which is higher than that of GAO, XI A, UCLA because of adopting the clique network and the triple of AS path.Secondly, the hierarchy of AS-level Internet topology is studied. For a start, observing the evolution of the number of ASes and inter-AS links overall and studying the growth in different regions according to the AS number registered in the AS-level Internet macroscopic-topology, we find that network development in North America and Europe led the Internet growth today. Then analyzing the hierarchy evolution of AS-level Internet topology by using some metrics, such as the structure entropy, network depth, network core and network clustering which depict the network hierarchy, the results show that the AS-level Internet topology is in the direction of flatten hierarchy. And then beginning with one AS in the clique, the top-down conical hierarchy structure are built by using the P2C links. Last, we analyze this conical hierarchy from clique network, the level distribution of nodes, the level preference attachment of the P2C/P2P and the preference attachment of nodes degree. According to the evolution results of Internet topology, we get the conclusions as fllows:there is a steady clique network in the Internet; the level distribution of nodes is positive-skewed; the P2C prefer to the high-level connection between the conical hierarchies and the P2P prefer to cross-level connection between the conical hierarchies; the nodes of the P2C/P2P obey power-law distribution of node degree.Finally, a modeling method are proposed.This thesis proposes an Internet modeling method based on level preference attachment (HPA) which consider four characteristics of Internet topology:dynamic growth of nodes and links, hierarchical structure, preference attachment of level and node degree and the commercial relationship of P2C/P2P. After simulation with different parameters in HPA model and comparison with the real Internet topology, HPA model can present the dynamic, hierarchy, preference and commercial relationship of the Internet very well.