Analysis Of AS-level Topology Under Skitter And Ark Measurement Infrastructure And Dynamic Coreness Model

Internet is a classical intstance of complex network. The research on characteristic analysis and modeling of its macroscopical topology is very important and has already attracted broad attention of academa. Recently, researches in this field have achieved great developments, especially in the autonomous system level(AS-level) Internet topology. In the early day of 2008, the Macroscopic Topology Project of CAIDA(The Cooperative Association for Internet Data Analysis) transited measurement infrastructure from Skitter to Ark. Facing the transition from old dataset to new, it's necessary to analyze the change of topolopy between Skitter and Ark. Meanwhile, although the modeling method used at present can give more and more accurate Internet model, this kind of method did't take the death of nodes and links into consideration and can't represent the hierarchies of network topology—coreness.Considering the problems above, after the design and implement of analysis tool about characteristics, research in this dissertation focus on the power-law, coreness and dynamic changes of topology and select the AS-level Skitter dataset from January 2003 to Decemeber 2007 and Ark dataset from January 2008 to Decemeber 2008 to analyze.Research began with the detailed statistics of various characteristics of Internet AS-level topology, as well as the power-law distribution and the evolution of connectivity of network. The analysis of the similarities and differences in several characteristics were given under Skitter and Ark which approved the power-law property of Internet didn't disappear with the transition of infrastructure. Take the born, and the dead of nodes and links into account, the analysis of the dynamic change of nodes and links showed the Internet topology trends plainly.Then by means of the analysis of the evolution of network coreness and time-slicing data, the law distribution of core was found. As inter-core links the major target, statistics on the distribution of link ratio in each core showed the maximum core has important influence on the network. Summarizing the analysis above, the extract of fitting formulas about the distribution of nodes and inter-core links in each core under Skitter and Ark were used to design the algorithm of dynamic coreness model. Results showed this model can reconstruct the evolution of network topology under Skitter and Ark and exbihit the essence of network better.