Research On Characteristics Of Birth Nodes And Death Nodes In Internet Macroscopic Topology

As a typical instance of complex networks, the Internet macroscopic topology is widely investigated with the combination of network science. Recently, researches have made considerable progresses, especially in the Internet autonomous system level (AS-Level) topology. However, because of the complexities, the internal laws within the IP-Level Internet topology need to be further explored. Therefore, the characteristics of birth nodes and death nodes from multiple perspectives should be investigated to thoroughly understand the Internet evolution nature and provide a valid basis for the next generation Internet design, which is of great significance.Firstly, degree evolution of birth nodes and death nodes is analyzed. By analyzing the massive IP-Level Internet data of CAIDA Ark project, we observe that both birth and death nodes'degree follow the power-law distribution. Since birth edges do not accord with preferential attachment, the rich get richer theory has little significance on dynamic topology and nodes with high degree do not bring great impact on birth nodes. In addition, stable edges have a stable probability distribution which derives the universal law of generating mechanism of the Internet.Secondly, taking k-core decomposition, the hierarchy characteristics of birth nodes and death nodes in IP-Level topology is analyzed by using the actual data measured by CAIDA Ark project. It is indicated that stable coreness distribution of birth and death nodes is similar, and both of them remain unchanged. Analyzing the relationship between node degree and node coreness on quantitation, we observe that there is no evident relation between them. The coreness distribution of birth nodes' adjacent nodes is similar to that of death nodes'adjacent nodes, which can help the Internet hierarchy control.Thirdly, the distribution of birth nodes and death nodes in the Internet communities is analyzed. It is found that the single type of nodes is less likely incorporated. Communities tend to consist of dynamical and stable nodes, and specifically, the proportion of dynamical nodes proportion from 50 percent to 60 percent has the highest possibility, which can help better understand the characteristics of the Internet local communities.Finally, in order to find out whether characteristics of birth nodes and death nodes are related to measurement levels or not, AS-Level actual data of CAIDA Ark project is studied and compared with IP-Level topology. The results show that metabolism of nodes, evolution of non-preferential attachment, and proportion of stable edge number all related to measurement levels. Characteristics of the dynamic nodes under different measure levels are different in essence. The Internet evolution natures are related to measurement level, which provides a valid basis for understanding the Internet evolution mechanism and forecasting the direction of Internet evolution.