Compound Complex Network Model Of Internet Based On Subnet Topology Evolution Model And Its Related Properties

At present, the theoretical study of complex networks has been widely used in the fields of sociology, biology, communications, computer and communication. But in those fields, theoretical research results are not enough deep in practice, therefore, complex network theory is applied into specific application, which will become the research focus in the future.Internet is a typical complex system, complex network theory is used to study the topology and evolution mechanism of Internet, establishment of the network topology evolution model will has important practical significance for estimating network security and survivability, optimizing network performance, designing an efficient network protocol. According to different granularity, Internet topology can be divided into autonomy system-level topology and router-level topology. For the existing autonomy system-level or router-level topology evolution model, there are a lots of problems that some network topological characteristics are different between network generated by model and actual network, and network evolution inherent causes and laws are not accurately reflected. In this dissertation, for the characteristics of different kinds entities and their relationships in Internet, Internet autonomy system-level topology and router-level topology are studied by multi-subnet composited complex network model. Based on these, the cascading failure of Internet is studied. The above studies could improve the development of complex system, which having theoretical and applied significance. The main works in this dissertation include:(1) Empirical research of Internet autonomy system-level topology and router-level topology. Through obtaining authoritative CAIDA-Skitter project topology measured data, from the perspective of network characteristics analysis, the basic characteristics, such as degree, clustering coefficient and so on, of autonomy system-level Internet topology is analyzed. Through self-developed network topology measurement software CERNET-MTR, the measured data of router topology and website page link of the China Education and Research Network (CERNET) are obtained. Using the theory of Multi-subnet Composited Complex Network, the basic characteristics such as degree, shortest path length, betweenness and so on, of topology of CERNET router-level Internet are analyzed. Based on these, accurate understanding of Internet autonomy system-level and router-level topology and its evolution law are obtained. The above studies could act as the theoretical foundation for follow-up Internet autonomy system-level and router-level topology evolution models. (2) Research of router-level Internet topology evolution model based on Multi-subnet Composited Complex Network Model. Based on the basic characteristics of topology and its evolution law from empirical research of Internet router-level topology, using the theory of Multi-subnet Composited Complex Network, through loading router subnet to website subnet, a Internet router-level topology evolution model is proposed. The actual impact of factors, such as. router load and limitations of router hardware are considered into the model, from the point of view of network performance optimization, router nodes and network traffic transmitted between nodes are taken into consider together. Through network characteristics contrast between a large number of simulation results and the same scale actual network, it shows that the model can simulate the basic characteristics of the actual router network topology.(3) Research of autonomy system-level Internet topology evolution model based on Multi-subnet Composited Complex Network Model. Autonomy system node types, the growth and aging of the network, local-world characteristics must be taken into account the actual impact of autonomy system-levellntemet evolution. A Internet autonomy system-level topology evolution model is proposed. According to their respective functions, autonomy system nodes in the Internet are divided into two categories:service provider node and service customer node, the relationship of line between autonomy system nodes in the Internet are divided into two categories:service provider to service customer relationship and peer to peer relationship. The model is based on the classic BA model, the location, price and service quality act as impact factor of the preferred attachment mechanism, and different strategy of establish connection is used for two different types of nodes, network aging is reflected by deleting nodes and edges. Through theoretical analysis and network characteristics contrast between200times simulation results and the same scale actual network, it shows that the model can simulate the basic characteristics of the actual autonomy system network topology and effectively and accurately portray the actual autonomy system-level Internet network characteristics.(4) Research of cascading failure based on Multi-subnet Composited Complex Network. Cascade failure is an important symbol of network vulnerability. A cascading failure ploy with an adjustable parameter based on Multi-subnet Composited Complex Network Model is proposed, and a new measure of network robustness is introduced. Through simulation and theoretical analysis, the ploy can effectively estimate the threshold of the network generated by cascading failure, effective recommendations for network topology optimization under limited resources and for suppressing network congestion are provided. The research can guide the actual network topology and load distribution for achieving the purpose to improve network robustness. The cascading failure model can be applied to a variety of typical complex network topologies, such as ER random networks, small-world networks, scale-free networks, the simulation results show that network topology generated by router-level Internet topology evolution model based on Multi-subnet Composited Complex Network Model are more robust than the three typical complex network topology.Autonomy system-level and router-level Internet topology evolution model based on Multi-subnet Composited Complex Network Model proposed in this dissertation solve the problem that existing Internet evolution models can not describe different kinds entities and their relationships in Internet, reflec the method that Internet evolution is jointly decided by different kinds entities and their relationships in Internet. They can reveal the evolution mechanism of Internet, and the network topology characteristics generated by evolution models are consistent with the same scale actual network.