The Internet Topology Discovery And Analysis Of Technical Studies

The knowledge relevant to Internet topology and topology quality is of significant importance to the management, expansion, performance enhancement, security and protocol optimization of the network. However, with the development of the Internet, the network is exploding and its structure is becoming more and more complex, thus it's rather difficult to acquire the topology of the network. Accordingly, the research on topology discovery is attracting more and more attention. Since the Internet is composed of many autonomous systems (Ases) and the interconnection among them reflects the macrostructure of the Internet, the research on the interconnection among ASes becomes a key in Internet topology discovery. The complexity of the network has gradually become a popular subject in network related research. It's proved that the seemingly disordered network in fact contains abundant disciplinarians and characteristics. Detailed research into these disciplinarians and characteristics can provide effective guideline for the development of the network and benefit the research on network modeling.This thesis first present the fundamental methods and relevant algorithms for topology discovery in the Internet, including the methods based on SNMP, interconnection protocol and those based on routing protocol. The research status and the characteristics of the interconnection among the ASes is also introduced.Then we focus on the technique of the topology discovery of the network at AS level and compare the advantages and weaknesses of several approaches in this category. After a comprehensive analysis of these methods, we present and implement an Internet topology discovery system and the system contains a topology generating module, an information processing module and a topology display module. To acquire the inherent topology properties and characteristics of the topology graph at AS-level, we introduce basic network model (Scale-free network and Small-world network) and evaluating parameters (such as cluster coefficient, average path length, degree distribution, Pearson correlation coefficient) in this thesis. Through thorough analysis of the topology of the generating topology at AS-level, we can see some apparent properties with the evaluating parameters and the inherent power-law distribution property of the derived topology. Besides, we discuss the perspective of the evolution of the network's topology based on these analyses. Finally we apply the classical BA model in complex network to generate a Scale-free network and then we analyze its topology property and compare it with that of the actual topology. The results show that the network generated by BA model has power-law distribution property, but it can't effectively describe some topological properties of actual network, such as cluster coefficient.