Measurement And Identification Of The Network Topology

With the rapid development of Internet, new Internet services and applications emerge, which makes the network scale gradually expanding and expanding. Large-scale network performance monitoring and analysis, is a new field in the study of internet. Network performance prediction and infer most of the known network topology needs to be the case. Network topology can only understand the link to the overall situation of the network to give accurate judgments. Accurately predict the network topology is a network performance measurement goals and premise. Accurate and timely access to a wide range of application of network topology on network management and network security is particularly important. Network topology measurement as the basis for network measurement is an important modern network management system components. It is the status of the development of modern computer networks significantly.Traditional network topology identification method is based on an agreement of collaboration, information feedback through the intermediate nodes to detect network topology information, access to connections between nodes. However, with the continuous expansion of network scale and security have become increasingly demanding, requiring coordination between the nodes become more and more difficult to identify the limits of the traditional method of application of the network topology. Chromatography technique based on network topology identification method does not require cooperation of internal nodes, which measured parameters through statistical analysis of end to end to speculation within the link performance and topology, which is the network monitoring and management of new technical means. This paper from the identification of network topology measurement method and implementation mechanisms are beginning to study the topology of traditional measurement methods and topology-based identification method of tomography, focused on large-scale network router level topology discovery. Its research and innovation are as follows:First, analysis based on tomographic imaging technology, the network topology identification method is mainly studied based on end-to-end measurement tomographic imaging technology, and the topological reconstruction is presented based on the end-to-end measurement improved topology identification algorithms. It used improved "sandwich" method to get the difference of RTT to infer the correlation. As a result, it reduced the cost of setting agents on every terminal node and the complexity of data collection. In terms of topology inference algorithm, the method refers to the tree depth-first sequence, through the junction compared to the similarity to reverse build network topology. It only needed to send less probe packets and improved the accuracy of topology identification. Then NS simulation was adopted after the light load on the network and the network under heavy feasibility analysis of the algorithm.Second, considering the current network heterogeneity and authority limits, making cooperative based on the internal nodes is very difficult to identify the traditional topology commonly used in large-scale networks; tomography-based identification method assumes that all intermediate routers do not collaborate, increase the complexity of the measurements. The actual collaboration network has no nodes, but also interoperable nodes; paper presents a first pass using the traditional methods of measurement TRACEROUTE partial topology network topology, end to end after the measurement by improving targeted to identify incomplete topology section, and some incorrect to be amended. The combination of the traditional measurement and tomography topology identification method can speed up the recognition speed, further reducing the amount of probes sent, topology recognition accuracy is also higher. Finally, the NS simulation show that topology identification method is feasible.