Network Topology Inference Research Based On Packet Loss Rate

Nowadays, computer network has gradually become the pillar of our information society, and the existing problems in the network are emerging, especially the detection and maintenance for large-scale network in the field of network security. Because there are a number of factors for causing the poor condition of the network in order to detect the cause of the adverse conditions produced, the network administrator must probe the network elements according to the topological structure to collect relevant data. But the network scale expands, the network administrator can’t direct access to the network elements, and the traditional topology inference method has been unable to meet the need of administrator to network elements. While they use the new technology—Network Tomography to conduct the network topology inference, under the condition of without collaboration of the network equipments, this technology only relies on the performance parameters obtained at the terminal device, and use the correlation of sibling nodes to determine the relationship between the known network equipments, then infer the connection of the internal network equipments, using this method can make up the shortage of traditional network topology inference method, so the thesis will have a research of the application of network tomography technology in network topology inference direction.The thesis first introduces the background, significance, measurement, principle and inference model of NT technology research.Secondly introduces the network topology inference methods, this chapter is the main content of the thesis, analyzes the limit of the application of traditional network speculation method and the advantage of network topology inference method based on the technology of NT. The inference method based on NT technology in the thesis uses network packet loss rate as the parameter of the inference, uses multicast measurement method to obtain a terminal node packet loss rate, analyze and compare the existing multicast network topology inference algorithm of advantages and disadvantages, On this basis, putting forward a more optimal topology inference algorithm—HTIP algorithm, the algorithm calculates the hamming distance between two terminal nodes on the base of node packet loss, the characteristic of the large similarity between siblings nodes is used to have a reasonable topology inference.Then introduces the simulation principle of NS2network simulation software, using NS2simulation multicast network, contracting out the measurement for the node performance parameters, and analyzes BFHC and HTIP algorithms and the existing algorithm, the results prove that HTIP algorithm has better performance. At the end of the thesis, introducing the application of information visualization of topology inference based on NT technology and current layout algorithms for the information visualization, and choosing the appropriate storage method for network equipment and link, having a dynamic display for topology inference process visualization, using c#language to program for the visualization software, finally show the topology inference process in a neat, beautiful fashion.