Research On Network Topology Inference Technology Based On Multi-parameter

Now, as the network scale expands unceasingly, at the beginning of the Internet which from building small user groups, centralized control methods, use of limited small network,the Internet grew into a complex network structure, distribution, collaboration, heterogeneous large interconnected networks. Therefore, it needs a steady, robust, free manipulation of the network environment. Research, so, internal topology structure is imperative. Accurately infer the internal topological structure is beneficial to the people managing, optimizing the operation environment.Medical topographic imaging extremely similars with network tomography, without internal node(router) cooperation, the active probe packets sent from end to end, and topographic imaging method is used to estimate the network internal performance parameters.The best advantage of this method is that it can complete internal performance parameters measurement to the network under the condition of no internal nodes. Therefore, when it put forward had been widely attention in the filed of academia and industry at home and abroad,and it becomes one of the most important research content in the field of network measurement.In this paper, using the network tomography methods, internal performance parameters estimated network, we study how to detect package. Hierarchical clustering method is used to infer. We obtained valuable results in the following aspects:(1) On the basis of the original proposed a new calculation method, it based on the information fusion of multi-parameter.The original calculation method including the success rate, queuing delay, delay covariance,delay jitter covariance. Before using this method need to use standardized formula for data conversion, and then in the information fusion.(2) Based on the hierarchical clustering inference algorithm, we proposed a algorithm inference general tree by a threshold. This algorithm selects the most likely two terminal nodes as their brother. Because when they are not brother, they at least have a link to separation. Based on this you can judge other siblings.Run the entire program cycling, until in the terminal nodes’ collection with only one node.(3)For network topology visualization work, we use the tree layout algorithm for layout, and use the LCBT, LCGT algorithm inference binary tree and the general tree, use tree layout algorithm to display on the canvas. To use NS2 simulation software for simulation experiment.By comparing the simulation results, in this paper, the method put forward is proofed validity and rationality.