Traffic Network Analysis And Classification Research

With the development of Internet, detection of network flow has become the important way of making use of network resources efficiently. Traffic analysis focuses on solving the problem of performance optimization in the operable network. It adopts many technologies and rules to scale, represent, characterize and control the traffic flow in order to transfer the data reliably and rapidly via network and make use of the network resources more efficiently.Based on the" Network security monitor technology and systems "project of National high-tech research development plan (863 plan), major in the network monitoring system, network traffic analysis and network traffic classification technology. Analysis aiming at network flow contributes to the basis network of national security, the pivotal network infrastructure and safe running of important information system on open net. Network traffic classification and disjunction technology provide premise of researching for dealing with emergency network security incidents, IP address trackback and forecasting of large-scale network security.The paper first introduce general analysis technology of self-similar traffic flow in detail, then, discusses how to make use of data mining technology in the offline analysis of traffic flow data. In this part, the author focuses on the mining associate rules from traffic flow data. By analyzing the relationship between different kinds of traffic flow or the different attributes of one kind of traffic flow by the association rules, users can predict the distribution and development of traffic flow. The paper makes improvement on the old PRAR (Path Restricted Association Rule) and compares the performance of the two algorithms. It makes the network managers predict the development trend of network traffic flow. Following that, we expound the significance of the traffic classification and existing classification, and according to the study of the status at home and abroad put forward a traffic flow classification method used Vector Quantization and Gaussian Mixture based on packet train length and packet train size. At last, we summarize the two reach work.