Research And Design Of Real-Time Traffic Classification For High-Speed Network

Next Generation Internet provides users with a rich and varied user experience, meanwhile, it has brought great challenges to network managers due to its huge network traffic. Traffic monitoring is the basis to ensure the controllability of the network. Nowadays, the equipments with lOGbps transmission speed have been broadly applied, and the equipments with 40Gbps will also appear in the backbone network, which gives higher requirements to traditional network traffic monitoring system. Nowadays, the research on traffic classification technologies is still far from the requirement of the development of business. It is mainly because the most of technologies today apdopts offline classification techniques, which cannot realize real-time monitoring. This paper mainly focuses on the study and design for traffic classification system in high-speed network, which is able to realize real-time accurate classification for applications in high-speed network and effectively control NGI traffic.This paper deeply studies traditional traffic classification technology and data mining based traffic classification method, and introduces a data mining based traffic classification technology. Moreoever, this paper also applies sampling technology to data packet collecting mechanism so as to reduce the burden of traffic classification system without affecting the accuracy of the classification. Meanwhile, this paper applies VFDT algorithm to real-time traffic classification, and analyzes how to apply VFDT algorithm on real-time traffic classification system in details. Meanwhile, we define the important characteristics of real-time attributes, and introduce real-time attribute set, which is suitable fit for the proposed traffic classification system. At last, according to experiment results and statiscally analysis under different situations, we intoduces how to use VFDT algorithm depending on the requirement of real-time traffic classification system. On the other hand, we fully describe classification accuracy of this traffic classification system through selecting different VFDT parameters and classification granularities.