Research On Intelligent Traffic Classification For High-speed Network

Accurate identification and categorization of network traffic according to application type is an important element of many network management tasks such as flow prioritization, trafflc shaping/policing, and diagnostic monitoring.At the same time, with the development of network technology, Gigabit networks have become the mainstream of campus network. Many backbone networks have reached 10G. Therefore, in order to deploy security management strategy and flow control measures, efficient traffic classification for high-speed networks should be researched.This paper has proposed a new classification method- Further Division of Partition Space(FDPS). This method has improved neuro network classifier by further divide the partition space of data set which has been mapped by neuro networks. In addition to this ,in order to further enhance the performance,this paper also researched the combination of FDPS and flexible neuro tree(FNT). Experiment results have shown that this method, compared to traditional neuro network classifier, has favorable performance especially with respect to the optimization speed and the accuracy of classified samples. More over, the generalization of FDPS can be further improved by using the combination of FDPS and FNT.This paper also studied several mainstream traffic classification method. Based on the comparison, we decide to choose transport layer statistics as our research object. This method has overcome the limitations of port-based and payload-based analysis. This classification technique relies on the fact that different applications typically have distinct behavior patterns when communicating on a network. Traffic can be identified using these behavior patterns. We pursued transport layer statistics, and utilized FDPS, and conducted experiments in the classic packet trace: Auckland IV. The experiments show that FDPS has favorable performance and prospect.However, if we want to extract statistics and identify traffic on high-speed networks, the processing speed of network equipment must be considered. Network processor is an efficient solution for high-speed network environment. It applies to each layer of network processing, combines the advantages of ASIC and RISC, increases flexibility and lowers development costsBased on these studies, a wire-speed traffic classification system, which uses transport layer statistics, is designed. The core classifier which is used in this system, is the combination of FDPS and FNT(FDPS+FNT).The hardware platform used in this system is IXP2400,deigned by Intel. Experiment results show that this system can identify traffic on high-speed network environment.