High performance payload signature-based Internet traffic classification system

Abstract

Internet traffic classification is an essential step for stable service provision and efficient network management. The payload signature-based-classifier is considered as a reliable method for Internet traffic classification, but is prohibitively and computationally expensive for real-time handling of large amounts of traffic on high-speed network. To solve this problem, most studies focused on the pattern matching algorithm or hardware-based approaches such as FPGA and network processor. However, in order to improve the performance of the classification system, It is also necessary to consider the classification criteria and signature model in accordance with the characteristics of various application protocols. In this paper, we newly define the classification criteria and signature model, and propose an optimized classification architecture in perspective of input data minimization and complexity of pattern matching algorithm to improve the processing speed of classification system. Each of them can be applied individually, or in any combination. The proposed method achieved an approximately 5-fold increase in processing speed over existing baseline classification system.