A hybrid method for network traffic classification

Abstract

In response to the growing requirements of traffic classification for increasing complex network environment, this paper introduces a hybrid method for network traffic classification. By combining port-based, signature string matching, regular expression matching and machine learning methods, our method can achieve high speed and accurate traffic classification. Moreover, a typical application of our method is proposed to identify encrypted traffic in high performance, which achieves 96.0% average accuracy. The experimental results show that our proposed method is able to achieve over 95.0% average accuracy for all experimental traces.