Traffic classification beyond application level: identifying content types from network traces

Yipeng Wang, Zhibin Zhang, Li Guo

Abstract: Prior works on traffic classification mainly focus their attentions on dividing Internet traffic into different categories based on their application layer protocols (such as BitTorrent, eDonkey etc.). Making traffic classification from another point of view, we divide Internet traffic into different content types. Our technology is an attempt to solve the classification problem of unknown and proprietary protocols. In this paper, we design a classifier which can distinguish Internet traffic into different content types using machine learning techniques, and features are the entropy of consecutive bytes and frequency of characters. The chief features of our classifier are high classification accuracy (about 81%) and small computing space (about 1K Bytes).