Application traffic classification using statistic signature

Networks today are becoming more complex and diverse because of the appearance of new applications and services. The importance, therefore, of application-level traffic classification is increasing daily. Application-level traffic classification has become a very popular area of study. Although many proposals have been presented, including port-based, payload-based and machine learning-based methods, the method that can manage all traffic has not yet been developed. More recently, methods based on statistical flow information have been studied. In this paper, we propose an application-level traffic classification methodology using the statistic signature. Our method creates a statistic signature using payload size, transmission order, and direction of the first N packets in the flow, and uses this to classify application traffic. Then, using a verification system, we prove the feasibility of our method and show its high accuracy.