Effective behavior signature extraction method using sequence pattern algorithm for traffic identification

Abstract

With the rapid development of the internet and a vigorous emergence of new applications, traffic identification has become a key issue. Although various methods have been proposed, there are still several limitations to achieving fine-grained and application-level identification. Therefore, we previously proposed a behavior signature model for extracting a unique traffic pattern of an application. Although this signature model achieves a good identification performance, it has trouble with the signature extraction, particularly from a huge amount of input traffic, because a Candidate-Selection method is used for extracting the signature. To improve this inefficiency in the extraction process, in this paper, we propose a novel behavior signature extraction method using a sequence pattern algorithm. The proposed method can extract a signature regardless of the volume of input traffic because it excludes certain unsatisfactory candidates using a predefined support value during the early stage of the process. We proved experimentally the feasibility of the proposed extraction method for 7 popular applications.