Real-time Traffic Classification Algorithm Based on Hybrid of Signature Statistical and Port to Identify Internet Applications

Internet traffic classification gained significant attention in the last few years. Most of the current classification methods were only valid for offline classification. The three common classification methods i.e. port, payload and statistics based have some limitations. This paper exploits the advantages of all the three methods by combining them to produce a new classification algorithm called SSPC (Signature Statistical Port Classifier). In the proposed algorithm, each of the three classifiers will individually classify the same traffic flow. Based on certain priority rules, SSPC makes classification decisions for each flow. The SSPC algorithm was used to classifying four types of Internet applications in two stages, initially offline and later online. The results of both cases show that SSPC is the higher accuracy when compared with other classifiers. In addition, as demonstrated in the real time online experiments done, SSPC algorithm uses a short time to classify traffic and thus it is suitable to be used for online classification