Online automatic traffic classification architecture in access network

Recently traffic classifications based on statistics methods and machine learning techniques have attracted a great deal of interest. Some challenging issues for these methods are that most of them need prior analysis to detect traffic applications and training data sets to generate classification model offline; some require a high amount computation and memory resource. These are infeasible to cope with the fast growing number of new applications and online traffic classifications. We propose an online automatic traffic classification architecture using unsupervised machine learning technique, in which flows are automatically clustered based on sub-flow statistical features instead of full flows. We select best-first features algorithm to find an optimal feature-sets which is suited for access network, then map the traffic flows to applications based on maximized probabilities applications in the clusters. The experiment results demonstrate the efficiency and capability of the proposed automated classification architecture.