SVM Based Network Traffic Classification Using Correlation Information

Abstract

Traffic classification is an automated process which categorizes computer network traffic according to various parameters into a number of traffic classes. Many supervised classification algorithms and unsupervised clustering algorithms have been applied to categorize Internet traffic. Traditional traffic classification methods include the port-based prediction methods and payload-based deep inspection methods. In current network environment, the traditional methods suffer from a number of practical problems, such as dynamic ports and encrypted applications. In order to improve the classification accuracy, Support Vector Machine (SVM) estimator is proposed to categorize the traffic by application. In this, traffic flows are described using the discretized statistical features and flow correlation information is modeled by bag-of-flow (BoF). This methodology uses flow statistical feature based traffic classification to enhance feature discretization. This approach for traffic classification improves the classification performance effectively by incorporating correlated information into the classification process. The experimental results show that the proposed scheme can achieve much better classification performance than existing state-of-the-art traffic classification methods.