Performance Analysis of Unsupervised Machine Learning Techniques for Network Traffic Classification

Network traffic classification is important for QoS, Network management and security monitoring. Current method for traffic classification such as port based or payload based suffered many problems. Newly emerged application uses encryption and dynamic port numbers to avoid detection. So we use unsupervised machine learning approach to classify the network traffic. In this paper unsupervised K-means and Expectation Maximization algorithm are used to cluster the network traffic application based on similarity between them. Performance of these two algorithms is compared in terms of classification accuracy between them. The experiment results show that K-Means and EM perform well but accuracy of K-Means is better than EM and it form better cluster.