Traffic classification using an improved clustering algorithm

Abstract

Accurate classification of Internet traffic is used in many fields such as network planning, network design, network management and monitoring of Internet traffic. In this paper, we apply an unsupervised machine learning approach based on clustering by exploiting the characteristic of applications. This approach uses an improved K-Means clustering algorithm named as I-K-Means. I-K-Means uses a transcendental initial value for K and assigns an individual weight value for each feature of the cluster. The results of the experiments show that I-K-Means has better performance than generic K-Means.