Title: An integrated approach for network traffic analysis using unsupervised clustering and supervised classification

Authors: Kothandapani Chokkanathan; S. Koteeswaran

Addresses: Department of Computer Science and Engineering, Vel Tech Dr.RR & Dr.SR Technical University, India ' Department of Computer Science and Engineering, Vel Tech Dr.RR & Dr.SR Technical University, India

Abstract: Traffic analysis is a significant task to control the network traffic in heterogeneous manner. The unsupervised learning system fails to expand the supervised classification model for network analysis. Several data mining techniques employed to classify the network traffic pattern using unsupervised learning approach. However, the continuous evaluation of network traffic on multi-dimensional data is a difficult task. In order to solve this issue, an integrated K-means unsupervised clustering and supervised C4.5 Classification (KUC-SC) technique is designed. At first, K means unsupervised learning algorithms is applied in KUC-SC technique to form a 'k' number of clusters. After that, C4.5 is used to classify the data whether it is real-time or non-real-time traffic through the construction of decision-tree. An experimental result shows that the KUC-SC technique obtains better performance in terms of classification accuracy, classification time, true positive rate and communication overhead compared to the state-of-the-art works.

Keywords: integrated approach; KUC-SC; traffic analysis; decision-tree; data mining; multi-dimensional data; unsupervised learning system; traffic pattern; real-time traffic; non-real-time traffic; network security.

DOI: 10.1504/IJITST.2019.102797

International Journal of Internet Technology and Secured Transactions, 2019 Vol.9 No.4, pp.517 - 536

Received: 11 Jul 2017
Accepted: 26 Jul 2017
Published online: 08 Oct 2019 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article