Your browser version is outdated. We recommend that you update your browser to the latest version.

Manuscript Title: Intrusion Detection System Using Unsupervised Immune Network Clustering with Reduced Features

Author : Murad Abdo Rassam, Mohd. Aizaini Maarof , and Anazida Zainal

Email : murad2009@gmail.com, anazida@utm.my, aizaini@utm.my

Abstract: Intrusion Detection Systems (IDS) are developed to be the defense against security threats. Current signature based IDS like firewalls and anti viruses, which rely on labeled training data, generally cannot detect novel attacks. The purpose of this study is to enhance the detection rate by reducing the network traffic features and to investigate the feasibility of bio-inspired Immune Network approach for clustering different kinds of attacks and some novel attacks. Rough Set method was applied to reduce the dimension of features in DARPA KDD Cup 1999 intrusion detection dataset. Immune Network clustering was then applied using aiNet algorithm to cluster the data. Empirical study revealed that detection rate was enhanced when most significant features were used to represent input data. The finding also revealed that Immune Network clustering method is robust in detecting novel attacks in the absence of labels.

Keywords: Feature Reduction, Artificial Immune Network, Intrusion Detection System.

Vol 2 (3)