An Ensemble Approach for Intrusion Detection in Collaborative Attack Environment

Data confidentiality and integrity are the main issue in modern era of Internet. Network security is compromised by intruders with help of attacks. Collaborative attack is a new attack used by intruder which is to be monitored by intrusion detection system (IDS). In this paper, we are proposing IDS which can handle collaborative attack with the help of ensemble classifiers. Proposed IDS makes use of various feature selection techniques to reduce number of feature required for attack detection. Various classifiers are used in ensemble to increase attack detection rate and reduce false alarm rate (FAR). Dataset used for experiments is UNSW NB15. Comparison of individual classifier is done with ensemble approach. Results show reduction of 5% in FAR and improvement of 2% in accuracy for ensemble approach. Ensemble approach and feature selection provide improved performance of IDS in terms of precision and recall.