Feature reduction-based dos attack detection system

Abstract

Denial-of-service (DoS) has still been popularly used by attackers. It can be seen in China and the USA who are major victims of DoS attacks in recent years. For this reason, the development of an intelligent intrusion detection system (IDS) remains a challenging task. This study proposes a system for the detection of DoS attacks with feature reduction using a rule-based PART classifier. The reduced feature set is identified based on the combination of information gain and correlation attribute evaluation methods. The system is implemented and tested on CICIDS 2017 dataset. Finally, the proposed system provides an accuracy of 99.9871% for the detection of DoS attacks with 56 reduced features.