Clustering Based Outlier Detection Method for Network Based Intrusion Detection

The discovery of objects with exceptional behavior is an outstanding challenge from a knowledge discovery standpoint and has received considerable attention in many applications such as network attacks, fraud detection. This paper proposes a simple clustering based algorithm to detect outlying objects. The main problem for network intrusion detection system is the ability to exploit ambiguities in the traffic stream. Network-Based Intrusion Detection monitors network traffic for particular network segment and analyzes the network and application protocol activity to identify suspicious activity. There are several recently developed outlier detection schemes to detect attacks in a network. In this paper, the proposed algorithm is applied to network intrusion detection system to detect ambiguities or violations in the network traffic stream.