Anomaly Detection on Intrusion Detection System Using CLIQUE Partitioning

Abstract

The development of information and network technology makes network security become important. Intrusion is one of the issues in network security. To prevent intrusion happens, intrusion detection system (IDS) is built. One of IDS category is anomaly detection. This category detects intrusion event based on data profile. Clustering is one way to observe data profile. There are a lot of clustering algorithms proposed for anomaly detection on IDS, but most of them find clusters in the highest dimension of data. CLIQUE Partitioning (CP) is one of the clustering algorithm that can find clusters from the subspace of data. Testing is done to analyze system's performance based on computational time, completeness, and false alarm rate. CP algorithm shows good performance from completeness point of view (94.59%) and false alarm rate (2.54%). From computational time, CP shows good performance based on the amount of tuple, but the performance is not too good from the quantity of feature side.