ABSTRACT
An Intrusion Detection System (IDS) seeks to identify unauthorized access to computer systems’ resources and data. Many methods have been propose to detect intrusion; for example, the pattern matching method to finding intrusion by matching sample data to know intrusive patterns and the statistical approach to detecting intrusion from normal activities at the user level or system level. In detecting processes, it usually will be obtained a large original data set. We investigated a principal component analysis method for intrusion detection in separating intrusions from normal activities by analyzing 5 system calls occurring on a host machine. The method generates a new set of variables. This method attempts to reduce the dimensionality of the data hopefully only 2 of the principal components will represent most of the variation in the data.
Keywords: Intrusion Detection, Principal Component, System Call
1. INTRODUCTION
The methodology of intrusion detection can divide in two-category [1]: anomaly intrusion detection and misuse intrusion detection. Anomaly intrusion detection refers to detecting intrusion base on anomalous behaviour of the attackers. Therefore, the distinction by categorizing the good or acceptable behaviour is very important. In the anomaly detection method, a statistical approach [2, 3, 4] and neural net approach are usually taken.