Design of Anomaly Detection System for Outlier Detection in Hardware Profile Using PCA

In this paper, we design an Anomaly Detection System for Outlier Detection in Hardware Profile by using Principal Component Analysis (PCA) that helps reduce the dimension of data. Anomaly detection methods can detect new intrusions, but they suffer from false alarms. Another approach is misuse detection that identifies only known attacks by matching with the previous patterns. Host based Intrusion Detection Systems (HIDSs) use anomaly detection approach to identify malicious attacks i.e. intrusion. Data being of large dimensional generates features in terms of large set of dimensions and hence the system takes considerable time for processing the huge amount of data. The PCA is used to reduce the dimensionality of the host based data without any loss of useful information such as non-redundant data. We experimentally show that the proposed intrusion detection system has detection rate in the range of 90% 97.5% and false alarm rate in the range of 2.5% 7.5% depending upon the major and minor principal components.