APPLICATION OF SUPPORT VECTOR MACHINE AND CONTEXTUAL OUTLIERS FOR INTRUSION DETECTION IN THE SCADA SYSTEM

Abstract

In this paper, we present an IDA-SCADA model based on Support Vector Machine (SVM) which is capable of detecting intrusion into SCADA systems with high accuracy. The distinction of our method used in this research is we applied contextual training data. To do that, the original dataset was reorganized to create context before training the SVM phase. The result of our work is the proposed system able to identify any attacks or normal patterns with precision from 95.02% to 99.03%.