Novelty detection system based on multi-criteria evaluation in respect of industrial control system

Abstract

The industrial processes and systems have become more sophisticated and also adopted in diverse areas of human activities. The Industrial Control System (ICS) or Internet of Things (IoT) have become essential for our daily life, and therefore vital for contemporary society. These systems are often included in Critical Information Infrastructure (CII) which is crucial for each state. Consequently, the cyber defense is and will be one of the most important security field for our society. Therefore, we use the novelty detection approach in order to identify anomalies which can be a symptom of the cyber-attack in ICS environment. To achieve the main goal of the article One-Class Support Vector Machine (OCSVM) algorithm was used. Moreover, the anomaly detection algorithm is adjusted via multi-criteria evaluation and classifier fusion. © 2019, Springer International Publishing AG, part of Springer Nature.