Infrequent Pattern Identification in SCADA Systems Using Unsupervised Learning
Mohiuddin Ahmed · Advances in information security, privacy, and ethics book series · 2016
In recent years, it has been revealed that these critical infrastructures such as SCADA systems have been the target of cyber-terrorism. In general cyber-attacks are infrequent in nature and hence infrequent pattern identification in SCADA systems is an important research issue. Therefore, design and development of an efficient infrequent pattern detection technique is a research priority. In this chapter, the effectiveness of co-clustering which is advantageous over regular clustering for creating more fine-grained representation of the data and computationally efficient is explored for infrequent pattern identification in SCADA systems. A multi-stage co-clustering based infrequent pattern detection technique is proposed and applied on seven benchmark SCADA datasets which includes practical industrial datasets. The proposed method shows its superiority over existing clustering based techniques in terms of computational complexity which is essential for practical deployment in a SCADA framework.