Attack Target Detection Using Machine Learning on SCADA Gas Pipeline Data

Supervisory Control and Data Acquisition (SCADA) architecture is at the core of industrial organizations, such as factories and power plants. Within the SCADA system is a network of devices gathering and monitoring data. As the number of devices or components increases, so does the number of potential attack points. It is crucial to prevent and mitigate attacks on SCADA systems in order to stop a complete loss of the system to an attacker. Detecting the specific target point of the attack will allow for quicker isolation and mitigation without requiring the shutdown of the whole system. This paper explores the use of machine learning models, particularly Random Forest (RF), Gradient Boosting Classifier (GB), AdaBoosted Classifier (AB), a Deep Neural Network (DNN), and other previously proposed models, comparing their performance in identifying the attack target in a SCADA environment for gas pipeline data.