Author(s): Kyoung Won Min, Young Hwan Choi, Joong Hoon Kim
Keywords: Cyber-Physical System; Cyber-Attack; Water quality; Detection Algorithm; Deep learning;
Abstract: Recently water distribution systems (WDSs) operation is applied based on cyber-physical systems (CPSs) for the efficient system operation and maintenance. Since CPSs are operated based on information and communication technology, it is exposed to the risk of cyber-attack which leads to a disruption in the operation of the WDSs such as water supply reduction, water pollution. Therefore, a few studies were proposed cyber-attack detection algorithms based on the various cyber-attack scenarios. These approaches considered only hydraulic factors (i.e., pipe velocity, nodal pressure) generated by WDSs components, and a statistical method or Deep-learning based model was applied as a detection method. However, the algorithm which considers only hydraulic factors are cannot prevent the various abnormal condition such as water quality problems. Since the hydraulic and quality factors have a correlation, it cannot be considered individually. Therefore, in this study, a framework was developed considering not only hydraulic factors but also water quality factors simultaneously. The proposed approach is applied to the deep learning-based model and the statistical approaches as a cyber-attack detection algorithm. By developing the cyber-attack detection algorithm in WDSs as CPSs considering hydraulic and water quality criteria to meet the purpose of WDSs, it is more efficient and realistic to be able to organize more scenarios that can occur when a cyber-attack occurs and it can be detected immediately and respond to the problem. Also it can contribute to the establishment of safe water supply infrastructure throughout the whole process of designing and operating WDSs as CPSs.