Dynamic Watermarking based defense of Closed Loop Cyber-Physical Systems

With increasing automation and control, the risk of chemical or other processes being cyber attacked by malicious agents has increased significantly. Such attacks need to be detected immediately to ensure that process performance and safety is not compromised. Various fault detection and diagnosis approaches are available in literature to detect process abnormalities and can potentially be used to detect such attacks. However, the performance of such approaches may not be adequate. Recently (in literature), an enhanced approach for detecting such attacks has been proposed. Labeled dynamic watermarking based approach, it involves deliberate inclusion of a random perturbation in the manipulated variable. The deviations in the plant measurements from the predicted values of the measurements are then analyzed to detect an attack. The current work will investigate the dynamic watermarking based approach for processes with closed loop control as well as consisting of noisy measurements.

Scope of the project: Computational, Theoretical, and Experimental.