Gradual increase in the number of successful attacks against Industrial Control Systems (ICS) has led to an urgent need to create defense mechanisms for accurate and timely detection of the resulting process anomalies. Towards this end, a class of anomaly detectors, created using data-centric approaches, are gaining attention. Using machine learning algorithms such approaches can automatically learn the process dynamics and control strategies deployed in an ICS. The use of these approaches leads to relatively easier and faster creation of anomaly detectors compared to the use of design-centric approaches that are based on plant physics and design. Despite the advantages, there exist significant challenges and implementation issues in the creation and deployment of detectors generated using machine learning for city-scale plants. In this work, we enumerate and discuss such challenges. Also presented is a series of lessons learned in our attempt to meet these challenges in an operational plant.

A method referred to as PbNN is proposed to detect cyber-physical attacks through the identification of resulting anomalies in the process dynamics of the underlying ICS. Unlike existing anomaly detectors based on an abstract knowledge acquired from operational data, PbNN utilizes the design knowledge of ICS to learn the complex relationships among the correlated components. Such relationships are accurately modeled using operational data through the application of the deep convolution neural network. The proposed detector was implemented and evaluated in an operational secure water treatment plant by launching several real-time stealthy and coordinated attacks. The results indicate that PbNN outperforms the existing state-of-the-art machine learning anomaly detectors when compared using detection accuracy and the rate of false alarms.