State-Based Modeling and Anomaly Detection in Industrial Systems Using DEVS

The Industrial Revolution changed the picture of industrial systems by making them highly efficient and more manageable. The inclusion of the internet, smart devices, and various protocols has collectively developed Industrial Control Systems (ICS), which are responsible for the entire industrial activity management. However, the devices in ICS may also malfunction and show abnormal behavior, affecting industrial activities. Hence, it is necessary to have a good abnormal condition detection system in ICS, but the presence of a real environment for developing such a detection system is challenging. Therefore, the proposed work addresses this challenge by modeling the ICS to mimic the actual behavior of the industries and collecting the relevant data to train a model for detecting anomalous behavior in the system. The work employed Discrete Event System Specification (DEVS) as a modeling and simulation formalism for designing the ICS. Through this modeling system, the work collects the states associated with every device and trains a machine-learning model to deal with anomalies in ICS further. This work also aims to show the significance of the DEVS formalism in modeling the ICS and how its state-based data may be used further to make an artificial intelligence-oriented solution.