ELLIOTT is developing a fast, reliable and robust intrusion detection model that can be applied to industrial processes.
Industrial control systems are evolving from isolated systems to those that use IoT sensors, Edge computing and wireless networks. This evolution improves performance and supports cost-effective production. However, it also increases the scope for attacks aimed at industrial espionage and sabotage. It is critical to design fast and reliable intrusion detection systems that are able to detect changes in the process. These systems also need to be robust to changes such as the aging of devices.
The project will investigate deep learning along with conventional machine learning models to develop a fast, reliable and robust detection model that can be applied to various industrial processes. The project has previously developed an AI-approach which uses evolutionary algorithms as a mechanism to generate attack examples to identify weaknesses in the detection. To test the performance of the detection models, ELLIOTT will test the developed AI-approach in conjunction with manual random attacks and off the shelf tools. To show the work in a wider context, ELLIOT will work with user partners to test the developed approaches in two real cyber physical systems within industrial control systems: i) electronic servo motors in factories and ii) building management systems.