Secure Industrial Control Systems (ICSs)
Project (7): Secure Industrial Control Systems (ICSs)
Sensors, IIoT, and CPSs can transform traditional manufacturing to smart manufacturing with focus on the increasing digitization and interconnection of products, value chains, and business models. Both IIoT systems and Industrial Control Systems (ICSs) are vulnerable to a variety of attacks. In this project, the REU students will study the following research topic.
Data-driven Early Detection of Intrusion. Different from general Information Technology (IT) systems, the states of ICSs change dynamically following physics laws. This ICS nature poses challenges to intrusion detection in ICSs because the dynamic behavior has to be precisely tracked to detect potential intrusions in real time. The predictor is a physical model that needs to be formed previously. Normally, the predictor can be built using linear system identification techniques, if sufficient operation data is available. Most of the existing intrusion detection systems assume that the ICSs can be modeled as discrete time memory-less Linear Time Invariant (LTI) systems. In real world, many ICSs have nonlinear behavior and memory effect. The REU students will design and examine an intrusion detector for a nonlinear ICS with memory. Simulated data will be used to identify the predictor using a model other than LTI system (e.g., neural network based), and few detection techniques will be adopted to generate alerts.
Qualifications: Matlab, c/c++, microcontroller programming, EE, CmpE or CmpSci major, knowledge of encryption.
Mentor: Dr. Guo (firstname.lastname@example.org) and TBD (TBD)