Detection of Cyber Attacks in IoT Enabled Cyber Physical Systems

Ensuring security for Cyber-Physical Systems (CPS) with Internet-of-Things (IoT) capabilities stands as a significant challenge, as current security solutions employed for traditional IT and OT, fails to preserve many of the Cyber Physical Systems unique design along with dynamic state. IoT gains traction in key industries like autonomous systems, manufacturing, healthcare, smart cities, and so on, cyber threats arise at an exponential rate especially. Even more, these systems are being subjected to many attacks (DoS, unauthorized access, adversarial command injection, and data manipulation) that can risk functions, safety, and data integrity. This project works to fulfill an identified gap in market demand for intelligent cyber security components for IoT-enabled CPS that require developing a ML dependent method for identifying cyber threats to CPS. Proposed method will implement ML approaches to utilize the real-time information collected from IoT sensors and actuators and networking logs, in order to identify system behavior and anomalous patterns for identifying cyber-attacks. Our proposed system will implement various advanced machine learning algorithms using, for example, Deep Neural Networks (DNN), Auto Encoder and Decision Trees, coupled with Principal Component Analysis (PCA) methods etc. As such, our system will have capable of accurately detecting and categorize an abundance of cyber-attacks on CPS as compared to alternative methods.