علي عابدي

This thesis studies a new method to detect cyber-physical attacks and faults as well as distinguish them from each other. Multi-agent systems are a subset of cyber-physical systems in which some subsystems called agents exchange data with each other through a communication network that is involved in generating the control law of each agent. Due to the high resemblance between the effects of false data injection cyber-attacks and the effects of physical additive faults in the system, conventional detectors are unable to distinguish between the two or do so with low accuracy. Therefore, a different approach is presented in this thesis, which is able to distinguish between these cyber-attacks and the physical faults with high accuracy. To avoid unnecessary mathematical complexity, a discriminant method was first proposed for a networked control system. Then the developed method is generalized to homogeneous and linear time-invariant multi-agent systems. In the structure presented for this method, a filter called a watermarking filter is utilized, the input of which is the output data of the system. This filter with two residual signal generators is designed in such a way that one of the two residual signals is highly insensitive to cyber-attacks, but the other residual signal is sensitive to both faults and cyber-attacks in the system. In fact, the designed watermarking filter contains the basis in the null space of the attack signal to residual signal transfer matrix. With this design, according to the decision logic, false data injection attacks and physical faults of the system can be distinguished from each other. In the rest of this thesis, in order to eva‎luate the performance of the proposed method, simulations have been performed in both networked control systems and multi-agent systems. The simulation results demonstrate that all the objectives of this thesis, including the high accuracy of distinguishing between cyber-attacks and physical defects, have been achieved.

تمييز عيب و حملات سايبري , حملات تزريق داده‌ي غلط , شناسايي عيب , تشخيص حملات سايبري , سيستم‌هاي چندعاملي , سيستم‌هاي سايبري-فيزيكي

Fault and Cyber-Attack Discrimination , False Data Injection Attack , Fault Diagnosis , Detecting Cyber-Attack , Multi-Agent Systems , Cyber-Physical Systems

Ali Abedi

Dr. Jahed