Flow-based attack detection and accommodation for networked control systems

This paper is concerned about detection and estimation of malicious attacks on the network and the linear physical system of a networked control system (NCS) by using linear matrix inequality (LMI)-based technique. Certain class of attacks on the communication network impacts the traffic flow causing network delays and packet losses to increase which in turn affects the stability of the NCS. Therefore in this paper, a novel observer-based scheme is proposed to capture the abnormal traffic flow at the bottleneck node of the communication network via the attack detection residual. An LMI-based design is proposed that ensures both system stability and H-infinity performance and also detects attacks on the network as well as on the physical system. Upon detection, the physical system is stabilised by adjusting the controller gains provided certain conditions are met. Both simulation and hardware implementation results are included to demonstrate the applicability of the proposed scheme.