Mitigating Denial-Of-Service Attacks In Wide-Area Lqr Control

In this work we study the impacts of Denial-ofService (DoS) attacks on the cyber-physical implementation of wide-area control. The controller is desgined as a Linear Quadratic Regulator (LQR) for damping power flow oscillations. We first derive a nominal mathematical model of the power system network considering a delay- aware LQR controller without any DoS. Thereafter, we model how DoS enters the closed-loop dynamics of this system by considering the Hadamard product of the LQR gain matrix with a matrix of attacked links. We propose two mitigation strategies to compensate for the loss of missing states in the feedback path, and test their effectiveness against three variables of interest - namely, location of attack, time of attack, and duration of attack. We design a decision tree classifier for each strategy based on these parameters to predict the severity of an attack. The results are illustrated with a DoS simulation on the standard IEEE 4-machine Kundur model.