Stealth Attacks on the SADI with Prior Information on the State Covariance Matrix

OSVADY AVALOS-ABREU; FRANCISCO MESSINA; LEONARDO REY VEGA

San Juan

Conferencia; 2022 IEEE Biennial Congress of Argentina (ARGENCON); 2022

Instituto de Ingeniería Eléctrica - CONICET- Universidad Nacional de San Juan

False data injection attacks (FDIAs), in which an attacker has access to a number of meters and can corrupt their measurements, is a critical cybersecurity issue in smart grids. In this work, we consider the effectiveness of FDIAs against the Argentine Interconnection System (SADI). We consider a situation in which the attacker is able to influence a small number of devices. Inspired by \cite{kosut_malicious_2011}, the problem is examined from the point of view of the grid operator and the attacker. For the control center, we examine the performance of classical LNR and $\mathcal{J}(x)$ detectors against FDIAs. From the point of view of the attacker we consider the maximal estimation error that he can induce at the control center subject to a bound on the detection probability of the attack. In addition, we analyze the influence of the grid state covariance matrix in the above problem. Using real and public data for the for the SADI we provides a estimation technique of the state covariance matrix exploiting is low rank structure. Then the resulting estimated covariance matrix is included in our analysis of the FDIAs in the SADI. The results of experiments demonstrate, at least for the SADI, the importance of the structure of the covariance matrix of the states for in the above mentioned trade-offs between the harmfulness of the attacks and the detection performance for the detectors at the control center. In fact, we show that a sufficiently informed attacker can leverage this information to generate stronger attacks.