False data injection attacks on phasor measurements that bypass low-rank decomposition

Abstract

This paper studies the vulnerability of phasor measurement units (PMUs) to false data injection (FDI) attacks. Prior work demonstrated that unobservable FDI attacks that can bypass traditional bad data detectors based on measurement residuals can be identified by detector based on low-rank decomposition (LD). In this work, a class of more sophisticated FDI attacks that captures the temporal correlation of PMU data is introduced. Such attacks are designed with a convex optimization problem and can always bypass the LD detector. The vulnerability of this attack model is illustrated on both the IEEE 24-bus RTS and the IEEE 118-bus systems.

Review of Cyber-Physical Attacks in Smart Grids: A System-Theoretic Perspective

“…Early works typically consider static SE-FDIA, that is, attacks at a given time instant. Some recent works, for example, [43], which focus on PMUs (which have a measurement frequency of tens of Hz), consider FDIA in a setting in which measurements over a block of time are collected and then processed in batch. This results in a measurement equation Z = XH T + E, where Z is the matrix of measurements (where z ij is the measurement collected by instrument j at time i), X the state vector, H the network matrix and E the measurement errors.…”

“…Unobservable attacks as usual take the form of A = CH T and therefore Z + A = (X + C)H T + E. Matrix A = CH T is column-sparse, meaning that only the columns corresponding to the compromised measurement units are non-zero. The measurement matrix Z instead is low-rank [43]. Hence, in literature, convex-optimizationbased decomposition methods have been developed, able to identify the attacked PMUs by separating the measurement matrix into a low rank matrix (the matrix of uncorrupted PMU measurements) and a column-sparse matrix (the attack matrix).…”

“…Hence, in literature, convex-optimizationbased decomposition methods have been developed, able to identify the attacked PMUs by separating the measurement matrix into a low rank matrix (the matrix of uncorrupted PMU measurements) and a column-sparse matrix (the attack matrix). Zhang et al show in [43] how sophisticated attacks can be fabricated that bypass such low-rank decomposition identification techniques.…”

Survey on synchrophasor data quality and cybersecurity challenges, and evaluation of their interdependencies

“…Quality checking methods check for this anomaly and may employ correlation across different timestamps to identify the corruption of data. FDI is one of the widely explored attacks on synchrophasor domain, with solutions like determining the mismatch between the values obtained from PMUs and that observed in SCADA, monitoring the line impedances which get affected when data is manipulated, and using density-based Local Outlier Filter (LOF) analysis [131][132][133][134][135].…”

Power system events classification using genetic algorithm based feature weighting technique for support vector machine

“…However, it is noteworthy that various sophisticated infrastructures including the PMUs are vulnerable. Various experiences and studies in the literature have discussed PMU technology vulnerabilities to numerous intrusions and attacks [ 10 , 11 , 12 , 13 , 14 ]. Pan et al.…”