Joint Detection and Localization of Stealth False Data Injection Attacks in Smart Grids Using Graph Neural Networks

Boyaci, Osman; Narimani, Mohammad Rasoul; Davis, Katherine; Ismail, Muhammad; Overbye, Thomas J.; Serpedin, Erchin

doi:10.1109/tsg.2021.3117977

Cited by 67 publications

(15 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We note that PSSF has so far been pursued via single-hidden-layers NNs [26,27], and further investigated by the Recurrent neural networks in [28] and Graph Recurrent neural networks [29]. The state-of-art neural network algorithms for FDI attack detection have been pursued by the Chebyshev GCN [30], CNN [31] and RNN [32].…”

Section: A Related Workmentioning

confidence: 99%

“…voltage magnitudes, active and reactive power injections) and GRNN [29] that combines a GNN1st layer and a LSTM layer together to capture the spatio-temporal correlations. Likewise, the stateof-art algorithms for false data detection and localization include ChebyGCN [30] that uses the absolute values of the admittance matrix and consider active and reactive powers as inputs, CNN [31] that takes both line and bus measurements 3 , and LSTM [32] that considers voltage phasors as inputs, and GSP algorithm [4].…”

Section: Numerical Experimentsmentioning

confidence: 99%

See 1 more Smart Citation

Complex-Value Spatio-temporal Graph Convolutional Neural Networks and its Applications to Electric Power Systems AI

Wu¹,

Scaglione²,

Arnold³

2022

Preprint

View full text Add to dashboard Cite

The effective representation, precessing, analysis, and visualization of large-scale structured data over graphs are gaining a lot of attention. So far most of the literature has focused on real-valued signals. However, signals are often sparse in the Fourier domain, and more informative and compact representations for them can be obtained using the complex envelope of their spectral components, as opposed to the original real-valued signals. Motivated by this fact, in this work we generalize graph convolutional neural networks (GCN) to the complex domain, deriving the theory that allows to incorporate a complex-valued graph shift operators (GSO) in the definition of graph filters (GF) and process complex-valued graph signals (GS). The theory developed can handle spatio-temporal complex network processes. We prove that complex-valued GCNs are stable with respect to perturbations of the underlying graph support, the bound of the transfer error and the bound of error propagation through multiply layers. Then we apply complex GCN to power grid state forecasting, power grid cyber-attack detection and localization.

show abstract

Section: A Related Workmentioning

confidence: 99%

Section: Numerical Experimentsmentioning

confidence: 99%

Complex-Value Spatio-temporal Graph Convolutional Neural Networks and its Applications to Electric Power Systems AI

Wu¹,

Scaglione²,

Arnold³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…In [22], the authors present a spatiotemporal feature extraction mechanism using cubature Kalman Filters and Gaussian process regression, following which a Deep Convolutional Neural Network is used to map a function between these features and the indicating labels detecting the presence of FDIA in the power system. In [23] the authors propose an Auto-Regressive Moving Average Graph Neural Network or ARMA -GNN based classification algorithm for joint detection and localization of FDIA attacks. These model free methods proposed in literature can be further classified into two different categories -supervised algorithms and unsupervised algorithms.…”

Section: Introductionmentioning

confidence: 99%

Deep Latent Space Clustering for Detection of Stealthy False Data Injection Attacks Against AC State Estimation in Power Systems

Bhattacharjee

Mondal

Verma

et al. 2023

IEEE Trans. Smart Grid

View full text Add to dashboard Cite

“…One of the main challenges is in the cascading failures caused by cyber-attacks or in the failure of a component in the power grid [3]. These attacks may occur when nodes or transmission lines in a smart grid fail [4]. When a power grid asset fails, this failure can propagate across the system due to the interdependencies between the power assets and the communication components.…”

Section: Introductionmentioning

confidence: 99%

“…Many studies investigated how to mitigate the cascading failure impacts caused by cyber-attacks. For example, [3], [4], [8]- [10] proposed algorithms to identify the mostvulnerable power lines and components in terms of their susceptibility to cascading attacks to strengthen power systems by protecting these vulnerable assets. These efforts show the importance of addressing the problem of cascading failure attacks on smart grids and the consequences of failing to do so.…”

Section: Introductionmentioning

confidence: 99%

Modeling Cascading Failures in Coupled Smart Grid Networks

et al. 2022

View full text Add to dashboard Cite

The smart grid connects components of power systems and communication networks in an interdependent two-way system that delivers electricity to consumers and collects data that enables it to react to usage levels and interference from threats, such as cyber-attacks. In this paper, we propose a novel cyber-attack failure propagation model in smart grids. Our realistic failure propagation model addresses the system's heterogeneity by assigning different roles to its components. We define rules for and interdependencies of failure propagation and propose a new approach to studying cascading failures. In addition, our graph model identifies the most-vulnerable nodes. The model implements power flow analysis to guarantee that all transmission lines work below capacity and remove lines exceeding capacity. The model also considers that control packets could encounter different delays regarding the communication network structure and investigates the impact of communication delay on the failure of power components. Our results establish that by considering both power and communication characteristics and interdependencies, cascading failures can be modeled more accurately. We show that when we run the power flow analysis, there are a negligible number of failed nodes, which means that our model accurately identifies system failures.

show abstract

Joint Detection and Localization of Stealth False Data Injection Attacks in Smart Grids Using Graph Neural Networks

Cited by 67 publications

References 39 publications

Complex-Value Spatio-temporal Graph Convolutional Neural Networks and its Applications to Electric Power Systems AI

Complex-Value Spatio-temporal Graph Convolutional Neural Networks and its Applications to Electric Power Systems AI

Deep Latent Space Clustering for Detection of Stealthy False Data Injection Attacks Against AC State Estimation in Power Systems

Modeling Cascading Failures in Coupled Smart Grid Networks

Contact Info

Product

Resources

About