1 June 2021 in Energy and AI
1 January 2024 in IEEE Transactions on Industrial Informatics
The stability of modern power grids relies heavily on effective frequency control, mainly achieved through automatic generation control (AGC) systems. However, AGC systems have become increasingly vulnerable to cyber-physical attacks due to their dependence on communication infrastructure and cyber-physical devices. In this article, we present a novel attack strategy that leverages the learning capabilities of a convolutional neural network and long short-term memory on encrypted data to predict the imminent operating state of a power system. By accurately forecasting vulnerable operational conditions through encrypted traffic analysis, attackers can orchestrate timely denial-of-service (DoS) attacks on AGC systems, significantly amplifying the physical impact of their cyber-attacks. Despite the implementation of network traffic protection measures through the utilization of the IPsec/ESP protocol, which offers encryption at the network layer to ensure the confidentiality of the original packet content and enhances the complexity of traffic analysis, our research demonstrates that attackers can still extract spatiotemporal features from high-entropy encrypted synchrophasor data packets. This finding underscores the inherent limitations of IPsec/ESP protocols in completely obstructing malicious analysis of network traffic and achieving absolute prevention of such nefarious activities. Consequently, the confidentiality of the power system's operational state remains compromised. To gauge the effectiveness of the proposed attack strategy, we conduct various test cases and simulations, revealing a significant increase in the success rate of DoS attacks. These findings underscore the urgency of implementing countermeasures to thwart attackers from exploiting traffic analysis techniques and emphasize the necessity for strengthened security measures in power grid infrastructure.
1 August 2019 in IEEE Systems, Man, and Cybernetics Magazine
To accommodate more energy from renewable sources, the existing paradigm in the electricity grid of supply follows demand has its limitations. On the other hand, demand response aims to exploit the available flexibility in electricity consumption to follow the supply of renewables. This is possible by decoupling the demand for electricity from the demand of the associated services (such as heating, cooling, and hot water) and presents many challenges, especially if flexibility is harvested at a local scale. For 10 years, research groups worldwide have looked into demand response. This article describes the open challenges of demand response in a residential context and outlines different coordination mechanisms that optimize market profit, mitigate technical issues, and reach prosumer objectives. Several considerations are added on how these cyberphysical solutions must be complemented by other, nontechnical fields to get customers engaged.
3 April 2024 in Sustainable Cities and Society
5 May 2025 in IEEE Transactions on Smart Grid
In cyber-physical power systems, the imperative convergence of information technology (IT) and operational technology (OT) engenders enhanced operational efficiency, yet increases susceptibility to cyberattacks. This scenario accentuates the necessity for deploying effective anomaly detection (AD) mechanisms to protect these critical infrastructures against the devastating impacts of cyberattacks. Traditional AD algorithms often focus on identifying anomalies through analyzing spatial and temporal patterns derived from the physical system measurements, an approach that typically proves effective only after a cyberattack has been successfully executed at the later stages of the cyberkill chain. This reactive approach fails to respond to evolving cyber threats. To tackle this issue, this article introduces an innovative framework for early-stage AD in integrated IT-OT power grid communication networks. Our approach models the IT-OT network as a traffic dispersion graph and applies wavelet transform to decompose the real-time traffic throughput of each information node into its constituent sub-bands to extract time-frequency features. These features are then weighted and prioritized using a graph attention mechanism and fed into a bidirectional long short-term memory for temporal analysis, ultimately classifying the nodes as normal or abnormal. Our experiments confirm the methodology’s precise detection and localization of active cyberattack sites, outperforming existing state-of-the-art AD models.