CAREER: Towards attack-resilient cyber-physical smart grids: moving target defense for data integrity attack detection, identification and mitigation

Kansas State University, Manhattan KS

Investigators

Abstract

This NSF CAREER project aims to provide a theoretical foundation and design guiding principles that will unlock the full potential of moving target defense (MTD) approaches and significantly enhance the resiliency of cyber-physical power grids under cyber data attacks. The project will transform existing bulk transmission system operations that rely on limited cyber-layer security mechanisms to proactive defense-in-depth approaches in both the cyber and physical layers using widely-deployed smart devices. The intellectual merits of the project include developing novel optimization, graph theory, low-rank matrix theory, and machine learning-based methods for optimal planning and operation of moving target defense devices, rapid detection, accurate identification, and robust mitigation of cyber data integrity attacks. The broader impacts of the project include promoting public awareness and understanding of smart grid cybersecurity, contributing to power engineering education, and preparing a diverse learning community, including middle and high school students, with requisite knowledge and skillsets to tackle future power grid security challenges. The successful completion of this project will provide power system operators with new tools to enhance situational awareness and better defend the power grid against cyber data attacks. MTD is an emerging concept originally introduced for computer and communication networks. Existing MTD approaches are limited to the cyber layer of a cyber-physical system. However, if field devices or internal communication networks are physically compromised, adverse consequences are trigged within the physical layer. Therefore, the cyber-layer MTD alone is inadequate for securing real-world power grids with significant attack surfaces. The goal of this CAREER project is to develop and validate physical-layer MTD approaches to detect, identify, and mitigate data integrity attacks by strong adversaries with state-of-the-art machine learning capabilities. The proposed MTD approaches feature three major technical innovations: 1) A minimum spanning tree-enabled planning scheme that maximizes MTD detection effectiveness while considering system economic and reliability metrics; 2) A novel alternating current optimal power flow operational framework, constrained by scalable voltage stability approaches, to ensure the MTD hiddenness and detection performance; and 3) A low-rank matrix decomposition method assisted by MTD approaches that radically improves the attack identification speed and measurement recovery accuracy. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.