Cise crosscutting small: CPS: enhancing the security and resilience of cyber-physical power systems against sophisticated cyber-attacks

The National Science Foundation's Division of Information and Intelligent Systems awarded $599,943 to The University Corporation (California State University, Northridge) under the Computer and Information Science and Engineering (CISE) program (CFDA 47.070) to develop advanced security and resilience solutions for cyber-physical power systems. Spanning from September 1, 2025, through August 31, 2028, this project delivers a three-pronged technical framework integrating power system physics with computational methods to detect and mitigate sophisticated cyber-attacks on critical electric grid infrastructure. The research outputs include region-based convolutional graph neural networks for anomaly detection, an analytical framework utilizing convex relaxation and bound-tightening methods to identify infeasible operating state transitions indicative of attacks, and a robust state estimation method capable of reconstructing accurate system states despite compromised data sources.

Beyond technical innovations in attack detection and mitigation, the project advances workforce development objectives by providing STEM training opportunities designed to engage participants from diverse institutional, geographic, and socioeconomic backgrounds. The research combines machine learning, optimization, and power system modeling to protect critical infrastructure against stealthy cyber-threats capable of evading conventional detection algorithms and potentially causing widespread outages or damage. As a federally funded research initiative, the project aligns with CISE's broader mission to support investigator-initiated research and education across computing and information science disciplines while addressing national security and critical infrastructure protection priorities.

Abstract

Cyber-attacks targeting critical infrastructure have increased dramatically in recent years, posing serious risks to national security, economic stability, and public safety. Among the most vulnerable systems are cyber-physical power systems, which integrate physical grid infrastructure with digital communication and control networks. These systems are increasingly exposed to sophisticated cyber threats capable of bypassing traditional defenses. This project aims to enhance the security and resilience of cyber-physical power systems by developing advanced tools to detect and mitigate stealthy cyber-attacks that could otherwise cause widespread outages or damage. The research combines machine learning, optimization, and power system modeling to protect critical electric power grid infrastructure. Beyond its technical innovations, the project will increase participation in stem by offering workforce development opportunities that are open to all students, with particular attention to engaging participants from a wide range of institutional, geographic, and socioeconomic backgrounds. This project introduces a three-pronged technical framework that tightly integrates the physics of power systems with state-of-the-art computational methods. First, IT uses region-based convolutional graph neural networks to detect anomalies in cyber-physical power systems by capturing spatial and temporal dynamics with high fidelity. Second, the project proposes a novel analytical framework based on convex relaxation and bound-tightening methods to identify infeasible transitions between power system operating states, thereby detecting sophisticated attacks that evade conventional detection algorithms. Third, IT develops a robust convex relaxation-based state estimation method that reconstructs the true system state even in the presence of compromised data, enabling rapid isolation and mitigation of attack impacts. Together, these efforts offer a scalable, physics-aware approach to improving the cyber resilience of power grids while contributing to the theoretical foundations of optimization and system verification. 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.- subawards are planned for this award.