Center for Agile and Intelligent Power Systems (CAIPS): Cybersecurity Research, Development, and Workforce Training
Award ID: DE-CR0000046
Lead PI: Mohammad Ashiqur Rahman;
Scholarship Amount: $2,500,000;
Scholarship Period: October 2024 – September 2026
About The Project
The goal of this project is to establish a regional cybersecurity center leveraging partnerships between regional universities and energy utilities, national laboratories, and system vendors to perform applied research and development in addition to curriculum and workforce development. The research and development focus of the project aims to bring the concept of agility into cyber-physical systems (CPS) by developing a comprehensive platform that spans proactive and reactive dynamic moves on both the physical and cyber components of power generation, transmission, and distribution systems.
Focus Areas of Research
Cyber-physical power systems face evolving cyber threats, especially with the integration of advanced sensing, control, and communication. CAIPS aims to introduce agility-based defense mechanisms using AI-driven Moving Target Defense (MTD) to secure the power grid against stealthy attacks like False Data Injection (FDI) and Denial-of-Service (DoS). The initiative is led by Florida International University with strong collaboration from national labs, utilities, and industry partners including Raytheon and BedRock Systems.
The technical approach includes the following major components:
Dynamic Reconfiguration of Physical Infrastructure: Introduce controlled changes in the power grid’s topology and line parameters using circuit breakers and D-FACTS devices. By dynamically modifying power flows and nodal injections, attackers are denied the static system knowledge required to execute stealthy disruptions. All physical changes are formally verified to ensure operational and economic feasibility.
Cyber-Layer Deception and Data Obfuscation: Alter the communication environment through randomized routing paths, shuffled sensor identifiers, and varied data resolutions. These changes, while transparent to system operators, obscure the attacker’s perception of the system and invalidate their reconnaissance. A provenance-based validation mechanism is employed to verify the integrity and source of incoming data.
Resilient Operational Control under Adversity: Develop intelligent control schemes capable of maintaining grid stability even when facing uncertain measurements or partial system views. Decision-making logic is infused with controlled randomness, ensuring that even if part of the system is compromised, attacker influence on the control layer is minimal. These methods are applied to applications such as Direct Load Control (DLC), balancing reliability and agility.
Out-of-Band Anomaly Detection at the Hardware Level: Leverage the physical side-effects of device operation—such as power consumption, electromagnetic emissions, and execution timing—as fingerprints for intrusion detection. Ensemble learning models detect deviations from expected behavior, including attempts by evasive malware to mimic benign activity. These detectors inform system-wide defense responses in real time.
Integrated Validation and Real-World Demonstration: All components are evaluated through high-fidelity simulations and hardware-in-the-loop setups, including a 1.4MW live microgrid at FIU and RTDS-based testbeds at NCSU. Coordinated attack scenarios, including red-team exercises, assess the effectiveness of each defense layer. The field validation ensures readiness for deployment in utility environments and sets the stage for future commercialization.