Zoleikha Biron, PhD, is applying novel machine learning techniques to address challenges facing cyber-physical systems (CPS) in a project recently funded by the National Science Foundation. Her $560k CAREER project, “Toward Resilient and Secure Cyber Physical Systems: Learn Fast from Experiences and Transfer the Knowledge,” aims to bring together analytical solutions and novel machine learning techniques to address interdisciplinary challenges of cyber-security and reliable operation of cyber physical systems with a case study on smart grids.
Distributed large-scale CPSs are complex, interconnected infrastructures that integrate physical processes with computational elements for control, monitoring, and decision-making. Biron’s work focuses on the resiliency of such distributed systems, which rely on data from multiple, geographically dispersed components to function effectively.
A typical example is a power system comprising renewable energy sources, residential and commercial consumers, and traditional power plants or generators. Ensuring reliable performance in these systems requires comprehensive data collection, evaluation, and analysis by decision-making units.
In modern implementations, data is often transmitted over Wi-Fi and communication networks—channels that are vulnerable to cyber-attacks and malicious activity. This research aims to detect cyber-attacks, assess their severity and impact on system performance, and apply machine learning (ML), artificial intelligence (AI), and control theory techniques to enable the system to maintain near-normal operations during and after an attack. Beyond existing resilience and control methods, this research introduces the concept of Anti-Fragility in CPSs. Unlike traditional approaches that focus solely on recovery, anti-fragile systems are designed to learn from past attacks and adapt, thereby strengthening their ability to identify and respond to future, previously unknown threats. This property empowers CPSs not only to withstand cyber-attacks but also to evolve and improve in robustness and adaptability by leveraging prior experiences.
This research helps make critical systems—like power grids or transportation networks—more secure, smarter, and stronger against cyber-attacks. Today, these systems depend on computers and the internet to collect data and make decisions. But that also makes them targets for hackers.
Biron’s project focuses on not just detecting cyber-attacks, but also helping systems recover and get even better after being attacked—similar to how our immune systems learn from past infections to protect us in the future. For power systems, this means:
- More reliable electricity and services, even during cyber threats
- Fewer outages or disruptions, keeping homes, hospitals, and businesses safe
- Smarter systems that can learn from past attacks and protect themselves better over time
In short, the work aims to build self-improving technologies that can keep critical systems running smoothly and safely—even under attack.