Sensor Verification for Cyber-Physical Models of Power Systems

Abstract

This project explores the ways that data from sensors in power systems can be authenticated by enhancing the security of power systems from a cyber-physical point of view. This is a continuation of the work for the NSF project “CPS: Synergy: Collaborative Research: Distributed Just-Ahead-Of-Time Verification of Cyber-Physical Critical Infrastructure.” Adversaries who gain access to a cyber-physical system can cause significant physical damage and financial loss by injecting false data into a sensor node. Identifying adversarial action in a system can mitigate unsafe actions made based off of bad data. The technique presented in this work combines topology analysis with real-time probing to create a measure of trustworthiness of sensors in a system. A previously developed tool called Cyber Physical Security Assessment (CyPSA) gives each node a topology vulnerability score based on the cyber accessibility and potential physical impact should it be compromised. We develop a real-time vulnerability score by simulating attack and non-attack scenarios with PowerWorld. The data from these simulations is processed in MATLAB. Results show improved attack detection over current methods. The measure of trustworthiness developed will improve attack detection in power systems, and it may be used to help prevent a system from entering an unstable state.