Topology Attack: Optimal Sensor Placement and Fast Topology Attack Detection in Power Distribution Systems

PI: Soummya Kar, Carnegie Mellon University
PI: Rick Blum, Lehigh University

A cyber topology attack occurs when incorrect information about the network topology is transmitted to the system operator/controller, possibly by manipulating/attacking system (topology) databases or through malicious reporting of instantaneous topological states, and incorporated as the truth. A topology attack, if not detected and localized, leads to incorrect understanding of system physical dynamics at the operator level, and, thus, potentially triggering incorrect control actions with disastrous consequences.

The project proposes to develop fast real-time (decentralized) methods for power system topology attack detection and localization, both for transmission and distribution systems. To perform meaningful data analytics toward topology attack detection and localization, we propose to develop a family of cyber-physical type models corresponding to both nominal system operation and hypothesized models corresponding to different types of topology attacks that capture the dependencies between physics, sensing data and attacks.