Stealthy MTD Against Unsupervised Learning-Based Blind FDI Attacks in Power Systems

This paper examines how moving target defenses (MTD) implemented in power systems can be countered by unsupervised learning-based false data injection (FDI) attack and how MTD can be combined with physical watermarking to enhance the system resilience. A novel intelligent attack, which incorporates dimensionality reduction and density-based spatial clustering, is developed and shown to be effective in maintaining stealth in the presence of traditional MTD strategies. In resisting this new type of attack, a novel implementation of MTD combining with physical watermarking is proposed by adding Gaussian watermark into physical plant parameters to drive detection of traditional and intelligent FDI attacks, while remaining hidden to the attackers and limiting the impact on system operation and stability.2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. M. Higgins, F. Teng, and T. Parisini, "Stealthy MTD Against Unsupervised Learning-Based Blind FDI Attacks in Power Systems," IEEE Transactions on Information Forensics and Security, vol. 16, pp. 1275-1287, 2021, doi: 10.1109/TIFS.2020.3027148