@inproceedings{ashebo_covert_2025,
author={Ashebo, Betelihem Kebede and Talkington, Samuel and Zonouz, Saman and Molzahn, Daniel K.},
booktitle={57th North American Power Symposium (NAPS)},
title={{Covert Distribution Load Tripping Attacks}},
year={2025},
volume={},
number={},
doi={10.1109/NAPS66256.2025.11272195}}
The increasing integration of distributed energy resources (DERs), particularly solar photovoltaic (PV) systems, has introduced new cybersecurity challenges in distribution networks. This paper presents a data-driven attack model that examines how an adversary can exploit direct load control (DLC) mechanisms to selectively disconnect downstream loads during periods of high solar generation. Such targeted load tripping forces excess PV output to flow back toward the substation transformer, potentially causing power imbalances and transformer overloading. We model both PV output and load demand as multivariate Gaussian distributions to capture their inherent temporal and spatial uncertainties. A probabilistic power imbalance metric is defined to quantify the extent of reverse flow under compromised conditions. To identify the most impactful combinations of load disconnections and timing, we employ a multi-armed bandit approach based on the Upper Confidence Bound (UCB) algorithm. Simulation results demonstrate the feasibility and effectiveness of the attack strategy under realistic variability in solar output and demand.
Admittance Matrix Concentration Inequalities for Understanding Uncertain Power Networks
Samuel Talkington, Cameron Khanpour, Rahul K. Gupta, Sergio A. Dorado-Rojas, Daniel Turizo, Hyeongon Park, Dmitrii M. Ostrovskii, and Daniel K. Molzahn