Load Oscillating Attacks of Smart Grids: Vulnerability Analysis

We investigate the vulnerability of a power transmission grid to load oscillation attacks. We demonstrate that an adversary with a capability to oscillate a small fraction of load and monitor some tie line flows can launch a successful load oscillation attack to destabilize the grid. The adversary is assumed to be able to compromise smart meters at a subset of load buses and control their switches. In the reconnaissance stage of the proposed attack, the adversary estimates the line flow sensitivity factors (LFSFs) associated with the monitored tie lines by perturbing a small amount of load at compromised buses and observing how the monitored line flow changes. The learned LFSF values are used to select a target line and optimize the oscillation attack magnitudes at the compromised load buses to cause the target line to trip. In the execution stage, the oscillation attack frequency is adjusted in real time, based on the monitored target line flow measurements, in order to maximize the attack impact under the system condition at the attack time. We experimentally demonstrate that the proposed attack strategy can result in a much more severe consequence than benchmark oscillation attacks, using the COSMIC time-domain simulator with two test cases, the IEEE RTS-96 and Polish 2383-Bus Systems. The proposed attack strategy succeeded in causing a full blackout by oscillating only 8% of the load in the IEEE RTS-96 test system, and 1.7% of the load in the Polish test system.