Undetectable GPS-Spoofing Attack on Time Series Phasor Measurement Unit Data

TL;DR

This paper introduces an undetectable GPS-spoofing attack on PMUs that gradually alters phase measurements over time, potentially causing power grid breaches without detection, and proposes a new detection method based on low-rank Hankel-matrix approximation.

Contribution

It presents a novel undetectable gradual GPS-spoofing attack model on PMUs and a low-rank Hankel-matrix based detection technique to identify such attacks.

Findings

The attack can cause significant power flow deviations without detection.

The proposed detection method effectively identifies gradual GPS-spoofing attacks.

The attack exploits the vulnerability of GPS time synchronization in PMUs.

Abstract

The Phasor Measurement Unit (PMU) is an important metering device for smart grid. Like any other Intelligent Electronic Device (IED), PMUs are prone to various types of cyberattacks. However, one form of attack is unique to the PMU, the GPS-spoofing attack, where the time and /or the one second pulse (1 PPS) that enables time synchronization are modified and the measurements are computed using the modified time reference. This article exploits the vulnerability of PMUs in their GPS time synchronization signal. At first, the paper proposes an undetectable gradual GPS-spoofing attack with small incremental angle deviation over time. The angle deviation changes power flow calculation through the branches of the grids, without alerting the System Operator (SO) during off-peak hour. The attacker keeps instigating slow incremental variation in power flow calculation caused by GPS-spoofing relentlessly over a long period of time, with a goal of causing the power flow calculation breach the MVA limit of the branch at peak-hour. The attack is applied by solving a convex optimization criterion at regular time interval, so that after a specific time period the attack vector incurs a significant change in the angle measurements transmitted by the PMU. Secondly, while the attack modifies the angle measurements with GPS-spoofing attack, it ensures the undetectibility of phase angle variation by keeping the attack vector less than attack detection threshold. The proposed attack model is tested with Weighted Least Squared Error (WLSE), Kalman Filtering, and Hankel-matrix based GPS-spoofing attack detection models. Finally, we have proposed a gradient of low-rank approximation of Hankel-matrix based detection method to detect such relentless small incremental GPS-spoofing attack.