Towards Attack Resilient Data Analytics for Power Grid Operations
Date
April 27, 2018
Description
With an abundance of real-time data from a power grid, real-time data analytics is expected to revolutionize the way that critical decisions are made in power system operations. Nevertheless, employing data analytics for critical decision making may create a new attack surface unless a proper countermeasure is accompanied. In particular, an adversary may perturb power system operations by manipulating part of input data to the analytics. Therefore, an attack-resilient framework of designing and implementing such data analytics is necessary. In this talk, we will present an attack-resilient data analytics framework for power system protection. The proposed data analytics employs an artificial intelligence method equipped with power system simulator modules. The analytics processes real-time PMU data streams to compute an optimal protection decision in the presence of a GPS spoofing attack on PMUs. We will present the overview of our attack-resilient analytics and demonstrate its robustness against GPS spoofing attacks.