False data injection attacks against nonlinear state estimation in smart power grids

Abstract

False data injection attacks are recently introduced as a class of cyber attacks against smart grid's monitoring systems. They aim to compromise the readings of grid sensors and phasor measurement units. Recent studies have shown that if the operator uses the DC, i.e., linear, state estimation to determine the current states of the power system, the attacker can adjust the attack vector such that the attack remains undetected and successfully passes the commonly used residue-based bad data detection tests. However, in this paper, we examine the possibility of implementing a false data injection attack when the operator uses the more practical AC, i.e., nonlinear, state estimation. We characterize such attacks when the attacker has perfect and imperfect knowledge of the current states of the system. To the best of our knowledge, this is the first paper to address false data injection attacks against non-linear state estimation.