CONSUMER: A novel hybrid intrusion detection system for distribution networks in smart grid

Abstract

Smart meters have been deployed worldwide in recent years that enable real-time communications and networking capabilities in power distribution systems. Problematically, recent reports have revealed incidents of energy theft in which dishonest customers would lower their electricity bills (aka stealing electricity) by tampering with their meters. The physical attack can be extended to a network attack by means of false data injection (FDI). This paper is thus motivated to investigate the currently-studied FDI attack by introducing the combination sum of energy profiles (CONSUMER) attack in a coordinated manner on a number of customers' smart meters, which results in a lower energy consumption reading for the attacker and a higher reading for the others in a neighborhood. We propose a CONSUMER attack model that is formulated into one type of coin change problems, which minimizes the number of compromised meters subject to the equality of an aggregated load to evade detection. A hybrid detection framework is developed to detect anomalous and malicious activities by incorporating our proposed grid sensor placement algorithm with observability analysis to increase the detection rate. Our simulations have shown that the network observability and detection accuracy can be improved by means of grid-placed sensor deployment.