Minimizing Energy Theft by Statistical Distance based Theft Detector in AMI

Abstract

To minimize energy theft attacks in Advanced Metering Infrastructure (AMI), we propose statistical distance based theft detection method. In the proposed method, different statistical distance indices (Jensen-Shannon distance, Hellinger distance, and Cumulative Distribution Function based distance) are computed using historical measurement variations between adjacent time steps. When adversary launches energy theft attacks in AMI, distance indices increases. A threshold is set to detect malicious measurement samples. We tested the performance of the proposed method under different attack scenario using real smart meter data. Test results show that the proposed method minimizes energy theft attacks efficiently.