Electricity theft is a billion-dollar problem faced by utilities around the world, and current measures are ineffective against sophisticated theft attacks that compromise the integrity of smart meter communications. Our goal is to detect and mitigate electricity theft by identifying anomalous consumption patterns reported to utility data centers. We plan to use a large dataset obtained from a real smart meter deployment as an example to model normal consumption patterns. Deviations from these patterns, which indicate anomalies, may be detected using statistical methods (such as the Kullback-Leibler divergence), dimensionality reduction methods (such as Principal Component Analysis and its variants), clustering methods (such as DensityBased Spatial Clustering of Applications with Noise), and time series methods (such as AutoRegressive Integrated Moving Average models). We plan to evaluate these methods, and hope to discover new ones in the process of exploration. The evaluation will be performed by simulating electricity theft attacks on consumers in our smart-meter dataset and measuring false positive and false negative rates (receiver operating characteristic).
Investigators include William H. Sanders (PI), Varun Badrinath Krishna, and Gabriel Weaver.
(Funded by the Siebel Energy Institute)