Privacy Preservation and Classification of Energy Consumption into Theft Using Benign Dataset in Advanced Metering Infrastructure

Abstract

Internet of Things (IoT) is an integral part of smart grid which provides energy consumption data in real-time from the smart meters to the utility companies as well as the consumers. In an advanced metering infrastructure (AMI), the deployment of smart meters makes it easier for bidirectional communication between the utility companies and the consumers. Energy consumption data of consumers is crucial and sensitive, thus, it is imperative that the privacy of consumers be preserved in an AMI. Somehow, if the AMI is breached and the data is altered before reaching the utility companies, then there may be a chance of incorrect information about energy consumption reaching the utility companies by using various malpractices. Electric utility companies face the toughest challenge of energy theft while supplying energy to the consumers. This is a serious problem all around the world as aggregate technical and commercial (AT&C) loss due to energy theft is an alarming Figure in every country. This work is aimed at developing a machine learning algorithm with regard to the past energy consumption patterns of the customers and to identify the defaulters by training classifiers while keeping the privacy of the consumers using Paillier algorithm for encryption.