Hypothesis Testing for Privacy of Smart Meters With Side Information

Abstract

Privacy concern is often underlined as the main obstacle to the path for the large deployment of smart meters (SMs). An energy management unit (EMU) along with a proper battery system charge/discharge schedule can mask the consumer’s actual energy use and to some extent alleviate privacy concerns. This paper speculates a hypothesis testing problem focused on SM privacy. The proposed model consists of an energy consumer, battery storage, EMU, and SM. The EMU, which controls the energy consumption policy, chooses between two hypotheses. The first one considers a normal-mode consumption policy in which EMU utilizes the consumer’s entire energy demand from the grid. In the second hypothesis, EMU takes some part of the energy consumption from the grid independent of the consumer’s actual demand. In this situation, the battery with a limited capacity undertakes supplying the remaining load or absorbing the excessive energy. The second approach would conceal the consumer’s profile from an adversary. The energy provider, which has access to a side information, could predict the hypothesis utilized by the consumer. We denote the reliability criterion as probability of error in choosing a correct hypothesis. A case study is given in the support of effectiveness of proposed methodology.