Abstract
Thanks to the roll-out of smart meters, availability of fine-grained electricity usage data has rapidly grown. Such data has enabled utility companies to perform robust and efficient grid operations. However, at the same time, privacy concerns associated with sharing and disclosure of such data have been raised. In this paper, we first demonstrate the feasibility of estimating privacy-sensitive household attributes based solely on the energy usage data of residential customers. We then discuss a framework to measure privacy gain and evaluate the effectiveness of customer-centric privacy-protection schemes, namely redaction of data irrelevant to services and addition of bounded artificial noise.