Key points are not available for this paper at this time.
The solutions offered to-date for end-user privacy in smart meter measurements, a well-known challenge in the smart grid, have been tied to specific technologies such as batteries or assumptions on data usage without quantifying the loss of benefit (utility) that results from any such approach. Using tools from information theory and a hidden Markov model for the measurements, a new framework is presented that abstracts both the privacy and the utility requirements of smart meter data. This leads to a novel privacy-utility tradeoff problem with minimal assumptions that is tractable. For a stationary Gaussian model of the electricity load, it is shown that for a desired mean-square distortion (utility) measure between the measured and revealed data, the optimal privacy-preserving solution: i) exploits the presence of high-power but less private appliance spectra as implicit distortion noise, and ii) filters out frequency components with lower power relative to a distortion threshold; this approach encompasses many previously proposed approaches to smart meter privacy.
Sankar et al. (Thu,) studied this question.