Preserving user privacy in the smart grid by hiding appliance load characteristics

Abstract

As data transmitted in the smart grid are fine-grained and private, the personal habits and behaviors of inhabitants may be revealed by data mining algorithms. In fact, nonintrusive appliance load monitoring (NALM) algorithms have substantially compromised user privacy in the smart grid. It has been a realistic threat to deduce power usage patterns of residents with NALM algorithms. In this paper, we introduce a novel algorithm using an in-residence battery to counter NALM algorithms. The main idea of our algorithm is to keep the metered load around a baseline value with tolerable deviations. Since this algorithm can utilize the rechargeable battery more efficiently and reasonably, the metered load will be maintained at stable states for a longer time period. We then implement and evaluate our algorithm under two metrics, i.e., the step changes reduction and the mutual information, respectively. The simulations show that our algorithm is effective, and exposes less information about inhabitants compared with a previously proposed algorithm. © Springer International Publishing Switzerland 2013.