Structure of optimal privacy-preserving policies in smart-metered systems with a rechargeable battery

Abstract

In smart-metered systems, fine-grained time-series power usage data (load profile) is communicated from a user to the utility provider. The correlation of the load profile with a user's private activities leaves open the possibility of inference attacks. Using a rechargeable battery, the user can partially obscure its load profile and provide some protection to the private information using various strategies for charging and discharging the battery (battery management policies). Using mutual information as the privacy metric, we study optimal battery management policies for discrete alphabets. We show that the problem can be formulated as a Markov Decision Process, identify the associated state and action space, and using this framework characterize the optimal policy for the binary alphabet case.