Smart-metering systems report electricity usage of a user to the utilityprovider on almost real-time basis. This could leak private information aboutthe user to the utility provider. In this work we investigate the use of arechargeable battery in order to provide privacy to the user. We assume thatthe user load sequence is a first-order Markov process, the battery satisfiesideal charge conservation, and that privacy is measured using normalized mutualinformation (leakage rate) between the user load and the battery output. Weconsider battery charging policies in this setup that satisfy the feasibilityconstraints. We propose a series reductions on the original problem andultimately recast it as a Markov Decision Process (MDP) that can be solvedusing a dynamic program. In the special case of i.i.d. demand, we explicitlycharacterize the optimal policy and show that the associated leakage rate canbe expressed as a single-letter mutual information expression. In this case weshow that the optimal charging policy admits an intuitive interpretation ofpreserving a certain invariance property of the state. Interestingly analternative proof of optimality can be provided that does not rely on the MDPapproach, but is based on purely information theoretic reductions.
展开▼
