On the Lack of Anonymity of Anonymized Smart Meter Data: An Empiric Study

Abstract

After over a decade of research into the privacy of smart meters, the sensitivity of an individual household’s fine-grained energy readings is undisputed. A plethora of research contributions aim at protecting the privacy of end users while at the same time providing the energy supplier (and others) with sufficient data for safe operation, billing, and also forecasting purposes. The transmission of fine-grained readings is generally considered acceptable as long as they cannot be linked to the households they originate from (i.e., anonymized readings). Mart ́ınez et al. just recently pointed out that the typical provision of aggregated readings at the end of a billing period could compromise this anonymity, as the individual readings must sum up to the respective aggregate. In this short paper, we complement their research by examining the privacy implications of published aggregates of previously anonymized energy readings: We simulate attacks on a real world data set (Smart*), particularly investigating the implications of different parameter combinations such as aggregation group sizes, considered time spans, and reading precision to gain insights into theoretic risks, e.g., from an incautious choice of parameters.