Evaluation of the Precision-Privacy Tradeoff of Data Perturbation for Smart Metering

Smart grid users and standardization committees require that utilities and third parties collecting metering data employ techniques for limiting the level of precision of the gathered household measurements to a granularity no finer than what is required for providing the expected service. Data aggregation and data perturbation are two such techniques. This paper provides quantitative means to identify a tradeoff between the aggregation set size, the precision on the aggregated measurements, and the privacy level. This is achieved by formally defining an attack to the privacy of an individual user and calculating how much its success probability is reduced by applying data perturbation. Under the assumption of time-correlation of the measurements, colored noise can be used to even further reduce the success probability. The tightness of the analytical results is evaluated by comparing them to experimental data