Publication Details

Brackenbury, J., O'Shaughnessy, P. Y. & Lin, Y. (2020). On Masking and Releasing Smart Meter Data at Micro-level: the Multiplicative Noise Approach. PSD2020 - Privacy in Statistical Databases (pp. 1-13).


Smart meter electricity data presents privacy risks when malicious agents gain insights of private information, including residents’ lifestyle and daily habits. When allowing access to record-level data, we apply the multiplicative noise method to mask individual smart meter data, which simultaneously aims to minimise disclosure of a dwelling’s consumption signal to any third party and to enable accurate estimation of the sum of a cluster of households. Three testing criteria are introduced to measure the performance of multiplicative noise masking approach relevant to the smart meter data. We propose a novel ‘Twin Uniform’ noise distribution and derive relevant theoretical results. We then implement the multiplicative noise approach in the real smart meter data from ESSnet Big Data. Results are assessed based on privacy, utility and practicality. We conclude that the multiplicative noise method has outstanding practical values. It preforms reasonably well in term of individual value protection and estimation accuracy of the sum when noise distribution is carefully selected .