Privacy lab
Smart meter data can be identifying and revealing - more so than many people appreciate. For example, someone may volunteer to share their energy data, but does not want to reveal their personal religious practices. In some cases it may be possible to infer such information, when energy demand patterns differ during Ramadan or on other religious occasions.
To make an informed decision for consent, one needs to understand what information can and cannot be revealed and what measures can be put in place to safeguard against undesirable disclosure. This applies to both the person giving consent and the organisation obtaining consent to store and process the data.
In this lab, consenting participants will submit detailed personal information. We train models to identify which of these features may uniquely identify them and once identified, what personal information could be inferred about them. The first challenge relates to data privacy, the latter to data utility.
With a robust analytical understanding of privacy and utility of a given dataset, we are able to apply methods to obfuscate, aggregate and synthesis the data until they can be pronounced sufficiently ‘benign’.
An application of such methods can be found in the MSc thesis of John Corsten (Energy Systems MSc, 2024):
Download thesis