In a new briefing paper from the Imperial College Energy Futures Lab, the researchers point to the ‘enabling’ role of smart meter data in the energy transition but state that access to this data faces a challenge due to consumer privacy concerns.
Currently, once a smart meter is installed, customers can choose whether they wish to share their consumption data at a monthly, daily or half-hourly frequency, whether their supplier may use the data for marketing purposes and whether the data can be passed on to third parties.
However, several recent developments could fundamentally change how and by whom smart meter and associated consumption data can be accessed. Smart appliances and demand response are expected to play a significant role, which will generate complementary data streams with additional, more granular data on energy consumption.
In addition, there are calls for widening access to smart meter data for uses beyond the day-to-day operation of the electricity network as part of a move to digitalise the energy sector. A key component of this is to develop open data platforms which will provide access to public interest actors to inform and shape policy.
Privacy-preserving techniques
The researchers suggest various privacy-preserving techniques of which one is data obfuscation such as pseudonymisation and aggregation, which alter data to remove information that may be considered sensitive.
Another is differential privacy, which introduces noise into datasets to provide mathematical guarantees of anonymity, while a third is homomorphic encryption, which allows mathematical operations on the encrypted data.
Others are user demand shaping, which functions behind the meter by altering actual consumption patterns, and distributed learning techniques such as federated and peer-to-peer learning, which ensure the raw consumption data is kept locally, e.g. in the smart meter.
Warning that without significant changes to how data is stored, processed and used, the adoption and resulting benefits of smart meters may not be realised, the researchers recommend the introduction of one or a combination of such privacy-preserving techniques.
In particular, differential privacy is suggested as a prime candidate, given that it can be implemented either centrally or in a distributed manner and provides provable, tuneable and future proof guarantees of privacy protection. Moreover, it has been implemented in other sectors, e.g. in the publication of the 2020 US census.
“Privacy-preserving techniques provide a framework within which the privacy-utility trade-off can be explored and quantified They give consumers greater control over their data and force those wishing to access their smart meter data to incentivise them,” write the researchers in the paper.
“Going beyond permission control, the use of privacy-preserving mechanisms can ensure data is not easily re-identifiable. Even in the case of a data breach, any inferences cannot be linked back to individual consumers. Importantly, this is done while preserving the utility of the data so that the benefits can be realised.”
