Wavelet-based load profile representation for smart meter privacy

A significant portion of (potential) end-users at this point in time are wary about possible disadvantages of smart grid technologies. A critical issue raised by end-users in various studies is the lack of trust in the level of privacy. Smart metering is the component in the end-user domain around which the most intense debate on privacy revolves, because load profiles are made available at high resolutions. Non-intrusive load monitoring (NILM) techniques allow the analysis of these load profiles to infer user behaviour, such as sleep-wake cycles. We investigate and compare the utility of different variants of the wavelet transform for creating a multi-resolution representation of load profiles. In combination with selective encryption, this multi-resolution representation allows end-users to grant or deny access to different resolutions on a “need-to-know” basis. Access to the different resolutions is thereby only granted to parties holding the needed keys. The whole datastream can be transmitted over the smart grid communications network. The lifting implementation of the wavelet transform has computationally low demands and can be run in embedded environments, e.g. on ARM-based architectures, in acceptable time. The proposed approach is evaluated based on the provided level of security, computational demands and feasibility in an economic sense.