Privacy-Friendly Forecasting for the Smart Grid Using Homomorphic Encryption and the Group Method of Data Handling

While the smart grid has the potential to have a positive impact on the sustainability and efficiency of the electricity market, it also poses some serious challenges with respect to the privacy of the consumer. One of the traditional use-cases of this privacy sensitive data is the usage for forecast prediction. In this paper we show how to compute the forecast prediction such that the supplier does not learn any individual consumer usage information. This is achieved by using the Fan-Vercauteren somewhat homomorphic encryption scheme. Typical prediction algorithms are based on artificial neural networks that require the computation of an activation function which is complicated to compute homomorphically. We investigate a different approach and show that Ivakhnenko's group method of data handling is suitable for homomorphic computation. Our results show this approach is practical: prediction for a small apartment complex of 10 households can be computed homomorphically in less than four seconds using a parallel implementation or in about half a minute using a sequential implementation. Expressed in terms of the mean absolute percentage error, the prediction accuracy is roughly 21%.