Privacy-Preserving Inverse Distance Weighted Interpolation

Abstract

Inverse distance weighted (IDW) interpolation is one of the well-known geo-statistics techniques. On the one hand, one party (server) holding some measurements for specific locations wants to provide predictions; on the other hand, another party (client) is looking for a prediction for a particular point. However, due to privacy concerns, neither the server nor the client wants to reveal their confidential data to each other. We propose privacy-preserving schemes to provide IDW-based predictions without violating confidentiality. We analyze our enhanced scheme in terms of privacy and performance. Such analyses show that our improved method does not violate privacy and provides predictions efficiently. We also perform real data-based experiments to show how it affects accuracy. Empirical results show that it is able to estimate accurate predictions.