A Location Privacy Estimator Based on Spatio-Temporal Location Uncertainties

Abstract

The proliferation of mobile devices and location-based services (LBS) is strongly challenging user privacy. Users disclose a large volume of sensitive information about themselves to LBS. Indeed, such services collect user locations to operate and can thus use them to perform various inference attacks. Several privacy mechanisms and metrics have been proposed in the literature to preserve location privacy and to quantify the level of privacy obtained when these mechanisms are applied on raw locations. Although the use of these metrics is relevant under specific threat models, they cannot anticipate the level of location privacy on the sole basis of the altered location data shared with LBS. Therefore, we propose a location privacy estimator that approximates the level of location privacy based on spatio-temporal uncertainties resulting from location alterations produced when a location privacy preserving mechanism is applied on user raw locations. This estimator also takes into account spatial-temporal user privacy parameters. We also describe the computation of the spatio-temporal uncertainties through the sampling, the Gaussian perturbation as well as the spatial cloaking. Finally, we compare the results of our estimator with those of the success of two localization attacks. The findings show that our estimator provides reasonable or conservative estimates of the location privacy level.

Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.