Poster Abstract

Mobile users face privacy risks when providing their locations to untrusted location based services (LBSs). We develop CacheCloak, a system that enables realtime anonymization of location data. A trusted anonymizing server receives the location of a user, processes it using CacheCloak, and exposes the processed data to LBSs. We show that the quality of the location data does not degrade in space or time, and hence, LBSs are able to respond accurately in real time. Yet, privacy is preserved as LBSs are unable to track any given user for more than a short duration. The key is to exploit mobility patterns to intelligently extrapolate each user's path until it meets another. A snapshot of intersecting paths is sent to the LBS, preventing reliable tracking of any given user. We use stochastic methods to generate mobility predictions and entropy measurements to quantify a user's anonymity. We demonstrate a high tracking resilience against a reasonably powerful adversary. Tracebased traffic simulation using real GIS data shows that CacheCloak can achieve realtime location privacy without degrading location accuracy or service availability.