TraceMixer: Privacy-preserving crowd-sensing sans trusted third party

Crowd-sensing promises cheap and easy large scale data collection by tapping into the sensing and processing capabilities of smart phone users. However, the vast amount of fine-grained location data collected raises serious privacy concerns among potential contributors. In this paper, we argue that crowd-sensing has unique requirements w.r.t. privacy and data utility which renders existing protection mechanisms infeasible. We hence propose TraceMixer, a novel location privacy protection mechanism tailored to the special requirements in crowd-sensing. TraceMixer builds upon the well-studied concept of mix zones to provide trajectory privacy while achieving high spatial accuracy. First in this line of research, TraceMixer applies secure two-party computation technologies to realize a trustless architecture that does not require participants to share locations with anyone in clear. We evaluate TraceMixer on real-world datasets to show the feasibility of our approach in terms of privacy, utility, and performance. Finally, we demonstrate the applicability of TraceMixer in a real-world crowd-sensing campaign.