Universal location referencing and homomorphic evaluation of geospatial query

Abstract

Location data reveals users’ trajectories, yet it is often shared to enable many location-based services (LBS). In this paper, we propose a privacy-preserving geospatial query system with geo-hashing and somewhat homomorphic encryption. We geo-hash locations using space-filling curves for locality-preserving dimension reduction, which allows the users to specify granularity preference of their location and is agnostic to specific maps or precoded location models. Our system features three homomorphic algorithms to compute geospatial queries on encrypted location data and encrypted privacy preferences. Comparing with previous work, one of our algorithms reduces the multiplicative depth of a basic homomorphic computation approach by more than half, which significantly speeds it up. We then present an optimized prototype and experimentally demonstrates its utility in spatial cloaking.