Evaluating k Nearest Neighbor Query on Road Networks with no Information Leakage

The development of positioning technologies and pervasiveness of mobile devices make an upsurge of interest in location based services LBS. The k nearest neighborkNN query in road networks is an important query type in LBS and has many real life applications, such as map service. However, such query requires the client to disclose sensitive location information to the LBS. The only existing method for privacy-preserving kNN query adopts the cloaking-region paradigm, which blurs the location into a spatial region. However, the LBS can still deduce some information albeit not exact about the location. In this paper, we aim at strong privacy wherein the LBS learns nothing about the query location. To this end, we employ private information retrivial PIR technique, which accesses data pages anonymously from a database. Based on PIR, we propose a secure query processing framework together with flexible query plan for arbitrary kNN query. To the best of our knowledge, this is the first research that preserves strong location privacy for network kNN query. Extensive experiments under real world and synthetic datasets demonstrate the practicality of our approach.