( k, R, r )-anonymity: a light-weight and personalized location protection model for LBS query

This paper studies the problem of location and query content preserving in location based service (LBS) systems. Based on the private information retrieval (PIR) theory and location k -anonymity model, we propose a new privacy preserving model, called ( k, R, r )-anonymity, which is a light-weight and personalized resolution that can be implemented on mobile terminals. Main idea of the proposed model is to replace the user's real location and the query target by a specific area and a set of location types, respectively. A user may control his privacy preserving degree according to his specific demand by dynamically adjusting parameters of the proposed model, such as query contents, the size and location of the anonymous region, etc. Taking the nearest neighbor query as example, we evaluate performance of the proposed model and make a brief comparison with k -anonymity and PIR. Results show that the proposed model provides stronger anonymity protection than k -anonymity, whereas it its simpler and induces less additional overheads than PIR. In addition, it does not need a third-party privacy anonymity server either.