Location Oblivious Privacy Protection for Group Nearest Neighbor Queries

Finding a convenient meeting point for a group is a common problem. For example, a group of users may want to meet at a restaurant that minimizes the group’s total travel distance. Such queries are called Group Nearest Neighbor (GNN) queries. Up to now, users have had to rely on an external party, typically a location service provider (LSP), for computing an optimal meeting point. This implies that users have to trust the LSP with their private locations. Existing techniques for private GNN queries either cannot resist sophisticated attacks or are computationally too expensive to be implemented on the popular platform of mobile phones. This paper proposes an algorithm to efficiently process private GNN queries. To achieve high efficiency we propose an approach that approximates a GNN with a high accuracy and is robust to attacks. Unlike methods based on obfuscation, our method does not require a user to provide an imprecise location and is in fact location oblivious. Our approach is based on a distributed secure sum protocol which requires only light weight computation. Our experimental results show that we provide a readily deployable solution for real life applications which can also be deployed for other geo-spatial queries and applications.