Lightweight private proximity testing for geospatial social networks

Abstract.

Our paper introduces a novel lightweight protocol for private proximity testing.The proposed protocol clearly outperforms current state of the art.We formally prove that the protocol is secure in the semi-honest model.We present experimental results in Python and C++ and compare it with its peers.We demonstrate an Android application that uses are protocol and proves its efficacy. The wide adoption of smart phones has enabled Online Social Networks (OSNs) to exploit the location awareness capabilities offering users better interaction and context aware content. While these features are very attractive, the publication of users' location in an OSN exposes them to privacy hazards. Recently, various protocols have been proposed for private proximity testing, where users are able to check if their online friends are near, without disclosing their locations. However, the computation cost of the required cryptographic operations utilized in such protocols is not always efficient for mobile devices. In this paper we introduce a lightweight and secure proximity testing protocol, suitable for online mobile users. We show that our protocol is provably secure under the well-known factoring problem and we analyze its efficiency. Our results show that our approach outperforms other existing protocols, by significantly reducing the computational cost and making it practical for devices with limited resources. Finally, we demonstrate the applicability of our proposal in an actual OSN location-based, mobile application.

Keywords: private proximity testing; Location privacy; geospatial social netowrks;

Download: (pdf)