A Fine-grained Privacy-Preserving Profile Matching Scheme in Mobile Social Networks

Tao Peng, Wentao Zhong, Kejian Guan, Yipeng Zou, Jiawei Zhu, Guojun Wang

Abstract: Mobile Social Networks (MSN) have made it easy for us to communicate with our friends. Friend discovery is a common feature in the MSN application, which can recommend friends to requesters by comparing the similarity of attributes between users. However, in the process of attribute matching, users' personal information may be stolen by the server or other malicious users, leading to privacy leakage. Many solutions have been proposed for this problem, but the existing solutions have not considered the different ranges of user attribute values when calculating their similarity, leading to the final matching results not accurate. The existing solutions also do not consider users' location attributes and cannot find users according to the requester's query range. To address these issues, we propose a fine-grained privacy-preserving profile matching scheme (FPPM) that supports precise queries. In our scheme, the requester can flexibly set the query range. We use order-preserving encryption to compare the cipher-text submitted by the user to precisely achieve similarity matching. To further protect users' privacy security, we design a secure dot product protocol (SDPP) that uses two servers to jointly compute the dot product of cipher-text vectors to avoid privacy leakage during the computation. Our proposed scheme supports the requester to define the querying range to obtain fine-grained query results. Based on it, we also consider multi-dimensional privacy protection such as user feature attributes and location attributes.