Design And Implemention Of A Lightweighted And Fine-grained Privacy-preserving Profile Matching Application For Mobile Social Networking In Proximity

With the rapid development of mobile social network and wide usage of smart mobile terminal, recently the academic and industrial literatures have witnessed the booming deployment of a large number of location-based mobile social networking applications. Especially, mobile social networks in proximity(MSNP) is quickly emerging, which is wireless peer-to-peer(P2P) network of spontaneously and opportunistically connected nodes, and uses geo-proximity as the primary filter in determining who is discoverable on the social network.This thesis mainly deal with the issue of privacy-preserving profile matching in mobile social networking in proximity. Specifically, in order to meet the growing demand for personalized service,MSNP applications need to gather and analyze users’ sensentive profile information to decide whether the pair of users match or not. However, in this procedure, privacy leaking will harm to the safety of the user. Therefore, it is a challenging in MSNP to provide personalized service as well as ensure the safety of users’ privacy.The contributions of this thesis are following threefold:1. Considering there exist two main categories of profile matching protocols in MSNP: coarse-grained and fine-grained, this thesis comprehensively summarized several typical schemes belonging to those two categories, and draw the conclusion that,in comparison with coarse-grained scheme, the fine-grained profile matching protocol can compute similarity degree between users(defined by various formal and strict similarity metrics,e.g.,cosine similarity), which is the basis of providing personalized services. This thesis also explicitly pointed out that due to the fact that the existing fine-grained matching protocols usually employ homomorphism encryption algorithm to protect privacy, which needs heavily exponential computation operations, it brings a high computational complexity to resource constraint mobile terminals.2. In order to satisfy people’s growing demand for personalized service and protect the privacy of users, this thesis proposes a lightweighted and fine-grained profile matching mechanism for MSNP, LIP3.Instrad of using the traditional heavy homomorphism encryption algorithm, LIP3 utilized the encryption algorithm based on matrix transformation, only has multiplication and addition calculations, and it can obtain a certain improvement to reduce the computational complexity. Meanwhile, it uses the vector similarity to represent the attribute similarity, and can intuitively show the similar degree between users.3. Finally, on the Android operating system this thesis uses the Bluetooth communication technology to realize the design of the lightweighted and fine-grained profile matching mechanism, and evaluate the system performance. Experiments show that the designed profile matching protocol has achieived a certain improvement on the computational complexity compared with the original profile matching protocol based on El Gamal algorithm, and verifed the validity of the theoretical analysis.