Design And Implementation Of A Location-Based Social Network Service System Based On Privacy-preserving

Recently, the privacy-preserving has become one of the key research issues in location-based social network service(LBSNS), in which individual’s private information includes not only her location information, but also her personal attribute data(e.g., age, sex, interests, etc.). The purpose of LBSNS applications is to consolidate the existing relationship between friends and to find potential friends in close location. The latter needs to utilize users’ personal attribute information in the operation of private matching, which may cause privacy breach. Therefore, this thesis designs a privacy-protection enabled LBSNS system, in which an effective and lightweight privacy protection method is proposed and implemented in private matching without exposing user’s personal attribute information.Firstly, this thesis introduces the background knowledge of current LBSNS and makes an explanation about the classification of privacy information and their corresponding protection methods.Secondly, considering the architecture and characteristics of LBSNS, this thesis summarizes two typical approaches to private matching(i.e., coarse-grained private matching and fine-grained private matching), and the limitation of coarse-grained private matching and the advantage of fine-grained private matching are discussed.Then, through adopting homomorphic encryption in secure multi-party computation(SMC), an ElGamal crypto-system based private matching protocol is proposed to protect user’s personal attribute information, which can perform the similarity computation between any pair of users without exposing their private attributes. Then, we theoretically analyze our proposal’s computation complexity(including computation cost and communication cost) and compare with the existing Paillier-based private matching protocol. The following results can be inferred. The computation complexity of ElGamal-based private matching protocol is only related to the parameter d(the number of attributes information) in matching process, while the computation complexity of Paillier-based private matching protocol is relevant to d and ?(the parameter of extending attributes vector). Furthermore, when ??2, both computation cost and communication cost of ElGamal-based private matching protocol are less than those in Paillier-based private matching protocol with the increase of d. We also design a LBSNS system with built-in ElGamal crypto-system based private matching protocol.Finally, on Android platform, using Bluetooth technology, the designed LBSNS system is designed and implemented, in which our proposed El Gamal-based private matching protocol is adopted. Moreover, this thesis makes a performance evaluation of this private matching protocol. The experimental results comply with the theoretical analysis results: the computation complexity of our proposed private matching protocol is less than that in the existing private matching protocol(Paillier-based) when ??2. Moreover, we also found that with the increase of the number of attributes used, our proposal performs better than the existing work.