Research On Privacy-preserving Technologies In Mobile Social Networks

Mobile Social Network(MSN)is a key application in the course of Internet de-velopment.MSN has contributed to the profound changes in people's social way,and has evolved into the most important information dissemination channels.A MSN ap-plication can be divided into three layers according to its data transmission direction:network layer,data layer and application layer.The rational use of MSNs and network-ing data can facilitate government decision making,corporate promotion and private customization services.However,the user's profile information and other data in the social network are highly sensitive,which may cause major privacy issues.In this pa-per,we first divide a MSN system into network layer,data layer and application layer.According to the characteristics of each layer,the privacy protection mechanisms in MSN are carefully designed.We analyze review the privacy protection scheme state of the art of MSN,and then propose and implement the privacy-preserving users' profile matching protocols,privacy-preserving streaming data aggregation schemes,privacy-preserving network data release algorithms and privacy-preserving collaborative filtering techniques.Proximity-based networking by users' profile matching can greatly expand the us-ability of MSN applications.However,the matching process must be related to personal interests,illnesses,health conditions or other private information,so we focus on se-cretly and securely finding users with the similar attributes.On the one hand,most related protocols can not solve the selective matching problem and do not take into ac-count the impact of the application environment on privacy requirements.In order to solve this problem,we design two Paillier cryptosystem based Selective Profile Matching Protocols,named Efficiency Preferred Selective Matching(EPSM)and Privacy Preferred Selective Matching(PPSM).EPSM and PPSM is proposed to find users who are similar among some prescribed attributes on the consideration of different privacy requirements and time limits.On the other hand,most of the existing profile-matching protocols were designed on the basis of homomorphic cryptosystem and were not quite efficient in en-cryption and decryption.In this paper,three efficient and privacy-preserving profile matching protocols,which do not use any homomorphic encryption,were proposed to deal with different privacy requirements in MSN.The proposed protocols were proved to be privacy-preserving and correct.The performances of our protocols were thoroughly analyzed and evaluated via real smartphone experiments,and the results show that the proposed protocols can decrease encryption and decryption time by at least an order of magnitude than the Paillier cryptosystem based protocol.Securely aggregate the data of MSN users is the key to widespread sharing of MSN data.AS the users' data is produced continuously,it is imperative to design privacy-preserving stream data aggregation schemes.The privacy-preserving aggrega-tion schemes oriented to the stream data can be divided into two types:the precise ag-gregation schemes based on the cryptography theory and the noisy aggregation schemes based on differential privacy(DP)theory.For the precise aggregation schemes,the re-lated papers either suffer from collusive attack or require a time-consuming initializa-tion at every aggregation request.In this paper,we proposed an efficient aggregation protocol which tolerates up to k passive adversaries who do not try to tamper the computation.The proposed protocol does not require a trusted key dealer and needs only one initialization during the whole time-series data aggregation.Furthermore,the implementation showed that the proposed protocol can be efficient for the time-series data aggregation.For the noisy aggregation schemes,both traditional DP and w-event differential privacy are trusted curator based paradigms.What's more,most work on differential private aggregation inherits the weakness of collectively disclose sensitive information in DP.In this paper,we release the requirement of the trusted curator and propose two schemes(dBD and dBA)to achieve w-event(?)-differential privacy for the semi-trusted data requester in a distributed way.Our solutions employ private distributed function monitoring technique to reduce the communication overheads for each node,and the semi-trusted data requester can get the aggregated statistics without disclosing the raw data of each node.We formally analyzed the utility of our proposed schemes and implemented our mechanisms on real datasets.Privately publishing MSN network structure data is the basis of social network research.It is very challenging to protect the privacy of individuals in social networks while ensuring a high accuracy of the statistics.Moreover,most of the existing work on differentially private social network publication ignores the facts that different users may have different privacy preferences and there also exists a considerable amount of users whose identities are public.In this work,we focus on a specific publication goal when public users are labeled,i.e.,the number of public users that a private user connects to within n hops(denoted as n-range Connection Fingerprints,or n-range CFPs for short).To this end,we proposed two schemes,DEB A and DUBA-LF,for privacy-preserving publication of the CFPs on the base of personalized differential privacy(PDP),and conduct a theoretical analysis of the privacy guarantees provided within the proposed schemes.The implementation showed that the proposed schemes are superior in pub-lication errors on real datasets.Collaborative filtering(CF)is the core application in MSN.However,the CF algo-rithm relies on the user's direct information to provide good recommendations,which may cause major privacy issues.To address these problems,Differential Privacy(DP)has been introduced into CF recommendation algorithms.In this paper,we propose a novel framework called Local-clustering-based Personalized Differential Privacy(LPDP)as an extension of DP.In LPDP,we take the privacy requirements specified at the item-level into consideration instead of employing the same level of privacy guarantees for all users.Moreover,we introduce a local-similarity-based item clustering process into the LPDP scheme,which leads to the result that any items within the same local cluster are hidden.We conduct a theoretical analysis of the privacy guarantees provided within the proposed LPDP scheme.We experimentally evaluate the LPDP scheme on real datasets and demonstrate the superior performance in recommendation quality.