Privacy-Preserving Attribute Matchmaking In Mobile Social Networks

Modernization of our societies and its attendant development in technology has improved the way we communicate and socialize. Mobile communication has significantly been improved in recent times since the advent of smart phones. These developments in smart phones have resulted in it having so many user friendly applications hence aiding easy communication. These new and improved interfaces of smart phones make users more likely to check their social networks on their smart phones than on computers or laptops. Social networks enable persons to make friends not only in their immediate environs but all over the world. Friends making as an integral part of social networks are usually based on similarities of their attributes such as – behaviour, beliefs, experiences,and so on.In order for two users to become friends on social networks, they usually check how compatible they are with each other. This compatibility is usually assessed by the number of attributes they have in common when they find the intersection of their attributes. However, since these attributes maybe private and sensitive they cannot openly compare them to find which attributes they have in common. This brings to the fore the need for limited(privacy-preserving) attribute matching so that, apart from the attributes they have in common, nothing else will be known. Hence, we define this problem as privacy-preserving attribute matchmaking. In this dissertation, we have researched various matchmaking techniques and we have found that the hybrid technique for matchmaking was more secured and privacy-preserving. This helped us to know the weakness,strengths and to better understand these techniques hence enabling us to formulate secure cryptographic matchmaking protocols. The research results are specified as follows:(1) Research on secure privacy-preserving attribute matchmaking cryptographic protocols. These protocols make our attribute matchmaking protocols devoid of leaking sensitive information and secure against semi-malicious and malicious attacks. Variants of authorized private set intersection(APSI) protocols were used hence making our protocols more efficient and secure than prior works. The use of variants of authorized private set intersection ensures that the attributes used in the protocol are authorized(certified)by a mutually trusted authority. The certification prevents arbitrary manipulation of input sets during the protocol. Furthermore, the matchmaking cryptographic protocols in this dissertation match-pair persons only when they fulfil a pre-set criteria defined by the initiator of the matchmaking. This requirement of the protocols in this dissertation is to correct one of the security flaw in some of the existing matchmaking protocols. This security flaw is that, some users terminate the protocol due to lack of enough common attribute. When the protocol is terminated, the individual attributes would have been known by the each other hence bringing the security and privacy of users’ attributes into question.(2) Research on secure privacy-preserving attribute matchmaking cryptographic protocols based on set. We present solutions to the problems in existing matchmaking protocols by proposing well-structured privacy-preserving attribute matchmaking protocols.Here, we consider the users’ attributes as a set, and we present secure privacy-preserving attribute matchmaking cryptographic protocols based on set. In our protocols, each user’s mobile devise is equipped with his/her attributes and the mobile devises can communicate using Bluetooth or Wi-Fi. The certification of users’ attributes by a mutually trusted certification authority(CA) using RSA to enhance the security of our protocols. Apart from the protocols we formulated, we used authenticated Diffie-Hellman protocol, and the technique from Agrawal et al.(2003) to design set-based attribute matchmaking protocols, so that they can further enhance the privacy and security of the users’ information.(3) Research on secure privacy-preserving attribute matchmaking cryptographic protocols based on binary vector. We also considered the attributes as a binary vector, and further present secure privacy-preserving attribute matchmaking cryptographic protocols. In these protocols, each user chooses a fixed number of attributes from the set of attributes of the CA. This attribute matchmaking protocol uses privacy-preserving scalar computation. Furthermore, based on a variant of the protocol by Lu et al.(2013), we present vector-based protocols which can compute the number of attributes users have in common. We also make use of one of the techniques in Wang et al.(2013) to design vector-based attribute matchmaking cryptographic protocols, the security analysis shows that our protocols can prevent collusion attacks by the malicious users.(4) Enhance the security of our proposed matchmaking protocols and give experimental implementation. In the protocols in this dissertation, we defined different levels of privacy. These different levels of privacy further enhance the security of our proposed matchmaking protocols and privacy of users’ information. Comparing with existing matchmaking protocols, our protocols are cryptographically secured and enable users compute the intersection of their attributes without leaking any information. We further investigated the usability of our proposed privacy-preserving attribute matchmaking protocols in the context of its practicality by extensive experimental implementation.Comparing with other existing protocols, we observed that our protocols performed better in the aspect of computation.