Research On Privacy Protection Issues In Distributed Social Networks

In recent years, Online Social Networks(OSNs) have gained rapidly impressive development and have become considerable parts of our Internet Life. Currently, the mainstream online social networks are almost completely based on centralized architecture which can benefit managements and manipulations extremely. What is more, the centralized architecture will facilitate service providers to assemble and analyze user data. In case of that service providers sell or disclose user data to incredible third-parties for business purpose,enormous threats will be posed to user privacy security. However, traditional privacy-preserving methods can't effectively eliminate these threats which are induced by providers to social users.Therefore, many researchers begin to focus on the reform of social system architecture to solve privacy and security issues and have proposed the Decentralized Online Social Network which is based on P2 P, DOSN for short. This novel architecture with excellent characteristics of privacy-preserving is believed as the development tendency of online social networks and attracts more and more researchers' attention. At present, major focus of interrelated work is about the study of storage, access control, communication, etc. Specific privacy protection methods aiming at particular application contexts have not been proposed in these models. To make up for the shortage of these researches, this paper mainly regards the privacy security of social location data as research problem in the environment of distributed social network and introduces the thought of grid-based spatial anonymity into studies of privacy protection of social location data, then proposes a new-type distributed privacy preserving method for location data and a novel social network model with good performance of privacy preserving for social location data. This work has forward-looking significant influences on further researches of distributed social network model and privacy protection.Based on above issues, research work of this paper mainly concentrates on the following aspects:(1) Based on analysis and summary of existing social network architectures systematically according to discrepancies of social service modes, this paper divides online social networks into three modes: the Centralized Online Social Network(COSN), the Decentralized Online Social Network(DOSN) and the Semi-Decentralized Online Social Networks(Semi-DOSN). This work will lay a good foundation for further study of DOSN.(2) In order to solve privacy issues more efficiently, this paper proposes a novel privacy-preserving method for protecting location data in DOSN based on grid-based spatial anonymity, called ClientGrid. It maps primitive user locations into the grid spatial through grid-based division and improves security assurances of location data through public cryptography mechanisms. ClientGrid which completely runs on client side is a distributed privacy-preserving method of location data. Simulation results show that ClientGrid can effectively balance the performance of protecting security of location privacy and the availability of anonymous location data.(3) ClientGrid is modified to satisfy demands of centralized mode and to develop a novel trusted privacy-preserving service module for location data, called TGPM. Then we apply this module to our previous academic achievement, HPOSN, which is a semi-distributed social network model, and then propose a novel social network model PP-HPOSN. When PP-HPOSN works in centralized schema, the service module TGPM can safeguard the privacy security of location data. And TGPM can also update grid granularity according to different security levels of privacy preserving in which way PP-HPOSN can provide different level privacy protections for users. When PP-HPOSN works in decentralized schema, ClientGrid is used to keep privacy security of user location data. Simulation results show that PP-HPOSN can solve the LSFP problem steadily, while having good performance of privacy-preserving for location data.