Preserving privacy of contents in Decentralized Online Social Networks

Traditional Online Social Networks (OSNs), which are based on a single service provider, suffer several drawbacks, first of all the privacy issues arising from the delegation of user data to a single entity, the OSN provider. In the last years, Distributed Online Social Networks (DOSNs) have been proposed to shift the control over user data from the OSN provider to the users of the DOSN themselves. Indeed, in contrast to centralized OSNs (such as Facebook), DOSNs are not based on centralized storage services which decide the term of service and the contents shared by the users are stored on the devices of the users themselves. However, the lack of a centralized entity introduces new interesting challenges, like that of guaranteeing the availability of user’s contents when the user disconnects from the DOSN and the privacy of the user’s contents. In this dissertation we investigate the problem of preserving the privacy of the contents shared by users of DOSNs by focusing on two different aspects: i) the need of defining proper privacy policies for content access and ii) the storage (allocation and replication) of these contents on the nodes which build up the DOSN system. When efficiency has to be taken into account, new solutions have to be devised that minimize as much as possible the overhead introduced by the enforcement of the privacy policy and enable a higher contents availability through distribution and replication. Current solutions fall short in meeting the above criteria, while in this dissertation we proposed two approaches which adopt two very different solutions to guarantee the protection of the contents according to the privacy preferences of users and efficiently reduce the overhead introduced by privacy enforcement mechanisms. The proposed approaches are validated by an experimental campaign based on data obtained from a real OSN which have enabled the definition of a set of simulations taking into account realistic scenarios. The experimental results obtained from a set of simulations performed on real traces show the effectiveness of our approaches.