Social-Based Content Diffusion in Pocket Switched Networks

Pocket switched networks (PSNs) enable mobile users in a social network to opportunistically exchange information by utilizing proximity-based connectivity, e. g., WiFi and Bluetooth. To cope with nonpersistent connectivity in a PSN, previous studies proposed routing protocols to enhance the probability that data would be delivered to the destinations. However, for some applications, such as dissemination of advertising and multimedia content, service providers might not know who would be interested in the content objects (messages); therefore, they tend to propagate the messages to every user in a PSN. To the best of our knowledge, this paper is the first study of how to forward a message to appropriate relay nodes, so that the message can be disseminated to as many users as possible before it expires. We formulate the problem as a forwarding scheduling model and propose a backward induction algorithm called DIFFUSE, which allows each relay to determine its optimal forwarding schedule in a distributed manner in pseudopolynomial time. Our experiment results show that, because DIFFUSE considers both the user's contribution and contact duration, it can increase the number of recipients in a PSN by up to 2.45 times compared with unicast routing protocols.