Co-evolution of Content Popularity and Delivery in Mobile P2P Networks

Mobile P2P technology provides a scalable approach to content delivery to a large number of users on their mobile devices. In this work, we study the dissemination of a \emph{single} content (e.g., an item of news, a song or a video clip) among a population of mobile nodes. Each node in the population is either a \emph{destination} (interested in the content) or a potential \emph{relay} (not yet interested in the content). There is an interest evolution process by which nodes not yet interested in the content (i.e., relays) can become interested (i.e., become destinations) on learning about the popularity of the content (i.e., the number of already interested nodes). In our work, the interest in the content evolves under the \emph{linear threshold model}. The content is copied between nodes when they make random contact. For this we employ a controlled epidemic spread model. We model the joint evolution of the copying process and the interest evolution process, and derive the joint fluid limit ordinary differential equations. We then study the selection of the parameters under the content provider's control, for the optimization of various objective functions that aim at maximizing content popularity and efficient content delivery.