Colocation aware content sharing in urban transport

People living in urban areas spend a considerable amount of time on public transport. During these periods, opportunities for inter-personal networking present themselves, as many of us now carry electronic devices equipped with Bluetooth or other wireless capabilities. Using these devices, individuals can share content (e.g., music, news or video clips) with fellow travellers that happen to be on the same train or bus. Transferring media takes time; in order to maximise the chances of successfully completing interesting downloads, users should identify neighbours that possess desirable content and who will travel with them for long-enough periods. In this thesis, a peer-to-peer content distribution system for wireless devices is proposed, grounded on three main contributions: (1) a technique to predict colocation durations (2) a mechanism to exclude poorly performing peers and (3) a library advertisement protocol. The prediction scheme works on the observation that people have a high degree of regularity in their movements. Ensuring that content is accurately described and delivered is a challenge in open networks, requiring the use of a trust framework, to avoid devices that do not behave appropriately. Content advertising methodologies are investigated, showing their effect on whether popular material or niche tastes are disseminated. We first validate our assumptions on synthetic and real datasets, particularly movement traces that are comparable to urban environments. We then illustrate real world operation using measurements from mobile devices running our system in the proposed environment. Finally, we demonstrate experimentally on these traces that our content sharing system significantly improves data communication efficiency, and file availability compared to naive approaches.