Opportunistic Content Distribution

Abstract: In recent decades, communication networks have had a profound effect on society. Wireless communication has affected our lifestyle and altered how humans communicate and the Internet has revolutionized how we access, publish and disseminate information. In recent years we have also witnessed a radical change in how information is generated on the Internet. Today, information is no longer only generated by a small group of professionals but it is created by the users themselves and shared with a broad community with matching interests. This is evident with ”Web 2.0” applications such as blogs, podcasts, YouTube and social platforms like Facebook and Flickr. As a result of these trends, the Internet is today mainly used to provide users with access to contents. With recent advances in mobile platforms, information generation and consumption has spread from personal computers and Internet into people’s palms. This calls for efficient dissemination of information to and from mobile devices.This thesis considers content-centric networking in the context of mobile wireless networks. The main focus is on opportunistic distribution of content where mobile nodes directly exchange content items when they are within communication range. This communication mode enables dissemination of content between mobile nodes without relying on infrastructure, which can be beneficial for several reasons: infrastructure may be absent, overloaded, unreliable, expensive to use, censored or limited to certain users or contents. Opportunistic networking also has different properties than infrastructure based wireless networking, particularly in terms of scalability, locality and dissemination delay.The contributions of this thesis lie in two areas. Firstly we study the feasibility and performance of opportunistic networking among mobile nodes in urban areas using both analytic models and simulations. In particular we study the effect of two enablers of opportunistic networking: cooperation and mobility. By applying models from epidemic modeling, we show that if nodes cooperate by sharing, even in a limited manner, content can spread efficiently in a number of common case scenarios. We also study in detail which aspects of human mobility affect wireless communication and conclude that performance is not very sensitive to accurate estimation of the probability distributions of mobility parameters such as speed and arrival process. Our results however suggest that it is important to capture the scenario and space in which mobility occurs since this may affect performance significantly. Secondly, we present our design and implementation of a middleware architecture for a mobile peer-to-peer content distribution. Our system uses a decentralized content solicitation scheme that allows the distribution of content between mobile devices without requiring Internet connectivity and infrastructure support. Our system is based on the publish/subscribe paradigm and we describe the design and implementation of key components. We evaluate the performance and correctness of the system using both large-scale simulations and small-scale experimentation with our implementation. Finally we present the design and evaluation of an energy-efficient radio subsystem for opportunistic networking.