Opportunistic Networking : Mobility Modeling and Content Distribution

Abstract: We have witnessed two main trends in recent years that have shaped the current state of communication networks. First, the Internet was designed with the initial idea to provide remote access to resources in the network; today it is overwhelmingly being used for content distribution. In addition, the community of content creators has evolved from a small group of professionals into a global community where every user can generate his contents and share it with other users. Second, the proliferation of personal mobile devices, such as smartphones and media tablets, has altered the way people access, create and share information, leading to a significant migration from wired to wireless networks and raising user expectations for ubiquitous connectivity. These trends have incited research on new communication modes and in this thesis we consider a specific mode, namely opportunistic networking.Opportunistic networking is a communication paradigm that utilizes intermittent connectivity between mobile devices to enable communication in infrastructure-less environments, and to provide complementary transport mechanisms in wireless networks where infrastructure is present. The thesis focuses on two main topics: understanding and modeling human mobility, and opportunistic content distribution.Mobility modeling is one of the key issues in opportunistic networking research. First, we discuss the structure of human mobility and introduce a framework to study mobility at different behavioural levels. We propose a queuing model, denoted by meeting-point model, for pedestrian mobility in smaller urban areas, such as city squares, parks, shops or at bus stops. The model is also a contribution to the second topic we address in the thesis, since we will use it to study characteristics of content distribution in smaller areas. We envision this model as a building block in a library of analytical models that would be used to study the performance of pedestrian content distribution in common scenarios of urban mobility. Furthermore, we show how the proposed model can be used to build larger, more complex models. In the area of opportunistic content distribution, we apply both analytical and simulation-based evaluation. We empirically study the performance of epidemic content distribution by using real-life mobility traces and investigate the fitness of a homogeneous stochastic model to capture the epidemic process.In addition, we present the design, implementation and evaluation of a mobile peer-to-peer system for opportunistic networking and discuss some promising application scenarios.