Modelling and analysis of communication traffic heterogeneity in opportunistic networks

In opportunistic networks, direct communication between mobile devices is used to extend the set of services accessible through cellular or WiFi networks. Mobility patterns and their impact in such networks have been extensively studied. In contrast, this has not been the case with communication traffic patterns, where homogeneous traffic between all nodes is usually assumed. This assumption is generally not true, as node mobility and social characteristics can significantly affect the end-to-end traffic demand between them. To this end, in this paper we explore the joint effect of traffic patterns and node mobility on the performance of popular forwarding mechanisms, both analytically and through simulations. Among the different insights stemming from our analysis, we identify

conditions under which heterogeneity renders the added value of using extra relays more/less useful. Furthermore, we confirm the intuition that an increasing amount of heterogeneity closes the performance gap between different forwarding policies, making endtoend routing more challenging in some cases, or less necessary in others. To our best knowledge, this is the first effort to model, analyze, and quantify effects of traffic heterogeneity. We believe this is an important step towards better protocol design and evaluation of the feasibility of applications in opportunistic networks.