Abstract EN:

Wireless communicating devices are everywhere and increasingly blend into our everyday lives, they form new opportunistic networks that allow data to flow across often unreliable, unorganized, and heterogeneous wireless networks. By developing new analysis techniques for temporal dynamic graphs, this thesis proposes and implements a strong use-case for opportunistic networks: data offloading. Analyzing real-life connectivity graphs is difficult. In this thesis, we develop the plausible mobility approach, which infers, from a given contact trace, a compatible node mobility. Furthermore, we define reachability graphs that capture space-time connectivity. When applied to common contact traces, they show that acceptable delivery ratios for point-to-point communications are often out of reach, regardless of the DTN routing protocol, but that the size of the space-time dominating set tends to be a small fraction of the total number of nodes. Accordingly, we show how opportunistic networks may be used to significantly offload broadcast traffic in situations were two radio technologies coexist, typically a pervasive, low-bitrate, and expensive radio, alongside a shorter-range, high-bitrate, and cheaper one. The latter forms the opportunistic network that is used for disseminating most of the content, whereas the former serves both as a control channel for monitoring and as a data channel for bridging the connectivity gaps in the opportunistic network. In this thesis we propose Push-and-Track, a mobility-agnostic framework that leverages an opportunistic network to reliably disseminate content to large numbers of mobile nodes, while minimizing the load on the pervasive radio