Enabling Energy Efficient Smart Object Networking at Internet-Scale : Experimental Tools, Software Platform, and Information-Centric Networking Protocols. (Internet des Objets : Bases Logicielles et Protocoles Réseaux pour unDéploiement Universel Efficace en Energie)

The Internet of Things aims to seamlessly integrate billions of so-called Smart Objects into traditional Internet infrastructures.From the hardware perspective, Smart Objects emerged when tiny, cheap computers became available, combining energy efficient micro-controllers, low-power radio transceivers, and sensors as well as actuators interacting with the physical world, often powered by batteries.Typically, Smart Objects are thus heavily constrained in terms of CPU, memory and energy resources.Furthermore, wireless links used for communication among Smart Objects or towards the Internet are often slow, subject to high packet loss, and may require spontaneous store-and-forward among peer Smart Objects to ensure connectivity.Such characteristics pose challenges, on one hand in terms of software running on Smart Objects, and on the other hand in terms of network protocols Smart Objects use to communicate.In consequence, novel evaluation methods and experimental tools are needed to study Smart Object networks in vivo, new software platforms are needed to efficiently operate Smart Objects, and innovative networking paradigms and protocols are required to interconnect Smart Objects.This thesis addresses these challenges by introducing new tools for large scale testbed-driven experimental research, a novel operating system (RIOT) applicable to a wide variety of connected Smart Objects, and several new mechanisms leveraging information-centric networking which significantly improve energy-efficiency compared to state-of-the-art network protocols in the Internet of Things.