Durant ces dernières années, de nombreux travaux de recherches ont été menés dans le domaine des réseaux multi-sauts sans fil à contraintes (MWNs: Multihop Wireless Networks). Grâce à l'évolution de la technologie des systèmes mico-electro-méchaniques (MEMS) et, depuis peu, les nanotechnologies, les MWNs sont une solution de choix pour une variété de problèmes. Le principal avantage de ces réseaux est leur faible coût de production qui permet de développer des applications ayant un unique cycle de vie. Cependant, si le coût de fabrication des nœuds constituant ce type de réseaux est assez faible, ces nœuds sont aussi limités en capacité en termes de: rayon de transmission radio, bande passante, puissance de calcul, mémoire, énergie, etc. Ainsi, les applications qui visent l'utilisation des MWNs doivent être conçues avec une grande précaution, et plus spécialement la conception de la fonction de routage, vu que les communications radio constituent la tâche la plus consommatrice d'énergie.Le but de cette thèse est d'analyser les différents défis et contraintes qui régissent la conception d'applications utilisant les MWNs. Ces contraintes se répartissent tout le long de la pile protocolaire. On trouve au niveau application des contraintes comme: la qualité de service, la tolérance aux pannes, le modèle de livraison de données au niveau application, etc. Au niveau réseau, on peut citer les problèmes de la dynamicité de la topologie réseau, la présence de trous, la mobilité, etc. Nos contributions dans cette thèse sont centrées sur l'optimisation de la fonction de routage en considérant les besoins de l'application et les contraintes du réseau. Premièrement, nous avons proposé un protocole de routage multi-chemin "en ligne" pour les applications orientées QoS utilisant des réseaux de capteurs multimédia. Ce protocole repose sur la construction de multiples chemins durant la transmission des paquets vers leur destination, c'est-à-dire sans découverte et construction des routes préalables. En permettant des transmissions parallèles, ce protocole améliore la transmission de bout-en-bout en maximisant la bande passante du chemin agrégé et en minimisant les délais. Ainsi, il permet de répondre aux exigences des applications orientées QoS.Deuxièmement, nous avons traité le problème du routage dans les réseaux mobiles tolérants aux délais. Nous avons commencé par étudier la connectivité intermittente entre les différents et nous avons extrait un modèle pour les contacts dans le but pouvoir prédire les future contacts entre les nœuds. En se basant sur ce modèle, nous avons proposé un protocole de routage, qui met à profit la position géographique des nœuds, leurs trajectoires, et la prédiction des futurs contacts dans le but d'améliorer les décisions de routage. Le protocole proposé permet la réduction des délais de bout-en-bout tout en utilisant d'une manière efficace les ressources limitées des nœuds que ce soit en termes de mémoire (pour le stockage des messages dans les files d'attentes) ou la puissance de calcul (pour l'exécution de l'algorithme de prédiction).Finalement, nous avons proposé un mécanisme de contrôle de la topologie avec un algorithme de routage des paquets pour les applications orientés évènement et qui utilisent des réseaux de capteurs sans fil statiques. Le contrôle de la topologie est réalisé à travers l'utilisation d'un algorithme distribué pour l'ordonnancement du cycle de service (sleep/awake). Les paramètres de l'algorithme proposé peuvent être réglés et ajustés en fonction de la taille du voisinage actif désiré (le nombre moyen de voisin actifs pour chaque nœud). Le mécanisme proposé assure un compromis entre le délai pour la notification d'un événement et la consommation d'énergie globale dans le réseau. / Great research efforts have been carried out in the field of challenged multihop wireless networks (MWNs). Thanks to the evolution of the Micro-Electro-Mechanical Systems (MEMS) technology and nanotechnologies, multihop wireless networks have been the solution of choice for a plethora of problems. The main advantage of these networks is their low manufacturing cost that permits one-time application lifecycle. However, if nodes are low-costly to produce, they are also less capable in terms of radio range, bandwidth, processing power, memory, energy, etc. Thus, applications need to be carefully designed and especially the routing task because radio communication is the most energy-consuming functionality and energy is the main issue for challenged multihop wireless networks.The aim of this thesis is to analyse the different challenges that govern the design of challenged multihop wireless networks such as applications challenges in terms of quality of service (QoS), fault-tolerance, data delivery model, etc., but also networking challenges in terms of dynamic network topology, topology voids, etc. Our contributions in this thesis focus on the optimization of routing under different application requirements and network constraints. First, we propose an online multipath routing protocol for QoS-based applications using wireless multimedia sensor networks. The proposed protocol relies on the construction of multiple paths while transmitting data packets to their destination, i.e. without prior topology discovery and path establishment. This protocol achieves parallel transmissions and enhances the end-to-end transmission by maximizing path bandwidth and minimizing the delays, and thus meets the requirements of QoS-based applications. Second, we tackle the problem of routing in mobile delay-tolerant networks by studying the intermittent connectivity of nodes and deriving a contact model in order to forecast future nodes' contacts. Based upon this contact model, we propose a routing protocol that makes use of nodes' locations, nodes' trajectories, and inter-node contact prediction in order to perform forwarding decisions. The proposed routing protocol achieves low end-to-end delays while using efficiently constrained nodes' resources in terms of memory (packet queue occupancy) and processing power (forecasting algorithm). Finally, we present a topology control mechanism along a packet forwarding algorithm for event-driven applications using stationary wireless sensor networks. Topology control is achieved by using a distributed duty-cycle scheduling algorithm. Algorithm parameters can be tuned according to the desired node's awake neighbourhood size. The proposed topology control mechanism ensures trade-off between event-reporting delay and energy consumption.

Several applications have been proposed for wireless sensor networks, including habitat monitoring, structural health monitoring, pipeline monitoring, and precision agriculture. Among the desirable features of wireless sensor networks, one is the ease of deployment. Since the nodes are capable of self-organization, they can be placed easily in areas that are otherwise inaccessible to or impractical for other types of sensing systems. In fact, some have proposed the deployment of wireless sensor networks by dropping nodes from a plane, delivering them in an artillery shell, or launching them via a catapult from onboard a ship. There are also reports of actual aerial deployments, for example the one carried out using an unmanned aerial vehicle (UAV) at a Marine Corps combat centre in California -- the nodes were able to establish a time-synchronized, multi-hop communication network for tracking vehicles that passed along a dirt road. While this has a practical relevance for some civil applications (such as rescue operations), a more realistic deployment involves the careful planning and placement of sensors. Even then, nodes may not be placed optimally to ensure that the network is fully connected and high-quality data pertaining to the phenomena being monitored can be extracted from the network. This work aims to address the problem of random deployment through two complementary approaches: The first approach aims to address the problem of random deployment from a communication perspective. It begins by establishing a comprehensive mathematical model to quantify the energy cost of various concerns of a fully operational wireless sensor network. Based on the analytic model, an energy-efficient topology control protocol is developed. The protocol sets eligibility metric to establish and maintain a multi-hop communication path and to ensure that all nodes exhaust their energy in a uniform manner. The second approach focuses on addressing the problem of imperfect sensing from a signal processing perspective. It investigates the impact of deployment errors (calibration, placement, and orientation errors) on the quality of the sensed data and attempts to identify robust and error-agnostic features. If random placement is unavoidable and dense deployment cannot be supported, robust and error-agnostic features enable one to recognize interesting events from erroneous or imperfect data.

Several applications have been proposed for wireless sensor networks, including habitat monitoring, structural health monitoring, pipeline monitoring, and precision agriculture. Among the desirable features of wireless sensor networks, one is the ease of deployment. Since the nodes are capable of self-organization, they can be placed easily in areas that are otherwise inaccessible to or impractical for other types of sensing systems. In fact, some have proposed the deployment of wireless sensor networks by dropping nodes from a plane, delivering them in an artillery shell, or launching them via a catapult from onboard a ship. There are also reports of actual aerial deployments, for example the one carried out using an unmanned aerial vehicle (UAV) at a Marine Corps combat centre in California -- the nodes were able to establish a time-synchronized, multi-hop communication network for tracking vehicles that passed along a dirt road. While this has a practical relevance for some civil applications (such as rescue operations), a more realistic deployment involves the careful planning and placement of sensors. Even then, nodes may not be placed optimally to ensure that the network is fully connected and high-quality data pertaining to the phenomena being monitored can be extracted from the network. This work aims to address the problem of random deployment through two complementary approaches: The first approach aims to address the problem of random deployment from a communication perspective. It begins by establishing a comprehensive mathematical model to quantify the energy cost of various concerns of a fully operational wireless sensor network. Based on the analytic model, an energy-efficient topology control protocol is developed. The protocol sets eligibility metric to establish and maintain a multi-hop communication path and to ensure that all nodes exhaust their energy in a uniform manner. The second approach focuses on addressing the problem of imperfect sensing from a signal processing perspective. It investigates the impact of deployment errors (calibration, placement, and orientation errors) on the quality of the sensed data and attempts to identify robust and error-agnostic features. If random placement is unavoidable and dense deployment cannot be supported, robust and error-agnostic features enable one to recognize interesting events from erroneous or imperfect data.