QoS routing is one of the major building blocks for supporting QoS in communication networks and, hence, a necessary component of future communication networks. Bandwidth- Constrained Routing Algorithm (BCRA) may help to satisfy QoS requirements such as end-to-end delay, delay-jitter etc when WFQ-like (Weighted Fair Queuing) scheduling mechanisms are deployed. The existing algorithms for bandwidth constrained routing suffer from high message overhead and have a high computational and space complexity. The work presented in the thesis, therefore, focuses on the different techniques that an be used to reserve bandwidth for a unicast connection with low protocol overhead in terms of number of messages. We have compared the performance of the proposed routing algorithms using simulation studies with other bandwidth constrained routing algorithms. The call blocking ratio and message overhead have been used as the performance metric to compare the proposed algorithm with the existing ones. We present three source routing algorithms for unicast connections satisfying the band- width requirement. The first two routing algorithms are based on the partitioning of the network. The link-state broadcasts are limited to the partition. In the first algorithm, the source node queries the other partitions for the state information on a connection request and computes the path based on the information received from the other partitions. The second algorithm is based on state aggregation. The aggregated state of other partitions is maintained at every node. The source node finds a feasible path based on the aggregated information. The path is expanded in every partition, if required, at the time of resource reservation. The third QoS routing algorithm uses the Distance Vector Tables to find a route for a connection. If the shortest path satisfies the bandwidth requirement, then it is selected; otherwise a random deviation is taken at the point where bandwidth requirement is not satisfied and shortest path algorithm is again followed. In all the three algorithms presented, the packets carry the entire path information to the destination node. Therefore, no per connection information is required to be maintained at the intermediate nodes. Simulation results indicate that the proposed algorithms indeed help educing the protocol overhead considerably, and at the same time they give comparable or better performance in terms of resource utilization across a wide range of workloads.

Cette thèse apporte des contributions à la modélisation, compréhension ainsi qu'à la communication efficace d'information dans les réseaux dynamiques peuplant la périphérie de l'Internet. Par réseaux dynamiques, nous signifions les réseaux pouvant être modélisés par des graphes dynamiques dans lesquels noeuds et liens évoluent temporellement. Dans la première partie de la thèse, nous proposons un nouveau modèle de mobilité - STEPS - qui permet de capturer un large spectre de comportement de mobilité humains. STEPS mets en oeuvre deux principes fondamentaux de la mobilité humaine : l'attachement préférentiel à une zone de prédilection et l'attraction vers une zone de prédilection. Nous proposons une modélisation markovienne de ce modèle de mobilité. Nous montrons que ce simple modèle paramétrique est capable de capturer les caractéristiques statistiques saillantes de la mobilité humaine comme la distribution des temps d'inter-contacts et de contacts. Dans la deuxième partie, en utilisant STEPS, nous analysons les propriétés comportementales et structurelles fondamentales des réseaux opportunistes. Nous redéfinissons dans le contexte des réseaux dynamiques la notion de structure petit monde et montrons comment une telle structure peut émerger. En particulier, nous montrons que les noeuds fortement dynamiques peuvent jouer le rôle de ponts entre les composants déconnectés, aident à réduire significativement la longueur du chemin caractéristique du réseau et contribuent à l'émergence du phénomène petit-monde dans les réseaux dynamiques. Nous proposons une façon de modéliser ce phénomène sous STEPS. À partir d'un réseau dynamique régulier dans lequel les noeuds limitent leur mobilité à leurs zones préférentielles respectives. Nous recablons ce réseau en injectant progressivement des noeuds nomades se déplaçant entre plusieurs zones. Nous montrons que le pourcentage de tels nœuds nomades est de 10%, le réseau possède une structure petit monde avec un fort taux de clusterisation et un faible longueur du chemin caractéristique. La troisième contribution de cette thèse porte sur l'étude de l'impact du désordre et de l'irrégularité des contacts sur la capacité de communication d'un réseau dynamique. Nous analysons le degré de désordre de réseaux opportunistes réels et montrons que si exploité correctement, celui-ci peut améliorer significativement les performances du routage. Nous introduisons ensuite un modèle permettant de capturer le niveau de désordre d'un réseau dynamique. Nous proposons deux algorithmes simples et efficaces qui exploitent la structure temporelle d'un réseau dynamique pour délivrer les messages avec un bon compromis entre l'usage des ressources et les performances. Les résultats de simulations et analytiques montrent que ce type d'algorithme est plus performant que les approches classiques. Nous mettons également en évidence aussi la structure de réseau pour laquelle ce type d'algorithme atteint ses performances optimum. Basé sur ce résultat théorique nous proposons un nouveau protocole de routage efficace pour les réseaux opportunistes centré sur le contenu. Dans ce protocole, les noeuds maintiennent, via leurs contacts opportunistes, une fonction d'utilité qui résume leur proximité spatio-temporelle par rapport aux autres noeuds. En conséquence, router dans un tel contexte se résume à suivre le gradient de plus grande pente conduisant vers le noeud destination. Cette propriété induit un algorithme de routage simple et efficace qui peut être utilisé aussi bien dans un contexte d'adressage IP que de réseau centré sur les contenus. Les résultats de simulation montrent que ce protocole superforme les protocoles de routage classiques déjà définis pour les réseaux opportunistes. La dernière contribution de cette thèse consiste à mettre en évidence une application potentielle des réseaux dynamiques dans le contexte du " mobile cloud computing ". En utilisant les techniques d'optimisation particulaires, nous montrons que la mobilité peut augmenter considérablement la capacité de calcul des réseaux dynamiques. De plus, nous montrons que la structure dynamique du réseau a un fort impact sur sa capacité de calcul.

La perspective de technologies nanométriques permet d'envisager l'avènement de processeurs constitués de centaines de coeurs de calcul. Néanmoins, l'utilisation de ces processeurs nécessitera de pallier aux problèmes de fiabilité et de variabilité inhérents à ces procédés de fabrication agressifs. Dans cette thèse, nous présentons un ensemble cohérent de techniques pour l'utilisation de processeurs multi-coeurs massivement parallèles, soumis à de forts taux de variabilité et de défaillance. Tout d' abord, la fiabilité du réseau d'interconnexion est abordée, avec la présentation de plusieurs algorithmes de routage tolérants aux fautes, sans interblocages et sans table de routage pour une meilleure scalabilité. Les différentes variantes de ces algorithmes permettent d'ajuster la complexité du réseau sur puce, en fonction des besoins en fiabilité des applications. A titre d'exemple, le plus performant des algorithmes de routage peut acheminer les paquets tant qu'il existe un chemin sans défaillance, et ce jusqu'à 40% de ressources défectueuses. Plusieurs évolutions ont également été étudiées afin d'améliorer les performances du réseau en présence d'un nombre important de fautes. Ensuite, nous proposons une technique auto-adaptative de gestion des applications parallèles, basée sur un routage tolérant aux fautes. L'affectation dynamique des tâches se base sur la recherche adaptative des noeuds de calcul, afin de diminuer la consommation énergétique de l'application en présence de variabilité. Enfin, nous présentons un modèle de simulation de haut-niveau appelé VOCIS (Versatile On-Chip Interconnect Simulator), développé pendant cette thèse. Il permet l'étude approfondie des réseaux d'interconnexion et des routages tolérants aux fautes dans des conditions complexes, afin de répondre aux contraintes propres à ce travail. Nous décrivons son architecture et ses capacités de visualisation. Finalement, nous analysons et illustrons plusieurs résultats expérimentaux originaux obtenus avec ce modèle. / The perspective of nanometric technologies foreshadows the advent of processors consisting of hundreds of computation cores. However, the exploitation of these processors will require to cope with reliability and variability issues inherent to these aggressive manufacturing processes. In this thesis, we present a coherent set of techniques for the utilization of many-cores processors subject to high defect and variability rates. First, the interconnection network reliability is addressed, with the presentation of several deadlock-free fault-tolerant routing algorithms, without routing tables for improving their scalability. The different variants of these algorithms allow for the tune-up of NoC complexity, depending on applications' reliability requirements. For example, the most performant routing algorithm is able to transmit packets as long as a fault-free path exists, with defect rates as high as 40%. Evolutions have also been studied, in order to improve the interconnect performances in the presence of a large number of faults. Second, we propose a self-adaptive technique for the management of parallel applications, based on a fault-tolerant interconnect. The dynamic tasks mapping is based on the adaptive search of computing nodes, in order to reduce the application's energy consumption in the presnece of variability. Third, we present a high-level simulation model named VOCIS (Versatile On-Chip Interconnect Simulator), developed during this thesis. The model allows in-depth study of interconnection networks and fault-tolerant routings under complex settings, in order to meet the specific constraints of this work. The architecture and visualization features are described. Finally, we analyse and illustrate original experimental results obtained with this model.

Made available in DSpace on 2016-06-02T19:05:56Z (GMT). No. of bitstreams: 1 4225.pdf: 931319 bytes, checksum: 2be2e5e88eb314cf8769dabd2749e671 (MD5) Previous issue date: 2011-06-27 / Financiadora de Estudos e Projetos / The usage of wireless sensor network is increasingly being applied to people s everyday lives everywhere: from energy consumption in households and buildings in general, to vital signs in assistive medicine, infrastructure monitoring, chemical or biological product leaking detection in industries, better surveillance, environmental monitoring, among many others. WSN can be deployed in different densities next to several thousands of nodes. However, the development of WSN solutions are limited mainly by energy resource restriction. The great challenge to WSN solutions is to increase the network longevity while guaranteeing data delivery, reliability and accuracy in an environment prone to different types of failures. The largest source of energy consumption is data transmission. Thus, solutions to WSN needs to avoid intense communication keeping energy consumption balance and so the network longevity. In applications in which high density of nodes is necessary, sensing process can produce a large amount of data which are similar or redundant, due to the special proximity among the nodes. This spatial proximity can be explored in routing solutions to reduce the amount of messages transmitted throughout the network. This work presents the Spatial Correlation Aware Routing Protocol - SCARP , which makes use of spatial correlation to reduce the number of network transmissions. With SCARP, the WSN is configured in cells and nodes of each cell are selected, in an alternated way, to transmit similar or redundant data, and so reducing the number of transmitted messages. This traffic reduction results in less energy consumption and longer network longevity. Evaluation results show that SCARP outperforms similar solutions described in the literature, such as DAARP, which uses clustering and aggregation. SCARP has a positive performance even for large node density scenarios. / Redes de Sensores Sem Fio (RSSFs) estão sendo cada vez mais utilizadas na vida diária das pessoas em aplicações que incluem desde monitoramento de gasto de energia em residências e prédios em geral, até monitoramento de sinais vitais para medicina assistida, monitoramento de infraestruturas físicas, vazamentos de produtos químicos ou biológicos em indústrias, vigilância para melhoria de segurança, monitoramento ambiental, dentre inúmeras outras. RSSFs podem ser implantadas em diferentes densidades podendo chegar a milhares de nós. No entanto, o desenvolvimento de soluções baseadas em RSSFs é limitado, principalmente, por recursos restritos dos nós sensores, em especial recursos energéticos. O grande desafio de soluções para RSSFs é aumentar a longevidade da rede e, ao mesmo tempo, garantir a entrega, confiabilidade e precisão dos dados coletados diante de um ambiente propício a falhas de diferentes tipos. A maior fonte de consumo de energia é a transmissão de mensagens. Assim, soluções de RSSF têm que evitar comunicação intensa, mantendo o balanceamento do consumo de energia e, assim, a longevidade da rede. Em aplicações onde é necessária alta densidade de nós sensores, o processo de sensoriamento pode produzir grande quantidade de dados similares ou redundantes devido à proximidade espacial entre esses nós. Esta proximidade espacial pode ser explorada em soluções de roteamento para reduzir a quantidade de mensagens transmitidas pela rede. Este trabalho apresenta o algoritmo de roteamento SCARP (Spatial Correlation Aware Routing Protocol), que faz uso da correlação espacial para reduzir o número de transmissões pela rede. Com o SCARP, a RSSF é configurada em células e nós de cada célula são escolhidos, de maneira alternada, para transmitir dados similares ou redundantes, reduzindo assim o número de mensagens transmitidas. Essa redução de tráfego resulta em menor consumo de energia e maior longevidade da rede. Resultados de avaliação de desempenho mostram que SCARP supera soluções semelhantes descritas na literatura como o DAARP, que utiliza clusterização e agregação de dados, e mantém o desempenho positivo mesmo em situações de grande densidade de nós.

A crescente demanda de tráfego em redes de acesso pressiona a melhor utilização das redes backbone, que são utilizadas para transporte de grandes taxas de dados em diversos domínios (Sistemas Autônomos, SAs). Com o aumento destas redes, aumenta-se a complexidade de topologia das interligações entre domínios. Desta maneira, roteamento de tráfego e pontos de interconexão de SAs (nós de borda) são questões importantes para o desempenho destas redes, que são operadas por diversos provedores que podem utilizar protocolos de comunicação distintos. Neste sentido, o roteamento interdomínio apresenta desafios como a publicação ou não de informações de parâmetros de rede de SAs e como tratar esta questão de maneira globalizada, com novos protocolos e suas especificações. Em termos de pontos de interconexão de SAs, a especificação dos locais onde enlaces inter-redes são conectados aos domínios são importantes para seu desempenho, já que são responsáveis por toda troca de tráfego entre redes distintas. O trabalho considera redes ópticas opacas e translúcidas em cenário multidomínio com bandas multigranulares. Neste cenário é estudado um algoritmo de roteamento multidomínio. No trabalho também é feito um planejamento, especificando em quais nós serão conectados enlaces interdomínio. A principal contribuição deste trabalho é o estudo de planejamento de enlaces interdomínio, com a proposta de um método para escolha de nós de borda (sistematização), com objetivo de diminuir a probabilidade de bloqueio interdomínio. A sistematização é baseada em estudos de resultados de algoritmo genético desenvolvido para o mesmo propósito e sua utilização diminui em até 42% o bloqueio interdomínio. Um algoritmo de alocação de banda também foi desenvolvido para redes multidomínio, que considera parâmetros da camada de rede e óptica para o cálculo de peso de enlaces para encontrar caminhos ópticos entre nós fonte e destino. Os resultados mostram diminuição de até 35% no bloqueio interdomínio com a modificação feita em algoritmo proposto na literatura. / The huge demand for traffic in last mile networks push the better utilization of backbone networks, which are used to transport large data rates in several domains (Autonomous Systems, ASs). With this growth, the topology complexity of interdomain links increases. Then, traffic routing and interconnection points of ASs (border nodes) are relevant questions for the performance of these networks, which are managed by several providers that can use distinct communications protocols. Thus, the interdomain routing presents challenges such as the decision on publishing or not the network´s parameters from ASs and how to deal with this issue in a global way, with new protocols and its specifications. For interconnection points between ASs, the points where interdomain links are connected are important for their performances, since they are responsible for all traffic exchange between distinct networks. This work considers opaque and translucent optical networks in a multidomain scenario with multigranular data rates. In this scenario a multidomain routing algorithm is studied and a network planning is developed, specifying the nodes where interdomain links are connected. The main contribution of this work is the planning of interdomain links, with the proposal of a method for border nodes selection (systematization), with the objective of decreasing the interdomain blocking probability. The systematization is based on the results from a genetic algorithm developed for the same purpose and its utilization decrease up to 42% of the interdomain blocking. A bandwidth allocation algorithm was also created for multidomain scenarios, that considers parameters from network and optical layer for the link weight calculation in order to find optimal paths. The results show a decreasing of up to 35% for interdomain blocking with a contribution based on literature\'s work.

The multi-net Global Routing Problem (GRP) in VLSI physical design is a problem of routing a set of nets subject to limited resources and delay constraints. Various state-of-the-art routers are available but their main focus is to optimize the wire length and minimize the over ow. However optimizing wire length do not necessarily meet timing constraints at the sink nodes. Also, in modern nano-meter scale VLSI process the consideration of process variations is a necessity for ensuring reasonable yield at the fab. In this work, we try to nd a fundamental strategy to address the timing-driven Steiner tree construction (i.e., the routing) problem subject to congestion constraints and process variation. For congestion mitigation, a gradient based concurrent approach (over all nets) of Erzin et. al., rather than the traditional (sequential) rip-and-reroute is adopted in or- der to propagate the timing/delay-driven property of the Steiner tree candidates. The existing sequential rip-up and reroute methods meet the over ow constraint locally but cannot propagate the timing constraint which is non-local in nature. We build on this approach to accommodate the variation-aware statistical delay/timing requirements. To further reduce the congestion, the cost function of the tree generation method is updated by adding history based congestion penalty to the base cost (delay). Iterative use of the timing-driven Steiner tree construction method and history based tree construction procedure generate a diverse pool of candidate Steiner trees for each net. The gradient algorithm picks one tree for each net from the pool of trees such that congestion is e ciently controlled. As the technology scales down, process variation makes process dependent param- eters like resistance, capacitance etc non-deterministic. As a result, Statistical Static Timing Analysis or SSTA has replaced the traditional static timing in nano-meter scale VLSI processes. However, this poses a challenge regarding the max/min-plus algebra of Dijkstra like approximation algorithm that builds the Steiner trees. A new approach based on distance between distributions for nding maximum/minimum at the nodes is presented in this thesis. Under this metric, the approximation algorithm for variation aware timing driven congestion constrained routing is shown to be provably tight and one order of magnitude faster than existing approaches (which are not tight) such as the MVERT. The results (mean value) of our variation aware router are quite close to the mean of the several thousand Monte Carlo simulations of the deterministic router, i.e the results converge in mean. Therefore, instead of running so many deterministic Monte Carlo simulations, we can generate an average design with a probability distribution reasonably close to that of the actual behaviour of the design by running the proposed statistical router only once and at a small fraction of the computational e ort involved in physical design in the nano regime VLSI. The above approximation algorithm is extended to local routing, especially non- Manhattan lambda routing which is increasingly being allowed by the recent VLSI tech- nology nodes. Here also, we can meet delay driven constraints better and keep related wire lengths reasonable.

Synchoros VLSI design style can be an alternative choice to fit the increasing complexity of embedded multi-processor architectures. SiLago Block is part of the synchoros blocks, which can effectively reduce the cost of logic and physical synthesis as it is hardened and highly centralized details from each layer of metal. Global NoCs play an essential part in system-level design and there is necessary to benchmark the SiLago global NoC against other existing NoC libraries. In this degree project, the structure of the NoC is established based on the SiLago models, including the wires and the switches. The whole structure has nine times nine grids and sixteen switches are placed inside symmetrically. The connection between two adjacent switches is built up by wires. The routing algorithm inside the switches can support the most common routing situations by destinations, routing states, and routing history. Except the routing algorithm, this essay provides some deadlock situations and also conclude some ways to solve them. The scripts developed from the NoC generator can be used to do the logical and physical synthesis for the SiLago models. The results from the synthesis can be explored to compare against other methods about the hability to estimate cost metrics from a high level of abstraction and the quality of results. The concept of partition is introduced to accomplish physical synthesis, and through this, the design can be more approach to the core idea of synchoros VLSI design. / Synchoros VLSI designstil kan vara ett alternativt val för att passa den ökande komplexiteten hos inbäddade flerprocessorarkitekturer. SiLago Block är en del av synchoros-blocken, som effektivt kan minska kostnaderna för logik och fysisk syntes eftersom det är härdat och mycket centraliserade detaljer från varje lager av metall. Globala NoC spelar en viktig roll i design på systemnivå och det är nödvändigt att jämföra SiLago globala NoC mot andra befintliga NoC-bibliotek. I detta examensarbete fastställs strukturen för NoC baserat på SiLago-modellerna, inklusive ledningarna och switcharna. Hela strukturen har nio gånger nio rutnät och sexton brytare är placerade inuti symmetriskt. Förbindelsen mellan två intilliggande brytare byggs upp av ledningar. Routingalgoritmen inuti switcharna kan stödja de vanligaste routingsituationerna efter destinationer, routingtillstånd och routinghistorik. Förutom routingalgoritmen ger den här uppsatsen några dödlägessituationer och kommer också fram till några sätt att lösa dem. Skripten som utvecklats från NoC-generatorn kan användas för att göra den logiska och fysiska syntesen för SiLago-modellerna. Resultaten från syntesen kan utforskas för att jämföras med andra metoder om förmågan att uppskatta kostnadsmått från en hög abstraktionsnivå och kvaliteten på resultaten. Begreppet partition introduceras för att åstadkomma fysisk syntes, och genom detta kan designen vara mer förhållningssätt till kärnidén med synchoros VLSI-design.

Cette thèse apporte des contributions à la modélisation, compréhension ainsi qu’à la communication efficace d’information dans les réseaux dynamiques peuplant la périphérie de l’Internet. Par réseaux dynamiques, nous signifions les réseaux pouvant être modélisés par des graphes dynamiques dans lesquels noeuds et liens évoluent temporellement. Dans la première partie de la thèse, nous proposons un nouveau modèle de mobilité - STEPS - qui permet de capturer un large spectre de comportement de mobilité humains. STEPS mets en oeuvre deux principes fondamentaux de la mobilité humaine : l’attachement préférentiel à une zone de prédilection et l’attraction vers une zone de prédilection. Nous proposons une modélisation markovienne de ce modèle de mobilité. Nous montrons que ce simple modèle paramétrique est capable de capturer les caractéristiques statistiques saillantes de la mobilité humaine comme la distribution des temps d’inter-contacts et de contacts. Dans la deuxième partie, en utilisant STEPS, nous analysons les propriétés comportementales et structurelles fondamentales des réseaux opportunistes. Nous redéfinissons dans le contexte des réseaux dynamiques la notion de structure petit monde et montrons comment une telle structure peut émerger. En particulier, nous montrons que les noeuds fortement dynamiques peuvent jouer le rôle de ponts entre les composants déconnectés, aident à réduire significativement la longueur du chemin caractéristique du réseau et contribuent à l’émergence du phénomène petit-monde dans les réseaux dynamiques. Nous proposons une façon de modéliser ce phénomène sous STEPS. À partir d’un réseau dynamique régulier dans lequel les noeuds limitent leur mobilité à leurs zones préférentielles respectives. Nous recablons ce réseau en injectant progressivement des noeuds nomades se déplaçant entre plusieurs zones. Nous montrons que le pourcentage de tels nœuds nomades est de 10%, le réseau possède une structure petit monde avec un fort taux de clusterisation et un faible longueur du chemin caractéristique. La troisième contribution de cette thèse porte sur l’étude de l’impact du désordre et de l’irrégularité des contacts sur la capacité de communication d’un réseau dynamique. Nous analysons le degré de désordre de réseaux opportunistes réels et montrons que si exploité correctement, celui-ci peut améliorer significativement les performances du routage. Nous introduisons ensuite un modèle permettant de capturer le niveau de désordre d’un réseau dynamique. Nous proposons deux algorithmes simples et efficaces qui exploitent la structure temporelle d’un réseau dynamique pour délivrer les messages avec un bon compromis entre l’usage des ressources et les performances. Les résultats de simulations et analytiques montrent que ce type d’algorithme est plus performant que les approches classiques. Nous mettons également en évidence aussi la structure de réseau pour laquelle ce type d’algorithme atteint ses performances optimum. Basé sur ce résultat théorique nous proposons un nouveau protocole de routage efficace pour les réseaux opportunistes centré sur le contenu. Dans ce protocole, les noeuds maintiennent, via leurs contacts opportunistes, une fonction d’utilité qui résume leur proximité spatio-temporelle par rapport aux autres noeuds. En conséquence, router dans un tel contexte se résume à suivre le gradient de plus grande pente conduisant vers le noeud destination. Cette propriété induit un algorithme de routage simple et efficace qui peut être utilisé aussi bien dans un contexte d’adressage IP que de réseau centré sur les contenus. Les résultats de simulation montrent que ce protocole superforme les protocoles de routage classiques déjà définis pour les réseaux opportunistes. La dernière contribution de cette thèse consiste à mettre en évidence une application potentielle des réseaux dynamiques dans le contexte du « mobile cloud computing ». / This thesis contributes to the modeling, understanding and efficient communication in dynamic networks populating the periphery of the Internet. By dynamic networks, we refer to networks that can be modeled by dynamic graphs in which nodes and links change temporally. In the first part of the thesis, we propose a new mobility model - STEPS - which captures a wide spectrum of human mobility behavior. STEPS implements two fundamental principles of human mobility: preferential attachment and attractor. We show that this simple parametric model is able to capture the salient statistical properties of human mobility such as the distribution of inter-contact/contact time. In the second part, using STEPS, we analyze the fundamental behavioral and structural properties of opportunistic networks. We redefine in the context of dynamic networks the concept of small world structure and show how such a structure can emerge. In particular, we show that highly dynamic nodes can play the role of bridges between disconnected components, helping to significantly reduce the length of network path and contribute to the emergence of small-world phenomenon in dynamic networks. We propose a way to model this phenomenon in STEPS. From a regular dynamic network in which nodes limit their mobility to their respective preferential areas. We rewire this network by gradually injecting highly nomadic nodes moving between different areas. We show that when the ratio of such nomadic nodes is around 10%, the network has small world structure with a high degree of clustering and a low characteristic path length. The third contribution of this thesis is the study of the impact of disorder and contact irregularity on the communication capacity of a dynamic network. We analyze the degree of disorder of real opportunistic networks and show that if used correctly, it can significantly improve routing performances. We then introduce a model to capture the degree of disorder in a dynamic network. We propose two simple and efficient algorithms that exploit the temporal structure of a dynamic network to deliver messages with a good tradeoff between resource usage and performance. The simulation and analytical results show that this type of algorithm is more efficient than conventional approaches. We also highlight also the network structure for which this type of algorithm achieves its optimum performance. Based on this theoretical result, we propose a new efficient routing protocol for content centric opportunistic networks. In this protocol, nodes maintain, through their opportunistic contacts, an utility function that summarizes their spatio-temporal proximity to other nodes. As a result, routing in this context consists in following the steepest slopes of the gradient field leading to the destination node. This property leads to a simple and effective algorithm routing that can be used both in the context of IP networks and content centric networks. The simulation results show that this protocol outperforms traditional routing protocols already defined for opportunistic networks. The last contribution of this thesis is to highlight the potential application of dynamic networks in the context of "mobile cloud computing." Using the particle optimization techniques, we show that mobility can significantly increase the processing capacity of dynamic networks. In addition, we show that the dynamic structure of the network has a strong impact on its processing capacity.