[pt] A crescente produção de petróleo em águas brasileiras torna cada vez mais importante a gestão logística de alívios das plataformas produtoras. A programação de alívios das plataformas deve ser realizada de forma antecipativa, evitando a parada de produção por falta de espaço disponível para armazenagem. Uma interrupção da produção de petróleo, por menor que seja, causa uma perda direta de receita para a empresa produtora. Alívios realizados muito próximos ao completo enchimento de todos os tanques da plataforma representam risco iminente de perda de produção e são denominados alívios críticos. Este trabalho tem como objetivo realizar um estudo estatístico com dados históricos de 2016 a 2019 para criar um modelo multivariado de previsão dos alívios críticos em uma grande empresa de petróleo brasileira. O modelo de regressão dinâmica foi utilizado para avaliar como as variáveis presentes no processo de programação de alívios se relacionam com o percentual mensal de alívios críticos. A partir do modelo gerado foram identificadas que as variáveis de produção mensal, estoque médio, previsão do tempo, lote médio e exportações mensais impactam no percentual de alívios críticos do mês. Foi realizada uma análise de sensibilidade, a partir da qual foi possível concluir que a gestão de estoques da empresa é o fator fundamental para a redução dos alívios críticos e consequentemente a redução das chances de perda de produção. / [en] The growing oil production in Brazilian waters makes the logistic management of offloadings from the platforms increasingly important. The platform offloading schedule must be carried out in advance, avoiding production stoppage due to lack of available storage space. An interruption in oil production, however small, causes a direct revenue loss for the producing company. Offloadings performed very close to the complete filling of all the platform tanks represent an imminent loss of production risk and are called critical offloadings. This work aims to carry out a statistical study with historical data from 2016 to 2019 to create a multivariate model for forecasting critical offloadings in a large Brazilian oil company. The dynamic regression model was used to evaluate how the variables present in the offloading scheduling process are related to the monthly percentage of critical offloadings. From the developed model, it was identified that the variables of monthly production, average stock, weather forecast, average batch and monthly exports impact the percentage of critical offloadings of the month. A sensitivity analysis was carried out, from which it was possible to conclude that the company s inventory management is the fundamental factor for the reduction of critical offloadings and, consequently, the reduction of the chances of production loss.

In both the private and work spheres, multitasking among three or more activities has become and is continuing to evolve as a pervasive element of everyday life, and recent technological advances only seem to be exacerbating the process. Despite attempts to understand the mental processes that let humans successfully multitask, little is known about the functional cognitive level at which these mental processes take place. This thesis makes a case for the involvement of spatial ability (among other cognitive abilities) in successful multitasking behavior. It focuses on the importance of the cognitive off-loading of executive control demands onto spatial ability, due to the inherent complexity of relationships between task goals and deadlines in multitasking scenarios. Importantly, it presents a working hypothesis—the spatiotemporal hypothesis of multitasking—as a tool for making specific predictions about multitasking performance, based on individual and sex differences in spatial ability. In Study 1, individual differences in spatial ability and executive functions emerged as independent predictors of multitasking performance. When spatial ability was decomposed into its subcomponents, only the coordinate (metric), but not categorical (nonmetric), processing of spatial relations was related to multitasking performance. Males outperformed females in both spatial ability and multitasking, and the effects were moderated by menstrual changes, in that sex differences in coordinate spatial processing and multitasking were observed between males and females in the luteal phase of the menstrual cycle, but not between males and females at menses. In Study II, multitasking performance reflected age- and sex-related differences in executive functioning and spatial ability, suggesting that executive functions contribute to multitasking performance across the adult life span, and that reliance on spatial skills for coordinating deadlines is reduced with advancing age. The results of Study III, in which the spatiotemporal hypothesis was directly scrutinized, suggest that the spatial disruption of multiple deadlines interferes with multitasking performance. Overall, these findings suggest that multitasking performance, under certain conditions, reflects independent contributions of spatial ability and executive functioning. Moreover, the results support the distinction between categorical and coordinate spatial processing, suggesting that these two basic relational processes are selectively affected by female sex hormones and are differentially effective, even across the age span, in transforming and handling temporal patterns as spatial relations in the context of multitasking. Finally, fluctuations of sex hormones exhibit a modulating effect on sex differences in spatial ability and multitasking performance. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 3: Manuscript.</p>

Avec l’avènement du nomadisme, des périphériques mobiles, de la virtualisation et du Cloud Computing ces dernières années, de nouvelles problématiques sont nées aux vues des considérations écologiques, de la gestion d’énergie, de la qualité de service, des normes sécuritaires et bien d’autres aspects liés à nos sociétés. Pour apporter une solution à ces problèmes, nous avons défini la notion de Cloudlet tel un Cloud local où peuvent se virtualiser des périphériques et ses applications embarquées. Ensuite, nous avons conçu une architecture distribuée basée sur ce pattern d’architecture lié au Cloud Computing et à la virtualisation de ressources. Ces définitions permettent de placer notre travail par rapport aux autres approches de déportation d’applications mobiles.D’autre part, un réseau de Cloudlets permet la protection de l’activité effectuée sur un périphérique mobile par la déportation d’applications embarquées dans une machine virtuelle s’exécutant dans la Cloudlet, ainsi que le suivi des usagers dans leur déplacement.Ces définitions nous ont guidées dans l’écriture de spécifications formelles via une algèbre de processus d’ordre supérieure. Elles autorisent le calcul de la sémantique opérationnelle pour les différentes études de cas basées sur ce concept de Cloudlet. Ces spécifications ont permis de décrire une nouvelle vision de la composition des périphériques virtuels applicables à tous les périphériques, les capteurs ou les actuateurs. L’ensemble des équations obtenues constitue une définition formelle de référence non seulement pour le prototypage d’une Cloudlet mais aussi pour la construction des automates temporisés.En se basant sur la structure de nos spécifications, nous avons construit un modèle d’automates temporisés pour un réseau de Cloudlets. Par l’emploi de technique de model checking, nous avons établi des propriétés temporelles montrant que toute exécution d’une application mobile sur un périphérique mobile pouvait être déportée dans une Cloudlet sous condition d’une structure applicative. Ces travaux ont abouti à des choix techniques donnant lieu à un prototype d’une telle architecture distribuée par l’emploi de serveurs OSGi. D’une part, nous fournissons une architecture logicielle d’application mobile. D’autre part, nous mettons en œuvre le principe de migration vers une Cloudlet voisine et son retour. Ces résultats sont une validation de nos choix initiaux et attestent de la réalité de nos travaux. Ils autorisent la prise de mesure permettant de définir le coût d’une migration vers une Cloudlet pendant une exécution, ainsi que son suivi au cours du déplacement de l’usager / With the advent of nomadism, mobile devices, virtualization and cloud computing in recent years, new problems have arisen taking into account ecological concerns, energy management, quality of service, security standards and many other aspects related to our societies. To solve these problems, we define the concept of Cloudlet as a local cloud where virtual devices and embedded applications can be virtualized. Then, we design a distributed architecture based on this architectural pattern related to cloud computing and virtualization of resources. These notions allow us to position our work among other approaches to offload mobile applications in a Cloudlet.On the other hand, a network of Cloudlets helps to secure the activity carried out on a mobile device by offloading embedded applications in a running virtual machine in the Cloudlet, and also to monitor users during their movements.These definitions guided us towards writing formal specifications via a higher order processes of algebra. They facilitate the calculation of operational semantics for different case studies based on this Cloudlet concept. These specifications foster a new vision for designing virtual devices suitable to all devices, sensors or actuators. This set of equations constitutes a formal definition relevant not only for prototyping a Cloudlet but also for constructing a timed automata system.Following the structure of our specifications, we built a model of timed automata for a network of Cloudlets. Exploiting the model checking techniques, we have established temporal properties showing that any execution of a mobile application on a mobile device could be offloaded in a Cloudlet depending on a given software architecture. This work resulted in making technical choices leading to a prototype of such a distributed architecture using an OSGi server. A first result leads us to define a software architecture for mobile applications. Secondly, we implement the principle of migration to a Cloudlet neighbor. Our tests validate our initial choices and confirm the hypotheses of our work. They allow taking measures in order to assess the cost of an offloading to a Cloudlet during runtime, as well as keeping track during user’s movements

Submitted by Luciclea Silva (luci@ufpa.br) on 2018-11-09T17:16:39Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Dissertacao_Estrategiaotimizacaooffloading.pdf: 4353496 bytes, checksum: 660c9fb62874c25c2071d6e88692d9a9 (MD5) / Approved for entry into archive by Luciclea Silva (luci@ufpa.br) on 2018-11-09T17:17:01Z (GMT) No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Dissertacao_Estrategiaotimizacaooffloading.pdf: 4353496 bytes, checksum: 660c9fb62874c25c2071d6e88692d9a9 (MD5) / Made available in DSpace on 2018-11-09T17:17:01Z (GMT). No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Dissertacao_Estrategiaotimizacaooffloading.pdf: 4353496 bytes, checksum: 660c9fb62874c25c2071d6e88692d9a9 (MD5) Previous issue date: 2018-08-31 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / O aumento no consumo de tráfego de dados é motivado pelo aumento do número de dispositivos como smartphone e tablets, já que há uma necessidade de estar conectado com tudo e com todos. As aplicações como streaming de vídeo e jogos online demandam por maior taxa de transmissão de dados, essa alta demanda corrobora para um a sobrecarga das redes móveis baseadas por radiofrequência, de modo a culminar em uma possível escassez do espectro RF. Por tanto, este trabalho busca otimizar o offloading entre LTE e VLC, e para isso é utilizado uma metodologia baseado em aprendizado por reforço denominada de Q-Learning. O algoritmo utiliza como entrada as variáveis do ambiente que estão relacionadas à qualidade do sinal, densidade e velocidade do usuário para aprender e selecionar a melhor conexão. Por tanto, os resultados da simulação mostram a eficiência da metodologia proposta em comparação com o esquema RSS predominante na literatura da área. já que provou por métricas de QoS, suportar maiores taxas de transmissão de dados, assim como, garantiu uma melhoria de 18% em relação as interrupções de serviço a medida que o número de usuários aumenta no sistema. / The increase in the consumption of data traffic is motivated by the increasing number of devices like smartphone and tablets, since there is a need to be connected with everything and with everyone. Applications such as streaming video and online games require a higher rate of data transmission, this high demand corroborates the overload of mobile networks based on radio frequency, so as to culminate in a possible shortage of the RF spectrum. Therefore, this work seeks to optimize offloading between LTE and VLC, and for this a methodology based on reinforcement learning called Q-Learning is used. The algorithm uses as input the environment variables that are related to the signal quality, density and speed of the user to learn and select the best connection. Therefore, the results of the simulation show the efficiency of the proposed methodology in comparison with the predominant RSS scheme in the area literature. as it has been proven by QoS metrics to support higher data rates, as well as ensuring an 18% improvement over service interruptions as the number of users increases in the system.

Mobile cloud computing is a way of delegating complex algorithms from a mobile device to the cloud to complete the tasks quickly and save energy on the mobile device. However, the cloud may not be available or suitable for helping all the time. For example, in a battlefield scenario, the cloud may not be reachable. This work considers neighbouring devices as alternatives to the cloud for offloading computation and presents three key contributions, namely a comprehensive investigation of the trade-off between computation and communication, Multi-Objective Optimisation based approach to offloading, and Queuing Theory based algorithms that present the benefits of offloading to neighbours. Initially, the states of neighbouring devices are considered to be known and the decision of computation offloading is proposed as a multi-objective optimisation problem. Novel Pareto optimal solutions are proposed. The results on a simulated dataset show up to 30% increment in performance even when cloud computing is not available. However, information about the environment is seldom known completely. In Chapter 5, a realistic environment is considered such as delayed node state information and partially connected sensors. The network of sensors is modelled as a network of queues (Open Jackson network). The offloading problem is posed as minimum cost problem and solved using Linear solvers. In addition to the simulated dataset, the proposed solution is tested on a real computer vision dataset. The experiments on the random waypoint dataset showed up to 33% boost on performance whereas in the real dataset, exploiting the temporal and spatial distribution of the targets, a significantly higher increment in performance is achieved.

Device-to-device communication has been suggested as a complement to traditional cellular networks as a means of offloading cellular traffic. In this thesis we explore a solution for device-to-device communication based on opportunistic content distribution in a content-centric network. Communication opportunities arise as mobile nodes roam around in an area and occasionally enter in direct communication range with one another. We consider a node to be a pedestrian equipped with a mobile device and explore the properties of opportunistic communication in the context of content dissemination in urban areas. The contributions of this thesis lie in three areas. We first study human mobility as one of the main enablers of opportunistic communication. We introduce traces collected from a realistic pedestrian mobility simulator and demonstrate that the performance of opportunistic networks is not very sensitive to the accurate estimation of the probability distributions of mobility parameters. However, capturing the space in which mobility occurs may be of high importance. Secondly, we design and implement a middleware for opportunistic content-centric networking, and we evaluate it via a small-scale testbed, as well as through extensive simulations. We conclude that energy-saving mechanisms should be part of the middleware design, while caching should be considered only as an add-on feature. Thirdly, we present and evaluate three different energy-saving mechanisms in the context of opportunistic networking: a dual-radio architecture, an asynchronous duty-cycling scheme, and an energy-aware algorithm which takes into account node selfishness. We evaluate our proposals analytically and via simulations. We demonstrate that when a critical mass of participants is available, the performance of the opportunistic network is comparable to downloading contents directly via the cellular network in terms of energy consumption while offloading large traffic volumes from the operator. / <p>QC 20151120</p>

Modern network interface controllers allow the host to offload packet processing to hardware in order to improve performance. At the present time, the advanced features are utilized in the Linux kernel by offloading the Traffic Control subsystem. Since this subsystem has been designed for a completely different purpose, its usage for hardware offloading is impractical and unreliable. Furthermore, in its current state the subsystem is not capable of utilizing all hardware features, which are often poorly documented. The presented work adopts a different approach to the problem. Five high-end controllers and their packet-processing pipelines were examined in detail. Accounting for their projected future development, common traits and features were identified. The researched information was used to draft a proposal for a new Linux subsystem, more compatible with hardware offloading than the current solution. The proposed subsystem defines a sufficiently descriptive interface to utilize the majority of hardware-offloaded features while avoiding common problems caused by excessively generalized approach of Traffic Control.

COSTA, Philipp Bernardino. Uma abordagem para offloading em múltiplas plataformas móveis. 2014. 104 f. Dissertação (Mestrado em ciência da computação)- Universidade Federal do Ceará, Fortaleza-CE, 2014. / Submitted by Elineudson Ribeiro (elineudsonr@gmail.com) on 2016-07-12T15:14:02Z No. of bitstreams: 1 2014_dis_pbcosta.pdf: 2579064 bytes, checksum: 96c47daf2c7ccea16a7e2f6945cb1a7a (MD5) / Approved for entry into archive by Rocilda Sales (rocilda@ufc.br) on 2016-07-20T13:52:31Z (GMT) No. of bitstreams: 1 2014_dis_pbcosta.pdf: 2579064 bytes, checksum: 96c47daf2c7ccea16a7e2f6945cb1a7a (MD5) / Made available in DSpace on 2016-07-20T13:52:31Z (GMT). No. of bitstreams: 1 2014_dis_pbcosta.pdf: 2579064 bytes, checksum: 96c47daf2c7ccea16a7e2f6945cb1a7a (MD5) Previous issue date: 2014 / The mobile devices, like smartphones and tablets, have evolved considerably in last years in computational terms. Despite advances in their hardware, these devices have energy constraints regarded to their poor computing performance. Therefore, on this context, a new paradigm called Mobile Cloud Computing (MCC) has emerged. MCC studies new ways to extend the computational and energy resources, on mobile devices using the offloading techniques. A literature survey about MCC, has shown that there is no support heterogeneity on reported studies. In response, we propose a framework called MpOS (Multi-platform Offloading System), which supports the offloading technique in mobile application development, for two mobile platforms (Android and Windows Phone). Two case studies were developed with MpOS solution in order to evaluate the framework for each mobile platform. These case studies show how the offloading technique works on several perspectives. In BenchImage experiment, the offloading performance was analyzed, concerning to its execution on a remote execution site (a cloudlet on local network and public cloud in the Internet). The Collision application promotes the analysis of the offloading technique performance on real-time application, also using different serialization systems. In both experiments, results show some situations where it was better to run locally on smarphone, than performing the offloading operation and vice versa. / Os dispositivos móveis, especificamente os smartphones e os tablets, evoluíram bastante em termos computacionais nos últimos anos, e estão cada vez mais presentes no cotidiano das pessoas. Apesar dos avanços tecnológicos, a principal limitação desses dispositivos está relacionada com a questão energética e com seu baixo desempenho computacional, quando comparado com um notebook ou computador de mesa. Com base nesse contexto, surgiu o paradigma do Mobile Cloud Computing (MCC), o qual estuda formas de estender os recursos computacionais e energéticos dos dispositivos móveis através da utilização das técnicas de offloading. A partir do levantamento bibliográfico dos frameworks em MCC verificou-se, para o problema da heterogeneidade em plataformas móveis, ausência de soluções de offloading. Diante deste problema, esta dissertação apresenta um framework denominado de MpOS (Multiplataform Offloading System), que suporta a técnica de offloading, em relação ao desenvolvimento de aplicações para diferentes plataformas móveis, sendo desenvolvido inicialmente para as plataformas Android e Windows Phone. Para validação foram desenvolvidas para cada plataforma móvel, duas aplicações móveis, denominadas de BenchImage e Collision, que demonstram o funcionamento da técnica de offloading em diversos cenários. No caso do experimento realizado com BenchImage foi analisado o desempenho da aplicação móvel, em relação à execução local, no cloudlet server e em uma nuvem pública na Internet, enquanto no experimento do Collision (um aplicativo de tempo real) foi analisado o desempenho do offloading, utilizando também diferentes sistemas de serialização de dados. Em ambos os experimentos houve situações que era mais vantajoso executar localmente no smartphone, do que realizar a operação de offloading e vice-versa, por causa de diversos fatores associados com a qualidade da rede e com volume de processamento exigido nesta operação.

GOMES, Francisco Anderson de Almada. Um serviço de offloading de dados contextuais com suporte à privacidade. 2017. 95 f. Dissertação (Mestrado em Ciência da Computação)- Universidade Federal do Ceará, Fortaleza, 2017. / Submitted by Jonatas Martins (jonatasmartins@lia.ufc.br) on 2017-05-26T13:43:05Z No. of bitstreams: 1 2017_dis_faagomes.pdf: 5747766 bytes, checksum: 678be7c6e0c8e999826aa6d7060bebb5 (MD5) / Approved for entry into archive by Jairo Viana (jairo@ufc.br) on 2017-05-26T17:04:18Z (GMT) No. of bitstreams: 1 2017_dis_faagomes.pdf: 5747766 bytes, checksum: 678be7c6e0c8e999826aa6d7060bebb5 (MD5) / Made available in DSpace on 2017-05-26T17:04:18Z (GMT). No. of bitstreams: 1 2017_dis_faagomes.pdf: 5747766 bytes, checksum: 678be7c6e0c8e999826aa6d7060bebb5 (MD5) Previous issue date: 2017 / Mobile devices became a common tool in our daily routine. Mobile applications are demanding access to contextual information increasingly. For instance, applications require user’s environment data as well as their profiles in order to adapt themselves (interfaces, services, content) according to this context data. Mobile applications with this behavior are known as context-aware applications. Several software infrastructures have been created to help the development of this applications. However, it was verified that most of them do not store history of the contextual data, since mobile devices are resource constrained. They are not built taking into account the privacy of contextual data either, due the fact that applications may expose contextual data without user consent. This dissertation addresses these topics by extending an existing middleware platform that help the development of mobile context-aware applications. This work present a service named COP (Contextual data Offloading service with Privacy Support) and is based in: (i) a context model, (ii) a privacy policy and (iii) synchronization policies. The COP aims to store and process the contextual data generated from several mobile devices, using the computational power of the cloud. To evaluate this work we developed an application that uses both the migration and the privacy mechanism of the contextual data of the COP. Other two experiments were made. The first experiment evaluated the impact of contextual filter processing in mobile device and remote environment, in which the processing time and energy consumption were measured. In this experiment was possible to conclude that the migration of data from mobile device to a remote environment is advantageous. The second experiment evaluated the energy consumption to send contextual data. / Dispositivos móveis tornaram-se uma ferramenta comum no dia a dia das pessoas. Aplicações móveis cada vez mais exigem o acesso às informações contextuais. Por exemplo, aplicações requerem os dados do ambiente do usuário, bem como dos seus perfis, a fim de se adaptarem (interfaces, serviços, conteúdo) de acordo com esses dados de contexto. Aplicações com esse comportamento são conhecidas como aplicações sensíveis ao contexto. Várias infraestruturas de software foram criadas para ajudar no desenvolvimento dessas aplicações. No entanto, foi verificado que a maioria delas não possui um histórico dos dados contextuais, uma vez que os dispositivos móveis são limitados em recursos de armazenamento. Também foi verificado que a maioria delas não é construída levando em conta a privacidade dos dados contextuais, o que pode levar à exposição desses dados sem o consentimento do usuário. Esta dissertação aborda tais tópicos, estendendo uma plataforma de middleware existente que ajuda o desenvolvimento de aplicativos móveis e sensíveis ao contexto. Este trabalho apresenta um serviço denominado COP (Contextual data Offloading service with Privacy Support) e é baseado em: (i) um modelo de contexto, (ii) uma política de privacidade e (iii) em políticas de sincronização de dados. O COP visa armazenar e processar os dados contextuais gerados a partir de vários dispositivos móveis, utilizando o poder computacional da nuvem. Para avaliar este trabalho foi desenvolvida uma aplicação que utiliza tanto a migração como o mecanismo de privacidade dos dados contextuais do COP. Outros dois experimentos foram feitos. O primeiro experimento avaliou o impacto da execução de filtros contextuais no dispositivo móvel e no ambiente remoto, em que foi medido o tempo e gasto energético desse processamento. Nesse experimento foi possível concluir que a migração de dados de um dispositivo móvel para um ambiente remoto é vantajosa. O segundo experimento avaliou o gasto energético para o envio dos dados contextuais.

La popularité des plateformes mobiles telles que smartphones et tablettes génère un volume croissant de données à transférer. La principale raison de cette croissance est l'accès simplifié aux contenus vidéo sur ces plateformes. La future génération (5G) de téléphonie mobile est en cours de développement et a pour objectif d'offrir une bande passante suffisante pour de tels volumes de données. Néanmoins, un déploiement en masse de la 5G n'est pas envisagé avant 2020. De plus, la croissance est telle qu'il sera forcément intéressant de développer des solutions alternatives et complémentaires capables de délester le réseau cellulaire. L'exemple actuel le plus représentatif est le délestage de données cellulaires vers des réseaux d'accès WiFi par les principaux opérateurs mobiles. Dans ce contexte, nous proposons de déployer un nouveau réseau de contenus qui s'appuie sur les réseaux de transports publics urbains. Cette solution déploie des bornes sans-fil dans les bus et sur certaines stations de bus pour offrir du contenu aux passagers des bus. Les bus enregistrent et transportent les données, et se comportent donc comme des mules qui peuvent s'échanger des données dans certaines stations de bus. L'ensemble des bus créé un réseau de transport de données tolérantes au délai telles que de la vidéo à la demande. La création d'un tel réseau soulève de nombreuses questions. Les questions traitées dans les trois parties de cette thèse sont les suivantes: (i) le choix des stations de bus sur lesquelles une borne sans-fil doit être déployée, (ii) le choix du protocole de routage des données, (iii) la gestion efficace de la contention dans les stations et enfin (iv) la réduction du coût d'une telle infrastructure. La première partie de la thèse présente notre réseau de contenu dont l'objectif principal est de transporter de larges volumes de données. Nous montrons pour cela qu'il suffit de déployer des bornes sans-fil aux terminus des lignes de bus. Ce résultat provient de l'analyse des réseaux de transports publics des villes de Toulouse, Helsinki et Paris. Connaissant les horaires et la topologie de ces réseaux de transports, nous proposons de pré-calculer les routes pour transmettre les données dans ce réseau. Nous montrons que ce routage statique permet de réduire drastiquement le nombre de réplications de messages quand on le compare à un routage épidémique. La seconde contribution de cette thèse s'intéresse à l'échange des messages au niveau des bornes sans-fil déployées aux terminus des lignes de bus. En effet, les protocoles d'accès actuels partagent équitablement la bande passante entre les bus et le point d'accès. Dans notre cas, il en résulte une congestion importante que nous proposons de résoudre en introduisant un codage réseau XOR de proche en proche. Les flux qui se croisent sont alors combinés par la borne. Les bus transportent des paquets codés qui seront décodés au prochain saut par la borne suivante. Une analyse théorique de ce mode de communication montre que la probabilité de réception des messages peut-être augmentée au maximum de 50% et la surcharge diminuée au maximum de 50%. Pour les 3 villes européennes considérées, nous montrons par simulation que ce protocole permet d'augmenter de 35% à 48% le nombre de messages reçus. La dernière partie de cette thèse a pour objectif de réduire le coût de déploiement d'une telle architecture. Elle classifie les terminus des lignes de bus en trois ensembles qui sont équipés par des bornes sans fil de nature différentes. Les résultats de simulation montrent que pour les trois villes il est possible de garantir la connectivité de bout-en-bout tout en réduisant les coûts de déploiement d'un facteur 3. Cette architecture, dénommée 3-tier, transporte 30% plus de messages que le déploiement basique proposé en première partie. Nous montrons qu'il est possible de décharger un grand volume de données avec notre architecture. Par exemple, pour Paris, notre architecture permet de / Mobile data traffic is increasing at an exponential rate with the proliferation of mobile devices and easy access to large contents such as video. Traffic demand is expected to soar in the next 5 years and a new generation of mobile networks (5G) is currently being developed to address the looming bandwidth crunch. However, significant 5G deployments are not expected until 2020 or even beyond. As such, any solution that offloads cellular traffic to other available networks is of high interest, the main example being the successful offloading of cellular traffic onto WiFi. In this context, we propose to leverage public transportation networks (PTNs) created by regular bus lines in urban centers to create another offloading option for delay tolerant data such as video on demand. This PhD proposes a novel content delivery infrastructure where wireless access points (APs) are installed on both bus stops and buses. Buses act as data mules, creating a delay tolerant network capable of carrying content users can access while commuting using public transportation. Building such a network raises several core challenges such as: (i) selecting the bus stops on which it is best to install APs, (ii) efficiently routing the data, (iii) relieving congestion points in major hubs and (iv) minimizing the cost of the full architecture. These challenges are addressed in the three parts of this thesis. The first part of the thesis presents our content delivery infrastructure whose primary aim is to carry large volumes of data. We show that it is beneficial to install APs at the end stations of bus lines by analyzing the publicly available time tables of PTN providers of different cities. Knowing the underlying topology and schedule of PTNs, we propose to pre-calculate static routes between stations. This leads to a dramatic decrease in message replications and transfers compared to the state-of-the-art Epidemic delay tolerant protocol. Simulation results for three cities demonstrate that our routing policy increases by 4 to 8 times the number of delivered messages while reducing the overhead ratio. The second part of the thesis addresses the problem of relieving congestion at stations where several bus lines converge and have to exchange data through the AP. The solution proposed leverages XOR network coding where encoding and decoding are performed hop-by-hop for flows crossing at an AP. We conduct a theoretical analysis of the delivery probability and overhead ratio for a general setting. This analysis indicates that the maximum delivery probability is increased by 50% while the overhead ratio is reduced by 50%, if such network coding is applied. Simulations of this general setting corroborate these points, showing, in addition, that the average delay is reduced as well. Introducing our XOR network coding to our content delivery infrastructure using real bus timetables, we demonstrate a 35% - 48% improvement in the number of messages delivered. The third part of the thesis proposes a cost-effective architecture. It classifies PTN bus stops into three categories, each equipped with different types of wireless APs, allowing for a fine-grained cost control. Simulation results demonstrate the viability of our design choices. In particular, the 3-Tier architecture is shown to guarantee end-to-end connectivity and reduce the deployment cost by a factor of 3 while delivering 30% more packets than a baseline architecture. It can offload a large amount of mobile data, as for instance 4.7 terabytes within 12 hours in the Paris topology.