Дисертації з теми "Calcul de réseau"
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-50 дисертацій для дослідження на тему "Calcul de réseau".
Біля кожної праці в переліку літератури доступна кнопка «Додати до бібліографії». Скористайтеся нею – і ми автоматично оформимо бібліографічне посилання на обрану працю в потрібному вам стилі цитування: APA, MLA, «Гарвард», «Чикаго», «Ванкувер» тощо.
Також ви можете завантажити повний текст наукової публікації у форматі «.pdf» та прочитати онлайн анотацію до роботи, якщо відповідні параметри наявні в метаданих.
Переглядайте дисертації для різних дисциплін та оформлюйте правильно вашу бібліографію.
Mangoua, sofack William. "Amélioration des délais de traversée pire cas des réseaux embarqués à l’aide du calcul réseau." Thesis, Toulouse, ISAE, 2014. http://www.theses.fr/2014ESAE0024/document.
Повний текст джерелаThe thesis addresses performance analysis of embedded real time network using network calculus. Network calculus is a theory based on min-plus algebra. We use network calculus to assess the quality of service of a residual flow in two context : aggregation with non-preemptive priority policy and DRR policy. The main contribution concerns the evaluation of residual service, given to each flow. We also present how to handle DRR and non-preemptive priority policy hierrachically
Vienne, Jérôme. "Prédiction de performances d'applications de calcul haute performance sur réseau Infiniband." Phd thesis, Grenoble, 2010. http://www.theses.fr/2010GRENM043.
Повний текст джерелаManufacturers of computer clusters require tools to assist them in making better decisions in terms of architectural design. To address this need, in this thesis work, we focus on the specific issues of estimating computation times and InfiniBand network congestion. These two problems are often dealt with globally. However, an overall approach does not explain the reasons of performance loss related to architectural choices. So our approach was to conduct a more detailed study. In this thesis work, we focus on the following : 1) the estimation of computation time in a Grid, and 2) the estimation of communication times over Infiniband networks. To evaluate the computation time, the proposed approach is based on a static or semi-static analysis of the source code, by cutting it into blocks, before making a micro-benchmarking of these blocks on the targeted architecture. To estimate the communication time, a model of bandwidth sharing for Infiniband networks has been developed, allowing one to predict the impact related to concurrent communications. This model was then incorporated into a simulator to be validated on a set of synthetic communication graphs and on the application Socorro
Vienne, Jérôme. "Prédiction de performances d'applications de calcul haute performance sur réseau Infiniband." Phd thesis, Université de Grenoble, 2010. http://tel.archives-ouvertes.fr/tel-00728156.
Повний текст джерелаDaigmorte, Hugo. "Analyse des interactions entre flux synchrones et flux asynchrones dans les réseaux temps réel." Thesis, Toulouse, ISAE, 2019. http://www.theses.fr/2019ESAE0003/document.
Повний текст джерелаComplex embedded systems (planes, satellites, drones ...) contain more and more calculators. From now on, these are tens or even hundreds of calculators that communicate through a shared network. A function is achieved by the collaboration of a set of devices that exchange a growing number of information. In an embedded real-time context, it must be ensured that these informations exchanged are correct but it must also be ensured that they verify their temporal constraints. From the network point of view, this means that informations must be exchanged respecting their deadlines. This implies being able to upper bound the traversal time of the network of each message in order to verify that it arrives in time. However, as embedded systems are more and more complex and as the amount of information exchanged is constantly increasing, this bound is increasingly difficult to compute. Furthermore, it is important that this upper bound to be the least pessimistic possible to avoid an oversized system.The goal of this work is to develop new methods of analysis in order to be able to compute these bounds.In order to achieve this, we used the Network Calculus method of analysis.This work focuses on the modeling of interactions between synchronous messages and asynchronous messages.The models presented in this work take into account the transmission dates on the network of synchronous messageswhen calculating the upper bounds of traversal time of the asynchronous messages.The main contributions are:1. the presentation of a new way of considering the use of the dates of emission on the CAN bus: the weak synchronization.As well as the complete modeling of such a system and finally the evaluation of the gain provided by this solution.2. a complete modeling of the TTEthernet network allowing to evaluate the impact of the synchronous flows on the traversal time of the asynchronous flows.3. a presentation of the use of synchronization in the TSN network and a complete model for analyzing this new technology
Pekergin, Mehmet Ferhan. "Optimisation combinatoire par le calcul neuronal et parallelisme optimal." Paris 5, 1992. http://www.theses.fr/1992PA05S017.
Повний текст джерелаLe, Potier Serge. "Optimisation stochastique distribuée : le problème de l'affectation dynamique de canal sur un réseau radio cellulaire." Rennes 1, 1999. http://www.theses.fr/1999REN1S151.
Повний текст джерелаBennouas, Toufik. "Modélisation de parcours du Web et calcul de communautés par émergence." Montpellier 2, 2005. http://www.theses.fr/2005MON20203.
Повний текст джерелаFrison, Julien. "Calcul Ab-initio de propriétés de résonances hadroniques légères." Thesis, Aix-Marseille, 2012. http://www.theses.fr/2012AIXM4082.
Повний текст джерелаIt has been established for now 40 years that the strong nuclear interaction can be described by Quantum Chromodynamics (QCD). However, the confinement of quarks in this theory hides its elementary degrees of freedom in low-energy processes, which have to be computed non-pertubatively. Lattice QCD is, to date, the only method allowing to go through such a computation, and is now reaching maturity. However, although the spectrum of stable hadrons is computed with precision, the study of decays is still delicate. We hereby present the application of Lüscher's formalism to the study of strong-decaying resonance properties, along with a precise computation of the decay width in the particular case of the rho meson
Lhoussaine, Cédric. "Réceptivité, mobilité et π-Calcul". Aix-Marseille 1, 2002. http://www.theses.fr/2002AIX11046.
Повний текст джерелаHajjar, Mansour. "Construction d'un calculateur spécialisé pour le calcul de la conductivité électrique d'un réseau de résistances aléatoires." Paris 11, 1987. http://www.theses.fr/1987PA112409.
Повний текст джерелаA special purpose computer for the calculation of the electric conductivity of a random resistor network. The special purpose computer PERCOLA is designed for long numerical simulations on a percolation problem in Statistical Mechanics of disordered media. Our aim is to improve the actual values of the critical exponents characterizing the behavior of random resistance networks at percolation threshold. The architecture of PERCOLA is based on an efficient iterative algorithm used to compute the electric conductivity of such networks. The calculator has the characteristics of a general purpose 64- bit floating point microprogrammable computer that can run programs for various types of problems with a peak performance of 25 Mflops. This high computing speed is a result of the pipeline architecture based on internal parallelism and separately microcode controlled units such as: data memories, a microcode memory, ALUs and multipliers (both WEITEK components), various data paths, a sequencer (ANALOG DEVICES component), address generators and a random number generator. Thus, the special purpose computer runs percolation problem program 10 percent faster than the supercomputer CRAY XMP
Bonté, Eric. "Calcul des extensions dans les théories de défauts en réseau : Application au raisonnement à profondeur variable." Paris 13, 1992. http://www.theses.fr/1992PA132015.
Повний текст джерелаRoyer, Alex. "Amélioration des méthodes de calcul thermique par réseau de neurones dans les chambres de combustion aéronautiques." Electronic Thesis or Diss., Université de Lorraine, 2023. http://www.theses.fr/2023LORR0141.
Повний текст джерелаAeronautical combustion chambers are constrained environments operating at temperatures and pressures that have been steadily increasing over the last decades. Under these conditions, radiative heat transfer is significant and its influence on flame structure and wall temperatures is noticeable. It is therefore essential to be able to accurately model this transfer mode on very heterogeneous industrial cases and on mesh of several hundred million cells. Common resolution methods do not provide satisfactory accuracy in view of the growing needs. A fine resolution and the use of spectral gas models would be necessary but would imply a prohibitive consumption of computing resources.To overcome this problem, we propose in this work a new radiation modelling methodology based on Monte Carlo methods and Bayesian neural networks. The principle of the concept detailed in this manuscript starts with the sampling of reference points randomly drawn in the study domain. The radiative quantity of interest is computed at each of these points using a Monte Carlo method and a fine spectral model. The results of these calculations are then stored in a database on which a neural network is trained using a Levenberg-Marquardt algorithm. The Bayesian paradigm allows us to define the optimal topology of the network and to automatically adapt the set of parameters inherent to the network without user intervention. The trained network allows to reconstitute the flux or flux divergence fields within the domain. After a theoretical description of the notions used in this work, results on academic cases are presented in different configurations. We then demonstrate the ability of the developed method to model radiation in a participatory environment with very high accuracy and low computational costs
Hablot, Ludovic. "Réseau longue distance et application distribuée dans les grilles de calcul : étude et propositions pour une interaction efficace." Phd thesis, Ecole normale supérieure de lyon - ENS LYON, 2009. http://tel.archives-ouvertes.fr/tel-00804813.
Повний текст джерелаLes applications parallèles s'appuient la plupart du temps sur le standard MPI qui fonctionne par passage de message. Initialement destiné aux grappes, celui-ci est toujours utilisé pour programmer les communications des applications s'exécutant sur les grilles. Cela permet la réutilisation d'anciennes applications.
Alors que différents problèmes ont été résolus pour les communications au sein des grappes, le réseau longue distance de la grille pose plusieurs problèmes. Tout d'abord, les messages MPI sont transmis de manière fiable sur le réseau longue distance via le protocole TCP. Or TCP, qui reste le protocole de transport utilisé dans la plupart des grilles, est basé sur un transfert de données à l'aide de flux ; il est donc peu adapté aux communications MPI. Ensuite, la grande latence du réseau longue distance implique des communications et des retransmissions de paquets perdus qui sont coûteuses. Enfin, le débit disponible sur le lien d'accès à ce réseau est généralement inférieur à la somme des débits nécessaires si tous les processus communiquent en même temps sur ce lien. Ceci crée de la congestion à la fois au sein d'une même application et à la fois avec les autres applications qui l'utilisent, et il devient nécessaire de gérer ce goulot d'étranglement.
L'objectif principal de cette thèse est d'étudier en détail les interactions entre les applications parallèles et la couche de transport dans les réseaux longue distance des grilles de calcul, puis de proposer des solutions à ces problèmes.
Mahmino, Ali. "Application du Codage Réseau aux Architectures à Garanties de Qualité de Service (QoS)." Thesis, Toulouse, INPT, 2009. http://www.theses.fr/2009INPT023H/document.
Повний текст джерелаThe Interest of network coding to improve the throughput or to optimize the use of the network capacity was clearly shown in various contexts. Certain work in particular showed that network coding allows to decrease the end-to-end transmission delay (maximum and average) of a package. This is due to the fact that the processing simultaneous of several packages in a coding node allows to reduce the maximum time spent by the packets in the buffers compared to a classical routing. In this thesis, we consider the application of network coding in the context of the networks providing quality-of-service (QoS) guarantees. Our contributions include the following. First, we propose three network coding strategies ensuring a level of QoS guaranteed expressed in terms of end-in-end delay. The first strategy, called "Network-Oriented Strategy (NOS)", is a random coding strategy. This coding strategy simply consists in combining the inputs packets present in the buffer of a node. It allows minimizing the time spent by the packets in the router's buffers. The weak point of this approach, as any random approach, is that it is not completely reliable. The two other strategies suggested implement a reliable strategy by using the concept of generation. The first, called "Flow-Oriented Strategy (FOS)" is based on the traditional definition of network coding whereas the second, called "Fast Forwarding Strategy (FFS)", allows reducing the packet's buffering delays by transferring them without awaiting all packets of the same generation. The maximum delays generated by different strategies have been evaluated at a coding node level by using network calculus. The end-to-end delay bounds have been then calculated for several types of networks. In most cases, these bounds are better than those obtained for the classical routing. The reliable network coding strategies and the routing strategy have been implemented and evaluated by simulation on networks studied previously. The results obtained show that the worst cases of end-in-end delays observed have the same behaviors as the calculated maximum theoretical bounds, thus validating the suggested strategies
Bahouche, Mebrouk. "Etude et traçabilité du calibrage "Line-Attenuator Reflect" pour des mesures sous pointes à l'aide de l'analyseur de réseau vectoriel." Paris, Télécom ParisTech, 2010. https://pastel.hal.science/pastel-00567066.
Повний текст джерелаThe LAR (Line-Attenuator-Reflect) calibration procedure is particularly attractive because it is already integrated into modern VNAs and it enables broadband measurements with a limited number of standards. On the other hand, only a few studies concern the traceability assessment of this method. That is the reason why the LNE (Laboratoire National de Métrologie) decided to estimate the traceability and accuracy of S parameters measurement when the LAR calibration method is used. In this context, the thesis can be summarized as follows: 1) Realization of a on wafer calibration kit allowing to execute both the LAR calibration and the Multiline TRL calibration which is considered as the reference calibration for on wafer measurements. 2) Development of a new method in order to estimate the errors due to the fact that the input and output impedances of the standard attenuator is different from 50 Ω. This solution is very efficient and it does not require an accurate determination of the reference impedance of the LAR calibration technique. 3) Development of an original method for determining the input and output impedances of the LAR calibration procedure leading to a precise measurement of the reference impedance. 4) Realization of a calibration kit for users, with three different methods to determine the reference impedance. ● Electrical model of the standard attenuator. ● Polynomial interpolation of the reference impedance measurement. ● Development of a new simplified and low cost technique named the LAR-L procedure. 5) Analysis of errors when the substrate of the calibration kit is different from the substrate of the device under test
Hoxha, Fatmir. "Calcul simultané des racines d'un polynome complexe : contribution à l'algorithmique et mise en oeuvre sur un réseau de processeurs." Toulouse, INPT, 1988. http://www.theses.fr/1988INPT048H.
Повний текст джерелаEstephan, Roger. "Contributions aux méthodes de calcul des groupes et des réseaux de micropieux." Phd thesis, Ecole des Ponts ParisTech, 2003. http://tel.archives-ouvertes.fr/tel-00005723.
Повний текст джерелаLa première partie est consacrée à la définition, la classification et les domaines d'application des micropieux. Les différentes méthodes usuelles de calcul des micropieux (isolés, en groupes ou en réseaux) sont également présentées.
La deuxième partie présente la synthèse de divers essais expérimentaux réalisés, dans le cadre du projet national FOREVER, sur des groupes et/ou des réseaux de micropieux. Ces essais sont réalisés en vraie grandeur ou en modèle réduit (centrifugeuse, chambre d'étalonnage ou cuve expérimentale).
La troisième partie traite de deux approches numériques différentes pour l'étude du comportement des micropieux. La méthode des fonctions de transfert de charge est utilisée à travers le programme de calcul des groupes de pieux (GOUPEG) qui est développé pour tenir compte de l'effet de l'inclinaison des micropieux. Le cas d'un réseau élémentaire de 4 micropieux et d'un groupe équivalent est étudié. Une étude paramétrique sur l'effet de l'inclinaison des micropieux au sein d'un réseau élémentaire est présentée. Enfin, la méthode des éléments finis est utilisée (le logiciel CESAR-LCPC) pour l'étude du comportement d'un réseau élémentaire de 4 micropieux (en double chevalet). Cette dernière approche fournit des résultats intéressants sur le comportement du massif du sol sous les micropieux inclinés.
Dimassi, Mohamed. "Modélisation de la structure poreuse du non-tissé par des techniques de calcul avancé." Thesis, Lille 1, 2009. http://www.theses.fr/2009LIL10017/document.
Повний текст джерелаNon-wowen are materials with a complicate porous structure. lndeed, the tangle of fibers and their consolidation, make the comprehension of their hydraulic properties complicated. After having explored three important tracks of research in this field, we chose the direct observation of the physical phenomenon by image analysis and especially the dynamics of the capil!ary rise in a non-woven material. For that, we set up a test bed using a digital camera and a well-adapted lighting to carry out tests on the non-woven materials. I developed and deployed algorithms, which allow an extraction of data in order to understand and simulate the capillary rise in the porous network ofthe non-woven material. These algorithms can be classified in the following way: -an algorithm for the extraction of metric and geometrical properties of pores contained in a non-woven sample.These data represent the structural parameters. -an algorithm which a!lows the extraction of the dynamic hydraulic property. This dynamic property corresponds to a functional property for these materials intended to be used for industrial and domestic hygiene, -an algorithm that determines the most relevant structural parameters based on a decision tree approach, -an algorithm allowing the determination and the simulation of a very wetting Iiquid's dynamic behaviour within the porous network. This approach draws its originality. from the importance attached to the, experimental model and the rules drawn from the Iatter in order.to build the simulation. ln this work, we start from experimental observatIons to deduce an algonthmlc descnptIon faithful to reality.The approach is supported by existing theones
Goglin, Brice. "Réseaux rapides et stockage distribué dans les grappes de calculateurs : propositions pour une interaction efficace." Lyon, École normale supérieure (sciences), 2005. http://www.theses.fr/2005ENSL0328.
Повний текст джерелаThis work aims at studying the exploitation of high-speed networks of clusters for distributed storage. Parallel applications running on clusters require both high-performance communications between nodes and efficient access to the storage system. Many studies on network technologies led to the design of dedicated architectures for clusters with very fast communications between computing nodes. Efficient distributed storage in clusters have been essentially developed by adding parallelization mechanisms so that the server(s) may sustain an increased workload. In this work, we propose to improve the performance of distributed storage systems in clusters by efficiently using the underlying high-performance network to access distant storage systems. The main question we are addressing is: do high-speed networks of clusters fit the requirements of a transparent, efficient and high-performance access to remote storage? We show that storage requirements are very different from those of parallel computation. High-speed networks of clusters were designed to optimize communications between different nodes of a parallel application. We study their utilization in a very different context, storage in clusters, where client-server models are generally used to access remote storage (for instance NFS, PVFS or Lustre). Our experimental study based on the usage of the GM programming interface of Myrinet high-speed networks for distributed storage did raised several interesting problems. Firstly, the specific memory utilization in the storage access system layers does not easily fit the traditional memory model of high-speed networks. Secondly, client-server models that are used for distributed storage have specific requirements on message control and event processing, which are not handled by existing interfaces. We propose different solutions to solve communication control problems at the file-system level. We show that a modification of the network programming interface is required. Data transfer issues need an adaptation of the operating system. We detail several propositions for network programming interfaces which make their utilization easier in the context of distributed storage. The integration of a flexible processing of data transfer in the new programming interface Myrinet/MX is finally presented. Performance evaluations show that its usage in the context of both storage and other types of applications is easy and efficient
Kovacs, Yves. "Modèles de simulation d'écoulement transitoire en réseau d'assainissement." Phd thesis, Ecole Nationale des Ponts et Chaussées, 1988. http://tel.archives-ouvertes.fr/tel-00520785.
Повний текст джерелаDiouri, Idriss. "Propositions de méthodes pour adapter le réseau aux contraintes d'applicatons temps-réel." Thesis, Nancy 1, 2010. http://www.theses.fr/2010NAN10091/document.
Повний текст джерелаThe study of the Networked Control Systems (NCS) is based both on the identification of the application functioning requirements called Quality of Control (QoC) and on the evaluation of the Quality of Service (QoS) offered by the network. The studies on the NCS are classified according to two approaches: the control over network and the control of network. This thesis addresses the second approach and models the scheduling mechanisms implemented in the Ethernet switches that are more and more used in the industrial applications. The specific QoS parameter studied in this thesis is the delay disturbing the controlled system. This thesis proposes two models of classification of service based on WRR (Weighted Round Robin) schedulers. The first modeling follows a constructive approach by using the network calculus theory. The second is based on an identification step from numerical simulations and from the fuzzy logic. In the two cases, the purpose is both to offer enough bandwidth for the time constrained traffic and to maximize the bandwidth dedicated to the others traffics to avoid famine effects. The network calculus approach is used to configure off-line the network in respecting the NCS strict time constraints. The solution based on the fuzzy logic enables a dynamic control of the scheduler in order to tune on-line the network according to the traffic variations. This latter can be applied only to NCS with soft time constraints
Cîrstea, Bogdan-Ionut. "Contribution à la reconnaissance de l'écriture manuscrite en utilisant des réseaux de neurones profonds et le calcul quantique." Electronic Thesis or Diss., Paris, ENST, 2018. http://www.theses.fr/2018ENST0059.
Повний текст джерелаIn this thesis, we provide several contributions from the fields of deep learning and quantum computation to handwriting recognition. We begin by integrating some of the more recent deep learning techniques (such as dropout, batch normalization and different activation functions) into convolutional neural networks and show improved performance on the well-known MNIST dataset. We then propose Tied Spatial Transformer Networks (TSTNs), a variant of Spatial Transformer Networks (STNs) with shared weights, as well as different training variants of the TSTN. We show improved performance on a distorted variant of the MNIST dataset. In another work, we compare the performance of Associative Long Short-Term Memory (ALSTM), a recently introduced recurrent neural network (RNN) architecture, against Long Short-Term Memory (LSTM), on the Arabic handwriting recognition IFN-ENIT dataset. Finally, we propose a neural network architecture, which we name a hybrid classical-quantum neural network, which can integrate and take advantage of quantum computing. While our simulations are performed using classical computation (on a GPU), our results on the Fashion-MNIST dataset suggest that exponential improvements in computational requirements might be achievable, especially for recurrent neural networks trained for sequence classification
Baron, Rémi. "Chromodynamique quantique sur réseau et propriétés du nucléon." Phd thesis, Université Paris Sud - Paris XI, 2009. http://tel.archives-ouvertes.fr/tel-00439372.
Повний текст джерелаDiakhaté, François. "Contribution à l'élaboration de supports exécutifs exploitant la virtualisation pour le calcul hautes performances." Phd thesis, Université Sciences et Technologies - Bordeaux I, 2010. http://tel.archives-ouvertes.fr/tel-00798832.
Повний текст джерелаDiouri, Idriss. "Propositions de méthodes pour adapter le réseau aux contraintes d'applications temps-réel." Phd thesis, Université Henri Poincaré - Nancy I, 2010. http://tel.archives-ouvertes.fr/tel-00610261.
Повний текст джерелаCraciun, Diana. "MODELISATION DES EQUIVALENTS DYNAMIQUES DES RESEAUX ELECTRIQUES." Phd thesis, Université de Grenoble, 2010. http://tel.archives-ouvertes.fr/tel-00602912.
Повний текст джерелаPruvost, Florent. "Méthodes numériques parallèles pour la simulation des réseaux électriques de grandes tailles." Thesis, Châtenay-Malabry, Ecole centrale de Paris, 2012. http://www.theses.fr/2012ECAP0011.
Повний текст джерелаPower system transient stability analysis enables to control the return to equilibrium of the system subjected to a disturbance. This systematic analysis of developing transport networks allows to optimize the production and the consumption of electric power and to protect the equipments such as power plants, transformers, highvoltage lines and so on. In order to improve the stability, the robustness, and the sustainability of these systems, a worldwide trend is to interconnect regional and national transport networks. This leads to analyze ever larger systems. The power-stability problem can be numerically simulated owing to the integration of a differential-algebraic system which is nonlinear and stiff. When considering a very large problem, numerical simulation is very time consuming and significantly slows down the work of professionals. This thesis aims at studying innovative parallel computing methods for the resolution of differential systems arising from the transient stability analysis of large power systems such as the European Transport Network. In this manuscript, we first deliver an analysis of the properties of these rather specific systems: sparse, irregular, nonlinear, stiff, and heterogeneous. We discuss the particular structure of these systems making the application of a domain decomposition method interesting. Thus, we study several space parallelization methods: the fine parallelization of each costly tasks, the resolution of the nonlinear system by decomposition into weakly coupled subnetworks, first on each integration step separately, and then by waveform relaxation method. We also address the time parallelization with a Parareal-based algorithm and a space-time parallel method which benefits from the coupled properties of waveform relaxation and Parareal methods. In this work, we focus on methods which ensure a fast convergence of domain decomposition methods whatever the number of subdomains/processors used. In order to achieve such a goal, we introduce space preconditioning techniques to improve the scalability of the parallelization methods considered
Seyedi, Seyed Mohammad. "Formation, propagation et coalescence dans un réseau de fissures en fatigue thermique." Phd thesis, École normale supérieure de Cachan - ENS Cachan, 2004. http://tel.archives-ouvertes.fr/tel-00133778.
Повний текст джерелаBichler, Olivier. "Contribution à la conception d'architecture de calcul auto-adaptative intégrant des nanocomposants neuromorphiques et applications potentielles." Phd thesis, Université Paris Sud - Paris XI, 2012. http://tel.archives-ouvertes.fr/tel-00781811.
Повний текст джерелаDiouri, Idriss. "Propositions de méthodes pour adapter le réseau aux contraintes d'applicatons temps-réel." Electronic Thesis or Diss., Nancy 1, 2010. http://www.theses.fr/2010NAN10091.
Повний текст джерелаThe study of the Networked Control Systems (NCS) is based both on the identification of the application functioning requirements called Quality of Control (QoC) and on the evaluation of the Quality of Service (QoS) offered by the network. The studies on the NCS are classified according to two approaches: the control over network and the control of network. This thesis addresses the second approach and models the scheduling mechanisms implemented in the Ethernet switches that are more and more used in the industrial applications. The specific QoS parameter studied in this thesis is the delay disturbing the controlled system. This thesis proposes two models of classification of service based on WRR (Weighted Round Robin) schedulers. The first modeling follows a constructive approach by using the network calculus theory. The second is based on an identification step from numerical simulations and from the fuzzy logic. In the two cases, the purpose is both to offer enough bandwidth for the time constrained traffic and to maximize the bandwidth dedicated to the others traffics to avoid famine effects. The network calculus approach is used to configure off-line the network in respecting the NCS strict time constraints. The solution based on the fuzzy logic enables a dynamic control of the scheduler in order to tune on-line the network according to the traffic variations. This latter can be applied only to NCS with soft time constraints
Zammar, Chadi. "Interactions coopératives 3D distantes en environnements virtuels : gestion des problèmes réseau." Phd thesis, INSA de Rennes, 2005. http://tel.archives-ouvertes.fr/tel-00908525.
Повний текст джерелаJurczuk, Krzysztof. "Calcul parallèle pour la modélisation d'images de résonance magnétique nucléaire." Thesis, Rennes 1, 2013. http://www.theses.fr/2013REN1S089.
Повний текст джерелаThis PhD thesis concerns computer modeling of magnetic resonance imaging (MRI). The main attention is centered on imaging of vascular structures. Such imaging is influenced not only by vascular geometries but also by blood flow which has to been taken into account in modeling. Next to the question about the quality of developed models, the challenge lies also in the demand for high performance computing. Thus, in order to manage computationally complex problems, parallel computing is in use. In the thesis three solutions are proposed. The first one concerns parallel algorithms of vascular network modeling. Algorithms for different architectures are proposed. The first algorithm is based on the message passing model and thus, it is suited for distributed memory architectures. It parallelizes the process of connecting new parts of tissue to existing vascular structures. The second algorithm is designed for shared memory machines. It also parallelizes the perfusion process, but individual processors perform calculations concerning different vascular trees. The third algorithm combines message passing and shared memory approaches providing solutions for hybrid parallel architectures. Developed algorithms are able to substantially speed up the time-demanded simulations of growth of complex vascular networks. As a result, more elaborate and precise vascular structures can be simulated in a reasonable period of time. It can also help to extend the vascular model and to test multiple sets of parameters. Secondly, a new approach in computational modeling of magnetic resonance (MR) flow imaging is proposed. The approach combines the flow computation by lattice Boltzmann method, MRI simulation by following discrete local magnetizations in time and a new magnetization transport algorithm together. Results demonstrate that such an approach is able to naturally incorporate the flow influence in MRI modeling. As a result, in the proposed model, no additional mechanism (unlike in prior works) is needed to consider flow artifacts, what implies its easy extensibility. In combination with its low computational complexity and efficient implementation, the solution is a user-friendly and manageable at different levels tool which facilitates running series of simulations with different physiological and imaging parameters. The goal of the third solution is to apply the proposed MR flow imaging model on complex vascular networks. To this aim, models of vascular networks, flow behavior and MRI are combined together. In all the model components, computations are adapted to be performed at various parallel architectures. The model potential and possibilities of simulations of flow and MRI in complex vascular structures are shown. The model aims at explaining and exploring MR image formation and appearance by the combined knowledge from many processes and systems, starting from vascular geometry, through flow patterns and ending on imaging technology
Pawlowski, Filip igor. "High-performance dense tensor and sparse matrix kernels for machine learning." Thesis, Lyon, 2020. http://www.theses.fr/2020LYSEN081.
Повний текст джерелаIn this thesis, we develop high performance algorithms for certain computations involving dense tensors and sparse matrices. We address kernel operations that are useful for machine learning tasks, such as inference with deep neural networks (DNNs). We develop data structures and techniques to reduce memory use, to improve data locality and hence to improve cache reuse of the kernel operations. We design both sequential and shared-memory parallel algorithms. In the first part of the thesis we focus on dense tensors kernels. Tensor kernels include the tensor--vector multiplication (TVM), tensor--matrix multiplication (TMM), and tensor--tensor multiplication (TTM). Among these, TVM is the most bandwidth-bound and constitutes a building block for many algorithms. We focus on this operation and develop a data structure and sequential and parallel algorithms for it. We propose a novel data structure which stores the tensor as blocks, which are ordered using the space-filling curve known as the Morton curve (or Z-curve). The key idea consists of dividing the tensor into blocks small enough to fit cache, and storing them according to the Morton order, while keeping a simple, multi-dimensional order on the individual elements within them. Thus, high performance BLAS routines can be used as microkernels for each block. We evaluate our techniques on a set of experiments. The results not only demonstrate superior performance of the proposed approach over the state-of-the-art variants by up to 18%, but also show that the proposed approach induces 71% less sample standard deviation for the TVM across the d possible modes. Finally, we show that our data structure naturally expands to other tensor kernels by demonstrating that it yields up to 38% higher performance for the higher-order power method. Finally, we investigate shared-memory parallel TVM algorithms which use the proposed data structure. Several alternative parallel algorithms were characterized theoretically and implemented using OpenMP to compare them experimentally. Our results on up to 8 socket systems show near peak performance for the proposed algorithm for 2, 3, 4, and 5-dimensional tensors. In the second part of the thesis, we explore the sparse computations in neural networks focusing on the high-performance sparse deep inference problem. The sparse DNN inference is the task of using sparse DNN networks to classify a batch of data elements forming, in our case, a sparse feature matrix. The performance of sparse inference hinges on efficient parallelization of the sparse matrix--sparse matrix multiplication (SpGEMM) repeated for each layer in the inference function. We first characterize efficient sequential SpGEMM algorithms for our use case. We then introduce the model-parallel inference, which uses a two-dimensional partitioning of the weight matrices obtained using the hypergraph partitioning software. The model-parallel variant uses barriers to synchronize at layers. Finally, we introduce tiling model-parallel and tiling hybrid algorithms, which increase cache reuse between the layers, and use a weak synchronization module to hide load imbalance and synchronization costs. We evaluate our techniques on the large network data from the IEEE HPEC 2019 Graph Challenge on shared-memory systems and report up to 2x times speed-up versus the baseline
Marié, Simon. "Etude de la méthode Boltzmann sur Réseau pour les simulations en aéroacoustique." Phd thesis, Université Pierre et Marie Curie - Paris VI, 2008. http://tel.archives-ouvertes.fr/tel-00311293.
Повний текст джерелаDans un premier temps, les élements historiques et théoriques de la LBM sont présentés ainsi que le développement permettant de passer de l'équation de Boltzmann aux équations de Navier-Stokes. La construction des modèles à vitesses discrètes est également décrite. Deux modèles basés sur des opérateurs de collision différents sont présentés : le modèle LBM-BGK et le modèle LBM-MRT. Pour l'étude des capacités aéroacoustiques de la LBM, une analyse de von Neumann est réalisée pour les modèles LBM-BGK et LBM-MRT ainsi que pour l'équation de Boltzmann à vitesse discrète (DVBE). Une comparaison avec les schémas Navier-Stokes d'ordre élevé est alors menée. Pour remédier aux instabilités numériques de la méthode Boltzmann sur Réseau intervenant lors de la propagation dans des directions particulières à M>0.1, des filtres sélectifs sont utilisés et leur effet sur la dissipation est étudié.
Dans un second temps, le code de calcul L-BEAM est présenté. La structure générale et les différentes techniques de calculs sont décrites. Un algorithme de transition de résolution est développé. La modélisation de la turbulence est abordée et le modèle de Meyers-Sagaut est implémenté dans le code. Enfin, des cas tests numériques sont utilisés pour valider le code et la simulation d'un écoulement turbulent complexe est réalisée.
Gotzig, Bernhard. "Recherche du schéma optimal d'exploitation d'un réseau de distribution électrique." Grenoble INPG, 1997. http://www.theses.fr/1997INPG0209.
Повний текст джерелаThe aim of this thesis is the exploration of the do main of the optimisation of the radial topology of a distribution system for different operation states on a real time basis. We have therefore developed several powerful computation utilities, especially a unified approach which may be used both for the optimisation in the normal state and for finding a restoration plan for the isolated llnfalllted areas in the case of an OLltage of an element in the network. In arder to satisfy the real time constraint, we have L1sed heuristic methods which permit the global optimum to be approached effeciently within a very short computation time. The verification of the results obtained by the heuristic methods is done using optimisation methods which are more dedicated for the research of the global optimum. In validating the proposed method with reallarge scale distribution networks, we have shown the efficiency of our approach in both computation time and the quality of the results
Gamboa, Dos Santos Carlos. "Apport de l'approche gestion de réseaux pour le placement de tâches dans le modèle de programmation par échange de messages." Nancy 1, 1998. http://www.theses.fr/1998NAN10244.
Повний текст джерелаAmari, Ahmed. "Conception et validation d'AeroRing - un réseau de communication Ethernet en double anneau pour les systèmes avioniques de nouvelle génération." Thesis, Toulouse, ISAE, 2017. http://www.theses.fr/2017ESAE0011/document.
Повний текст джерелаThe inherent complexity and bandwidth requirement of avionics communication architecturesare increasing due to the growing number of interconnected end-systems and theexpansion of exchanged data. The Avionics Full Duplex Switched Ethernet (AFDX) has beenintroduced to provide high-speed communication (100Mbps) for new generation aircraft.However, this switched network is deployed in a fully redundant way, which leads to significantquantities of wires, and thus increases weight and integration costs. To cope with thesearising issues, integrating ring-based Ethernet network in avionics context is proposed in thisthesis as a main solution to decrease the wiring-related weight and complexity. In this context,our main objective is to design and validate a new avionic communication network, calledAeroRing, based on a Gigabit Ethernet technology and supporting a Full Duplex ring topology.To achieve this aim, first, a benchmarking of the most relevant Real-Time Ethernet (RTE)solutions supporting ring topologies vs avionics requirements has been conducted, and weparticularly assess the main Performance Indicators (PIs), specified in IEC 61784-2. Thisbenchmarking reveals that each existing RTE solution satisfies some requirements better thanothers, but there is no best solution in terms of all the requirements
Courchelle, Inès de. "Vers une meilleure utilisation des énergies renouvelables : application à des bâtiments scientifiques." Thesis, Toulouse 3, 2017. http://www.theses.fr/2017TOU30196/document.
Повний текст джерелаThe work of this thesis deals with the optimization of energy and computer flows in an intelligent network aiming to supply a data center via renewable energies. In this thesis are treated the problems related to the pooling of energy and computer information in a strong reactivity constraint through the creation of an architecture for an intelligent network. The modeling of such a network must allow the decision making in a dynamic and autonomous way. The objective of this modeling, via an intelligent network, is the optimization of renewable resources in order to reduce the ecological footprint
Fabreges, Xavier. "Etude des propriétés magnétiques et du couplage spin/réseau dans les composés multiferroïques RMnO3 hexagonaux par diffusion de neutrons." Phd thesis, Université Paris Sud - Paris XI, 2010. http://tel.archives-ouvertes.fr/tel-00551271.
Повний текст джерелаTrahay, François. "De l’interaction des communications et de l’ordonnancement de threads au sein des grappes de machines multi-cœurs." Thesis, Bordeaux 1, 2009. http://www.theses.fr/2009BOR13870/document.
Повний текст джерелаThe current trend of constructors for scientific computation is to build clusters whose node include an increasing number of cores.The classical programming model that is only based on MPI is being replaced by hybrid approaches that mix communication and multi-threading. This evolution of the programming model leads to numerous problems since MPI implementations were not designed for multi-threaded applications. In this thesis, in order to guarantee a smooth behavior of communication, we propose a software module that interact with both the threads scheduler and the communication library. This module, by working closely with the thread scheduler, allows to make communication progress in the background and guarantees a high level of reactivity to network events, even when the node is overloaded. We show that this permits to make communication progress in the background and thus to overlap communication and computation. The parallelization of the communication library is also made easier thanks to a task onloading mechanism that is able to exploit the available cores while taking data locality into account. The results we obtain on synthetic application as well as real-life applications show that the interaction between the thread scheduler and the communication library allows to reduce the overhead of communication and thus to improve the application performance
Trahay, François. "De l'interaction des communications et de l'ordonnancement de threads au sein des grappes de machines multi-coeurs." Phd thesis, Université Sciences et Technologies - Bordeaux I, 2009. http://tel.archives-ouvertes.fr/tel-00469488.
Повний текст джерелаDimassi, Mohamed. "Modélisation de la structure poreuse du non-tissé par des techniques de calcul avancé." Electronic Thesis or Diss., Lille 1, 2009. http://www.theses.fr/2009LIL10017.
Повний текст джерелаNon-wowen are materials with a complicate porous structure. lndeed, the tangle of fibers and their consolidation, make the comprehension of their hydraulic properties complicated. After having explored three important tracks of research in this field, we chose the direct observation of the physical phenomenon by image analysis and especially the dynamics of the capil!ary rise in a non-woven material. For that, we set up a test bed using a digital camera and a well-adapted lighting to carry out tests on the non-woven materials. I developed and deployed algorithms, which allow an extraction of data in order to understand and simulate the capillary rise in the porous network ofthe non-woven material. These algorithms can be classified in the following way: -an algorithm for the extraction of metric and geometrical properties of pores contained in a non-woven sample.These data represent the structural parameters. -an algorithm which a!lows the extraction of the dynamic hydraulic property. This dynamic property corresponds to a functional property for these materials intended to be used for industrial and domestic hygiene, -an algorithm that determines the most relevant structural parameters based on a decision tree approach, -an algorithm allowing the determination and the simulation of a very wetting Iiquid's dynamic behaviour within the porous network. This approach draws its originality. from the importance attached to the, experimental model and the rules drawn from the Iatter in order.to build the simulation. ln this work, we start from experimental observatIons to deduce an algonthmlc descnptIon faithful to reality.The approach is supported by existing theones
Ruel, Silvain. "Évaluation des bornes des performances temporelles des Architectures d'Automatisation en Réseau par preuves itératives de propriétés logiques." Phd thesis, École normale supérieure de Cachan - ENS Cachan, 2009. http://tel.archives-ouvertes.fr/tel-00405783.
Повний текст джерелаKanso, Assem. "Evaluation des modèles de calcul des flux polluants des rejets urbains par temps de pluie : Apport de l'approche bayésienne." Phd thesis, Ecole des Ponts ParisTech, 2004. http://pastel.archives-ouvertes.fr/pastel-00001264.
Повний текст джерелаBreck, Damien. "Évaluation des performances du système d'agrégation implémenté dans les réseaux 802.11n." Thesis, Université de Lorraine, 2015. http://www.theses.fr/2015LORR0278/document.
Повний текст джерелаAfter the increase of the capacity of wireless networks, the improvement of the low efficiency allows users to benefit from higher bandwidths. This improvement is partly due to the creation of aggregation's systems. Now, if these systems allow to improve the offered bandwidth, several problems arise about their influence on the behavior of the system. In particular, the case of strong competition for the access to the medium remains unknown. The aggregation studied in this thesis makes a concatenation of several packets in a single frame. This operation allows the packets to share headers. Besides the loss of meaning about the value of a packet, the configuration of such a system remains today obscure for the access provider. As the interests of the provider and its customers may differ, it is also necessary to consider a compromise between their respective requirements. To address these issues, this thesis focuses on a thorough analysis of the behavior and the implications of a packet aggregation system (with IEEE 802.11n as case study). The impact of aggregation on the quality of service provided to an application is quantified through thousands of simulations with Riverbed Modeler. Recommendations are then proposed for the proper setup of the aggregation system to optimize delay and backlog. This thesis focuses on the performances offered to a particular flow in competition for the medium. Its aim is to show the influence of the aggregation on each customer. This thesis also develops an analytical model of the aggregation system using the network calculus theory. The limitations of such a model are demonstrated. Its pessimism is determined using a detailed analysis of the 802.11n aggregation. This study demonstrates a particular relative inadequacy of network calculus to the aggregation system considered. Adjustments of the theory are then proposed, particularly for the calculation of the residuel service offered to a customer. To help the access provider in the configuration of the aggregation system, this thesis lastly proposes an algorithm for a dynamic configuration of the aggregation. This algorithm is established from the recommendations previously established. Its originality is based on the deactivation of the aggregation according to the state of the competing stations. Implemented in Riverbed Modeler, this algorithm shows its efficiency. Then, new problems arise about its real implementation (distributed, centralized, ...)
Moreaud, Stéphanie. "Mouvement de données et placement des tâches pour les communications haute performance sur machines hiérarchiques." Phd thesis, Université Sciences et Technologies - Bordeaux I, 2011. http://tel.archives-ouvertes.fr/tel-00635651.
Повний текст джерелаBrunet, Elisabeth. "Une approche dynamique pour l'optimisation des communications concurrentes sur réseaux hautes performance." Thesis, Bordeaux 1, 2008. http://www.theses.fr/2008BOR13721/document.
Повний текст джерелаThe aim of this thesis is to optimize the communications of high performance applications, in the context of clusters computing. Given the massive use of multicore architectures, it is now crucial to handle a large number of concurrent communication flows. We highlighted and analyzed the shortcomings of existing solutions. We therefore designed a new way to schedule communication flows by focusing on the activity of the network cards. Its novelty consists in untying the activity of applications from that of the network cards. Our model takes advantage of the delay that exists between the deposal of the communication requests and the moment when the network cards become idle in order to apply some opportunistic optimizations. NewMadeleine implements this model, thus making possible to exploit last generation high speed networks. The approach of NewMadeleine is not only validated by synthetical tests but also by real applications
Fournier, Pierre-Luc. "Développement d'un modèle de mesure de la performance d'un processus administratif séquencé et non cadencé dans une organisation du réseau de la santé du Québec par le calcul du degré d'articulation." Thèse, Université du Québec à Trois-Rivières, 2012. http://depot-e.uqtr.ca/6918/1/030586134.pdf.
Повний текст джерелаBenaini, Abdelhamid. "Conception et validation des algorithmes systoliques." Phd thesis, Grenoble INPG, 1988. http://tel.archives-ouvertes.fr/tel-00329564.
Повний текст джерелаYang, Xuecan. "Approximate computing for embedded machine learning." Electronic Thesis or Diss., Institut polytechnique de Paris, 2021. http://www.theses.fr/2021IPPAT005.
Повний текст джерелаConvolutional Neural Networks (CNNs) have been extensively used in many fields such as image recognition, video processing, and naturallanguage processing. However, CNNs are still computational-intensive and resource-consuming. They are often constrained by the limit performanceand memory when deployed on embedded systems. This PhD research project aims at proposing CNNs which are more suitable for embedded systems withlow computing resources and memory requirements. Based on literature review, we propose three methods to accelerate the operation of neural networks : Selective Binarization, Quad-Approx Network and Min- ConvNets. Selective Binarization combines layers with different precisions in CNNs to achieve an acceptable speed and accuracy. As well an FPGA based hardware accelerator is proposed for these optimized structures. With the proposed signed PArameterized Clipping acTivation Function (signed PACT), the CNNs are quantized into 3 bits, and then a loss-less network is established by using approximate multiplier, which is named Quad-Approx Network. In addition to acceleration, what is more valuable is that Quad-Approx shows that CNNs are certain fault tolerance systems, which leads us to propose the MinConvNets. MinConvNet is a set of multiplication-less CNNs whose multiplications are replaced by approximate operations. MinConvNet can achieve negligible loss of prediction compared to exact image classification networks through transfer learning, meanwhile the multiplication which is more resource consuming to implement is replaced by easier implemented operations. Human is ushering the era of the artificial intelligence. In the meantime, the Internet of Things (IoT) makes our lives more convenient. These works bring more complex intelligent algorithms into the edge devices and helps us to create the era of Artificial intelligent Internet of Things (AIoT)