WorldCat Identities

Couturier, Raphaël

Works: 58 works in 109 publications in 3 languages and 719 library holdings
Roles: Editor, Opponent, Other, Author, Thesis advisor, Contributor, Publishing director
Publication Timeline
Most widely held works by Raphaël Couturier
Designing scientific applications on GPUs by Raphaël Couturier( )

17 editions published between 2013 and 2014 in English and held by 312 WorldCat member libraries worldwide

"This book covers designs of scientific applications for GPUs, beginning with a review of the principles of GPU programming. It then describes various scientific applications for GPUs and presents lessons learned. Scientific applications covered include computations on matrix operations, linear system solving, nonlinear system solving, image processing, and pseudo random number generation. Expert contributors discuss applications and the GPU porting in a pedagogical way, focusing their attention on the mechanisms they have used to obtain fast and interesting results"--
Parallel iterative algorithms : from sequential to grid computing by Jacques Mohcine Bahi( Book )

18 editions published between 2007 and 2019 in English and Occitan and held by 303 WorldCat member libraries worldwide

"Focusing on grid computing and asynchronism, Parallel Iterative Algorithms: From Sequential to Grid Computing explores the theoretical and practical aspects of parallel numerical algorithms. Each chapter contains a theoretical discussion of the topic, an algorithmic section that fully details implementation examples and specific algorithms, and an evaluation of the advantages and drawbacks of the algorithms. Several exercises also appear at the end of most chapters." "Providing the theoretical and practical knowledge needed to design and implement efficient parallel iterative algorithms, this book illustrates how to apply these algorithms to solve linear and nonlinear numerical problems in parallel environments, including local, distant, homogeneous, and heterogeneous clusters."--BOOK JACKET
Designing scientific applications on GPUs( )

1 edition published in 2014 in English and held by 11 WorldCat member libraries worldwide

PRESENTATION OF GPUs Presentation of the GPU Architecture and the Cuda Environment Raphaël CouturierIntroduction Brief history of video card GPGPU Architecture of current GPUs Kinds of parallelism Cuda multithreading Memory hierarchy Introduction to Cuda Raphaël CouturierIntroduction First example Second example: using CUBLAS Third example: matrix-matrix multiplication IMAGE PROCESSING Setting up the Environment Gilles PerrotData transfers, memory managementPerformance measurements Implementing a Fast Median Filter Gilles PerrotIntroduction Median filteringNVidia GPU tuning recipes A 3x3 media
Parallel iterative algorithms by Jacques Mohcine Bahi( Book )

1 edition published in 2007 in English and held by 3 WorldCat member libraries worldwide

"Providing the theoretical and practical knowledge needed to design and implement efficient parallel iterative algorithms, this book illustrates how to apply these algorithms to solve linear and nonlinear numerical problems in parallel environments, including local, distant, homogeneous, and heterogeneous clusters."--Jacket
Résolution de systèmes linéaires et non linéaires creux sur grappes de GPUs by Lilia Ziane Khodja( Book )

2 editions published in 2013 in French and held by 2 WorldCat member libraries worldwide

Or the past few years, the clusters equipped with GPUs have become attractive tools for high performance computing. In this thesis, we have designed parallel iterative algorithms for solving large sparse linear and nonlinear systems on GPU clusters. First, we have focused on solving sparse linear systems using CG and GMRES iterative methods. The experiments have shown that a GPU cluster is more efficient that its pure CPU counterpart for solving large sparse systems of linear equations. Then, we have implemented the synchronous and asynchronous algorithms of the Richardson and the block relaxation iterative methods for solving sparse nonlinear systems. We have noticed that the best solutions developed for the CPUs are not necessarily well suited to GPUs. Indeed, the experiments performed on a GPU cluster have shown that the parallel algorithms of the Richardson method are far more efficient than those of the block relaxation method. In addition, they have shown that the computing power of GPUs allows to reduce the ratio between the time of the computation over that of the communication, which favors the use of the asynchronous iteration on GPU clusters. Finally, we are interested in geographically distant clusters for solving large sparse linear systems. In this context, we have used a multisplitting two-stage method using parallel GMRES method adapted to GPU clusters. It uses the synchronous iteration to solve locally the sub-linear systems and the asynchronous one to solve the global sparse linear system
Data compression and deep learning for IoT healthcare applications based on physiological signals by Joseph Azar( Book )

2 editions published in 2020 in English and held by 2 WorldCat member libraries worldwide

In recent years, Internet of Things (IoT) technology has gained tremendous attention for its ability to relieve the burden on healthcare caused by an aging population and the increase in chronic disease. IoT technology facilitates the tracking of patients with different conditions and the processing of vast volumes of data, of which a substantial part of this data are physiological signals. Physiological signals are an invaluable source of data which helps to diagnose, rehabilitate, and treat diseases. The signals come from a Wireless Body Sensor Network or wearable devices placed on a patient's body. There are many difficulties in IoT-healthcare systems, such as data collection and processing especially that (1) wireless sensor nodes have limited energy, processing and memory resources, (2) the quantity of data collected periodically is enormous, (3) the quality of the data collected is not always satisfactory, and that such data are highly likely to include noisy or unreliable areas, and (4) the manual feature extraction process from the physiological signals requires significant human intervention and medical expertise.Firstly, an energy-efficient data compression technique is being proposed in this dissertation. The proposed scheme is based on an error-bound lossy compressor originally designed for high-performance computing applications and has been adapted for IoT devices that are resource constrained. The proposed solution is an easy-to-implement algorithm, which could reduce energy consumption by as much as 2.5 times. It also reduces the processing/transmission time by compressing large batches of data before their transfer from the IoT to the edge. Moreover, an empirical analysis was carried out to study the effect of lossy time series compression on the classification task. In addition to different variations of compression methods, various deep neural networks for time series classification were considered to identify the appropriate trade-off between the compression ratio and classification performance.Second, the problem of data distortion due to lossy compression was addressed. The reconstructed data may not always be satisfactory, given that the physiological signals collected are multivariate time series that are highly compressed with a lossy compressor prior to transmission. To solve this limitation, a convolutional autoencoder-based deep learning model was introduced. The proposed model was able to enhance the compressed data after reconstruction, thus fixing the shape of the physiological signal and allowing more accurate feature extraction.Then, the photoplethysmogram signals were given particular attention. Photoplethysmography (PPG) is used to measure the skin blood flow using infrared light. Photoplethysmography is a promising technique because of the capability of the new wrist-worn devices to provide the signal. The existence of motion artifacts and meaningless areas in the signal is a major challenge faced when working with this signal. A deep learning model for automatic motion artifacts detection based on a CNN-LSTM autoencoder architecture has been proposed to detect and discard irrelevant zones in photoplethysmogram signals to avoid analyzing and processing meaningless data.Finally, a deep learning model has been proposed for time series classification. Given that hand-designing features from physiological signals is a challenging task, the representation learning approach is a potential solution for automatically learning features from raw physiological data. A classification model based on the DenseNet architecture was proposed for the classification of multivariate time series. The results show that the proposed approach is capable of achieving and in several cases bypassing the performance of the state-of-the-art models on a benchmark dataset
Efficient and cryptographically secure generation of chaotic pseudorandom numbers on GPU by Christophe Guyeux( )

1 edition published in 2015 in English and held by 2 WorldCat member libraries worldwide

Blind digital watermarking in PDF documents using Spread Transform Dither Modulation by Ahmad W Bitar( )

1 edition published in 2015 in English and held by 2 WorldCat member libraries worldwide

Générateur de coprocesseur pour le traitement de données en flux (vidéo ou similaire) sur FPGA. by Gwenhael Goavec-Merou( Book )

2 editions published in 2014 in French and held by 2 WorldCat member libraries worldwide

Using Field Programmable Gate Arrays (FPGA) is one of the very few solution for real time processingdata flows of several hundreds of Msamples/second. However, using such componentsis technically challenging beyond the need to become familiar with a new kind of dedicateddescription language and ways of describing algorithms, understanding the hardware behaviouris mandatory for implementing efficient processing solutions. In order to circumvent these difficulties,past researches have focused on providing solutions which, starting from a description ofan algorithm in a high-abstraction level language, generetes a description appropriate for FPGAconfiguration. Our contribution, following the strategy of block assembly based on the skeletonmethod, aimed at providing a software environment called CoGen for assembling various implementationsof readily available and validated processing blocks. The resulting processing chainis optimized by including FPGA hardware characteristics, and input and output bandwidths ofeach block in order to provide solution fitting best the requirements and constraints. Each processingblock implementation is either generated automatically or manually, but must complywith some constraints in order to be usable by our tool. In addition, each block developer mustprovide a standardized description of the block including required resources and data processingbandwidth limitations. CoGen then provides to the less experienced user the means to assemblethese blocks ensuring synchronism and consistency of data flow as well as the ability to synthesizethe processing chain in the available hardware resources. This working method has beenapplied to video data flow processing (threshold, contour detection and tuning fork eigenmodesanalysis) and on radiofrequency data flow (wireless interrogation of sensors through a RADARsystem, software processing of a frequency modulated stream, software defined radio)
Energy-efficient secured data reduction technique using image difference function in wireless video sensor networks by Christian Salim( )

1 edition published in 2019 in English and held by 2 WorldCat member libraries worldwide

Data Reduction based energy-efficient approaches for secure priority-based managed wireless video sensor networks by Christian Salim( Book )

2 editions published in 2018 in English and held by 2 WorldCat member libraries worldwide

The huge amount of data in Wireless Video Sensor Networks (WVSNs) for tiny limited resources sensor nodes increases the energy and bandwidth consumption challenges. Controlling the network is one of the challenges in WMSN due to the huge amount of images sent at the same time from the sensors to the coordinator. In this thesis, to overcome these problems, several contributions have been made. Each contribution concentrates on one or two challenges as follows: In the first contribution, to reduce the energy consumption a new approach for data aggregation in WVSN based on shot similarity functions is proposed. It is deployed on two levels: the video-sensor node level and the coordinator level. At the sensor node level, we propose a frame rate adaptation technique and a similarity function to reduce the number of frames sensed by the sensor nodes and sent to the coordinator. At the coordinator level, after receiving shots from different neighboring sensor nodes, the similarity between these shots is computed to eliminate redundancies. In the second contribution, some processing and analysis are added based on the similarity between frames on the sensor-node level to send only the important frames to the coordinator. Kinematic functions are defined to predict the next step of the intrusion and to schedule the monitoring system accordingly. In the third contribution, on the transmission phase, on the sensor-node level, a new algorithm to extract the differences between two images is proposed. This contribution also takes into account the security challenge by adapting an efficient ciphering algorithm on the sensor node level. In the last contribution, to avoid slower detection of intrusions leading to slower reactions from the coordinator, a mac-layer protocol based on S-MAC protocol has been proposed to control the network. This solution consists in adding a priority bit to the S-MAC protocol to give priority to critical data
Fast GPU-based denoising filter using isoline levels by Gilles Perrot( )

1 edition published in 2013 in English and held by 2 WorldCat member libraries worldwide

Multiround Distributed Lifetime Coverage Optimization protocol in wireless sensor networks by Ali Kadhum Idrees( )

1 edition published in 2017 in English and held by 2 WorldCat member libraries worldwide

Algorithmes distribués pour l'optimisation de déploiement des microrobots MEMS by Hicham Lakhlef( Book )

2 editions published in 2014 in French and held by 2 WorldCat member libraries worldwide

Les microrobots MEMS sont des éléments miniaturisés qui peuvent capter et agir sur l'environnement. Leur taille est de l'ordre du millimètre et ils ont une faible capacité de mémoire et une capacité énergétique limitée. Les microrobots MEMS continuent d'accroître leur présence dans notre vie quotidienne. En effet, ils peuvent effectuer plusieurs missions et tâches dans une large gamme d'applications telles que la localisation d'odeur, la lutte contre les incendies, le service médical, la surveillance, le sauvetage et la sécurité. Pour faire ces taches et missions, ils doivent appliquer des protocoles de redéploiement afin de s'adapter aux conditions du travail. Ces algorithmes doivent être efficaces, évolutifs, robustes et ils doivent utiliser de préférence des informations locales. Le redéploiement pour les microrobots MEMS mobiles nécessite actuellement un système de positionnement et une carte (positions prédéfinies) de la forme cible. La solution traditionnelle de positionnement comme l'utilisation d'un GPS consommerait trop d'énergie. De plus, l'utilisation de solutions de positionnement algorithmique avec les techniques de multilatération pose toujours des problèmes à cause des erreurs dans les coordonnées obtenues.Dans la littérature, si nous voulons une auto-reconfiguration de microrobots vers une forme cible constituée de P positions, chaque microrobot doit avoir une capacité mémoire de P positions pour les sauvegarder. Par conséquent, si P est de l'ordre de milliers ou de millions, chaque noeud devra avoir une capacité de mémoire de positions en milliers ou millions. Parconséquent, ces algorithmes ne sont pas extensibles ou évolutifs. Dans cette thèse, on propose des protocoles de reconfiguration où les noeuds ne sont pas conscients de leurs positions dans le plan et n'enregistrent aucune position de la forme cible. En d'autres termes, les noeuds ne stockent pas au départ les coordonnées qui construisent la forme cible. Par conséquent, l'utilisation de mémoire pour chaque noeud est réduite à une complexité constante. L'objectif desalgorithmes distribués proposés est d'optimiser la topologie logique du réseau des microrobots afin de chercher une meilleure complexité pour l'échange de message et une communication peu coûteuse. Ces solutions sont complètement distribués. On montre pour la reconfiguration d'une chaîne à un carré comment gérer la dynamicité du réseau pour sauvegarder l'énergie, on étudie comment utiliser le parallélisme de mouvements pour optimiser le temps d'exécution et lenombre de mouvements. Ainsi, on propose une autre solution où la topologie physique initiale peut être n'importe quelle configuration initiale. Avec ces solutions, les noeuds peuvent exécuter l'algorithme indépendamment du lieu où ils sont déployés, parce que l'algorithme est indépendant de la carte de la forme cible. En outre, ces solutions cherchent à atteindre la forme de la cible avec une quantité minimale de mouvement
Distributed lifetime coverage optimization protocol in wireless sensor networks by Ali Kadhum Idrees( )

1 edition published in 2015 in English and held by 2 WorldCat member libraries worldwide

Energy-efficient data collection and fusion in wireless body sensor networks for continuous health monitoring by Carol Habib( Book )

2 editions published in 2018 in English and held by 2 WorldCat member libraries worldwide

Several challenges exist in Wireless Body Sensor Networks such as the data collection and fusion especially that (1) wireless sensor nodes have limited energy, processing and memory resources, (2) the amount of periodically gathered data is huge, (3) the gathered data are characterized by a heterogeneous nature and (4) the data interpretation to ensure decision-support is influenced byseveral external factors such as the provided context information of the monitored person.In this thesis, the aforementioned challenges were tackled by proposing scientific aproaches. Firstly, an energy-efficient data collection technique is proposed. This technique targets the energy consumed by biosensor nodes for sensing and transmitting vital signs. It consists of a real-timesampling rate adaptation mechanism and a local detection system which are provided at the level of the nodes. Second, in order to perform a health assessment based on the collected data, a multisensor data fusion model is proposed. In this approach, the coordinator of the network performs anassessment of the patient's health condition based on the collected measurements of his/her vital signs. Such data is interpreted in a human-reasoning way and are characterized by ambiguity and imprecision. Thus, we propose to use a Fuzzy Inference System. Then, given that vital signs are highly correlated to the context of the monitored person, a context-aware multi-sensor data fusionmodel for health assessment is proposed. The person's context include his/her physical activity status, medical record and personal information. This information highly influences the interpretation of vital signs. Hesitant fuzzy sets are used to subjectively evaluate the intensity of the person's physical activities based on his/her personal information and the activity's characteristics. Finally, a specific healthcare monitoring application is targeted. A real-time stress detection and evaluation framework is proposed while taking into consideration the energy consumption constraint. Shimmer 3 GSR+ is used as a wireless sensor node to sense the Photoplethysmogram (PPG) signal and the skin conductance. An android mobile application is developed to extract from the PPG signal stress correlated vital signs such as the heart rate, the respiration rate and the blood pressure
Efficient and secure cipher scheme for multimedia contents by Hassan N Noura( )

1 edition published in 2018 in English and held by 2 WorldCat member libraries worldwide

Gestion efficace de données et couverture dans les réseaux de capteurs sans fil by Hassan Moustafa Harb( )

2 editions published in 2016 in English and held by 2 WorldCat member libraries worldwide

Dans cette thèse, nous proposons des techniques de gestion de données pour économiser l'énergie dans les réseaux de capteurs périodiques basés sur l'architecture de clustering. Premièrement, nous proposons d'adapter le taux d'échantillonnage du capteur à la dynamique de la condition surveillée en utilisant le modèle de one-way ANOVA et des tests statistiques (Fisher, Tukey et Bartlett), tout en prenant en compte l'énergie résiduelle du capteur. Le deuxième objectif est d'éliminer les données redondantes générées dans chaque cluster. Au niveau du capteur, chaque capteur cherche la similarité entre les données collectées à chaque période et entre des périodes successives, en utilisant des fonctions de similarité. Au niveau du CH, nous utilisons des fonctions de distance pour permettre CH d'éliminer les ensembles de données redondantes générées par les nœuds voisins. Enfin, nous proposons deux stratégies actif/inactif pour ordonnancer les capteurs dans chaque cluster, après avoir cherché la corrélation spatio-temporelle entre les capteurs. La première stratégie est basée sur le problème de couverture des ensembles tandis que la seconde prend avantages du degré de corrélation et les énergies résiduelles de capteurs pour ordonnancer les nœuds dans chaque cluster. Pour évaluer la performance des techniques proposées, des simulations sur des données de capteurs réelles ont été menées. La performance a été analysée selon la consommation d'énergie, la latence et l'exactitude des données, et la couverture, tout en montrant comment nos techniques peuvent améliorer considérablement les performances des réseaux de capteurs
Utilisation des méthodes formelles pour le développement de programmes parallèles by Raphaël Couturier( Book )

2 editions published in 2000 in French and held by 2 WorldCat member libraries worldwide

Le travail décrit dans cette thèse a pour but d'étudier comment on peut appliquer les méthodes formelles à la parallélisassions, pour développer des programmes parallèles corrects. Comme un de nos objectifs est de travailler sur des applications en grandeur nature, nous avons, durant ce travail, collaboré avec des physiciens et chimistes de notre université afin de paralléliser trois de leurs applications. Ces applications ont été parallélisées, sur l'origin 2000 du centre Charles Hermite, soit avec openmp, soit avec mpi, soit avec ces deux paradigmes à la fois. Afin de prouver qu'une parallèlisation basée sur une décomposition de domaines est correcte, nous avons développé une méthodologie adéquate qui demande à l'utilisateur d'abstraire son code séquentiel afin d'en obtenir une post-condition. Celle-ci nécessite d'être généralisée pour le code parallèle. Ensuite, on doit prouver que les post-conditions du code parallèle, plus la post-condition du code réalisant le collage des informations obtenues en parallèle, impliquent la post-condition du programme séquentiel. Si on ne spécifie pas la post-condition du code réalisant le collage, la preuve échoue, mais les obligations de preuves constituent la post-condition du code d'assemblage des calculs parallèles. Nous avons appliqué cette méthodologie à deux des parallelisations que nous avons effectuées. Pour montrer que l'on peut élaborer un programme à partir d'une spécification formelle et en faire la preuve, nous avons prouvé que le tri bitonique, facilement parallelisable, est correct en utilisant pvs. Nous avons également construit un compilateur qui permet de transformer une spécification unity d'un programme parallèle déterministe en un programme fortran que l'on peut exécuter sur une machine avec openmp
Parallel sparse linear solver with GMRES method using minimization techniques of communications for GPU clusters by Lilia Ziane Khodja( )

1 edition published in 2014 in English and held by 2 WorldCat member libraries worldwide

moreShow More Titles
fewerShow Fewer Titles
Audience Level
Audience Level
  Kids General Special  
Audience level: 0.62 (from 0.54 for Designing ... to 0.97 for Parallel i ...)

Designing scientific applications on GPUs Designing scientific applications on GPUs
Parallel iterative algorithms : from sequential to grid computingDesigning scientific applications on GPUsParallel iterative algorithms
Alternative Names
Raphaël Couturier wetenschapper