Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Memory hierarchy in scheduling simulation: problems, implementation & return of experience
No full textInternational audienc
GpLMS: Generalized Parallel Least Mean Square Algorithm for Partial Observations
No full textInternational audienceWe propose a generalized parallel least mean square algorithm (GpLMS) to deal with partial observation scenarios. GpLMS takes advantage of a two stage parallel LMS architecture to enhance the convergence rate and updates weight vector based on observed entries to obtain a low computational complexity. We compare the results from our proposed algorithm with the state-of-the-arts in an adaptive beamforming context to illustrate its effectiveness
Measurement of the interfacial strain energy release rate of adhesively bonded structures with metallic substrates before and after water ageing
No full textInternational audienceThe aim of the presented work is to propagate cracks at interfaces between adhesive joints and metallic substrates and to measure the Strain Energy Release Rate (SERR). In order to propagate at a specific interface and not cohesively, a specific specimen was developed, based on a modification of the Tapered Double Cantilever Beam (TDCB) specimen. In particular, the dimensions of one substrate were reduced to create an asymmetry in the specimen and drive the propagation path towards one interface. The authors present experimental results showing repetitive interfacial failures and a procedure to evaluate the SERR. The non-linear behaviour responsible of non-linear deformation in the substrate was taken into account in the calculations of the SERR. Through a numerical inverse method using Finite Element Analysis (FEA) and Cohezive Zone Modelling (CZM), the SERR was calculated for an epoxy based structural adhesive at the interface with a stainless-steel substrate. Then, the procedure was carried out to measure the interfacial SERR after different ageing times
Energy-Aware HEVC Software Decoding On Mobile Heterogeneous Multi-Cores Architectures
No full textInternational audienceVideo content is becoming increasingly omnipresent on mobile platforms thanks to advances in mobile heterogeneous architectures. These platforms typically include limited rechargeable batteries which do not improve as fast as video content. Most state-of-the-art studies proposed solutions based on parallelism to exploit the GPP heterogeneity and DVFS to scale up/down the GPP frequency based on the video workload. However, some studies assume to have information about the workload before to start decoding. Others do not exploit the asymmetry character of recent mobile architectures. To address these two challenges, we propose a solution based on classification and frequency scaling. First, a model to classify frames based on their type and size is built during design-time. Second, this model is applied for each frame to decide which GPP cores will decode it. Third, the frequency of the chosen GPP cores is dynamically adjusted based on the output buffer size. Experiments on real-world mobile platforms show that the proposed solution can save more than 20% of energy (mJ/Frame) compared to the Ondemand Linux governor with less than 5% of miss-rate. Moreover, it needs less than one second of decoding to enter the stable state and the overhead represents less than 1% of the frame decoding time
Etalement de spectre par séquence directe et constellations tournées pour communications tactiques
No full textInternational audienceLes systèmes de communications sans fil sont vulnérables aux attaques par interférences car le média étant partagé, toute entité mal intentionnée peut facilement accéder au canal et envoyer des signaux de brouillage qui interfèrent avec le signal légitime et empêchent sa bonne réception. Les techniques à étalement de spectre par séquence directe (Direct-Sequence Spread Spectrum-DSSS) sont alors utiles pour limiter ce risque dans des situations critiques. Dans cet article, on propose d'utiliser conjointement le signal DSSS avec les constellations tournées. Les simulations, prenant en compte des phénomènes physiques comme un fort décalage Doppler, montrent que la méthode proposée permet un gain en robustesse de plusieurs dBs par rapport au DSSS utilisé avec des constellations conventionnelles
Multi-Level Modeling with Openflexo/FML: A Contribution to the Multi-Level Process Challenge
No full textInternational audienceModel federation is a multi-model management approach based on the use of virtual models and loosely coupled links. The models in a federation remain autonomous and represented in their original technological spaces whereas virtual models and links (which are not level bounded) serve as control components used to present different views to the users and maintain synchronization. In this paper we tackle the EMISAJ multi-level process modeling challenge, which consists in providing a solution to the problem of specifying and enacting processes. Solutions must fulfill a number of requirements for a process representation defined at an abstract process-definition level and at various more concrete domain-specific levels, resulting in a multi-level hierarchy of related models. We present a solution based on model federation and discuss the advantages and limitations of using this approach for multi-level modeling. Concretely, we use virtual models and more precisely the Federation Modeling Language (FML) that serves to describe them as the main building block in order to solve the process modeling challenge whereas the federation feature is used as a means to provide editing tools for the resulting process language. Our solution fulfills all the challenge requirements and is fully implemented with the Openflexo framework
Imagerie X de phase basée sur le capteur de front d’onde de Hartmann pour application sur l'étude des maladies neurodégénératives
No full textThe aim of this PhD thesis is to develop the technic of X-ray phase imaging with Hartmann wavefront sensor for various applications and to compare this new system against well-established phase-contrast techniques. The X-ray phase imaging will be mainly performed in 3D using tomographic setup. The main application includes the study of alteractions in the central nervous system induced by neurodegenerative diseases. The first introductory section describes the basic aspects of X-ray interaction with matter and of yhe coherence theory with specific application to the Hartmann wavefront sensor design. In the second chapter, an introduction to the free-space propagation technique. The third chapter examines the principles of tomography acquisitions and the available reconstruction algorithms. A separate chapter, labeled 4, is dedicated to the theory of Hartmann wavefront sensor. A 3D wave propagation model based on Fresnel propagator was developed to optimize the architecture of the full wavefront sensor including the Hartmann plate, the distances between the different elements of the set-up as well as the X-ray source. The model can manage any degree of spatial coherence, enabling the accurate simulation of a wide range of X-ray sources. Several simulations of standard experimental situations are described to valid the program. Then, the main wavefront reconstruction algorithms have been analyzed. In chapter 5, we will present experimental results obtained with the X-ray Hartmann wavefront sensor using both a parallel beam geometry (synchrotron measurements) and a cone beam geometry (laboratory measurements). Different Hartmann plates were used with the laboratory set-up to visualize a series of test and biological samples. also, using synchrotron, we tested the Hartmann sensor to retrieve the chemical composition of objects composed of known materials. The chemical composition could be inferred starting from direct and independent measurements of the real part (proportional to the phase) and the imaginary part (proportional to the absorption) of the sample refractive index. In chapter 6, the experimental results obtained with free space propagation X-ray phase contrast tomography will be discussed. We exploited the capability of X-ray phase contrast imaging to investigate the effects of neurodegenerative diseases on the central nervous system.L'objectif de cette thèse consiste en le développement d'une technique d'imagerie X de phase basée sur l'utilisation d'un senseur de front d'onde de type Hartmann pour être exploité sur différentes applications et de comparer ce nouveau système avec des techniques bien connues d'imagerie en contraste de phase. L'imagerie de phase sera principalement réalisée en 3D par tomographie. L'application principale inclue l'étude d'altérations du système nerveux central induites par des maladies neurodégénératives. La première section d'introduction décrit les aspects de base de l'interaction rayons X-matière et de la théorie de la cohérence avec des applications spécifiques à la conception de senseur de front d'onde de Hartmann. Dans le deuxième chapitre, une introduction à l'imagerie par contraste de phase est donnée, avec une attention particulière sur la technique en propagation libre. Le troisième chapitre examine les principes de la tomographie et des logiciels de reconstruction disponibles. Un chapitre séparé, numéroté 4, est dédié à la théorie des senseurs de front d'onde de Hartmann. Un modèle de propagation en 3D basé sur le propagateur de Fresnel a été développé pour optimiser l'architecture du senseur complet incluant la plaque de Hartmann, les distances entre les différents éléments du montage et enfin les propriétés de la source X. Le modèle peut gérer n'importe quel degré de cohérence spatiale, permettant de réaliser des modélisations précises d'une grande variété de source X. Différentes simulations de situation expérimentales sont décrites pour valider le programme. Puis, les programmes principaux de reconstruction du front d'onde ont été analysés. Dans le chapitre 5, nous allons présenter des résultats expérimentaux obtenus avec le senseur de font d'onde X en géométrie de faisceau parallèle (synchrotron) ou conique (mesure en laboratoire). Différentes plaques de Hartmann ont été utilisé sur le montage de laboratoire pour visualiser une série d'échantillons test et biologiques. De plus, sur synchrotron, nous avons testé le senseur de Hartmann pour retrouver la composition chimique d'objets composés de matériaux connus. La composition chimique peut être estimée à partir de mesures directes et indépendantes de la partie réelle (proportionnelle à l'absorption) de l'indice de réfraction de l'échantillon. Dans le chapitre 6, les résultats expérimentaux obtenus en tomographie X par contraste de phase en propagation libre seront discutés. Nous avons exploité la capacité de l'imagerie X par contraste de phase pour étudier les effets des maladies neurodégénératives du système nerveux central
SLRL: a simple least remaining lifetime file evicition policy for HPC multi-tier storage systems
No full textInternational audienceHPC systems are composed of multiple tiers of storage, from the top high performance tier (high speed SSDs) to the bottom capacitive one (tapes). File placement in such architecture is managed through prefetchers (bottom-up) and eviction policies (top-down). Most state-of-the-art work focus on the former while using algorithm flavors of LRU, LFU and FIFO for the latter. LRU was for long considered the best choice. However, recent studies has shown that the simplicity of FIFO could make it more scalable than LRU because of metadata management, and thus more adequate in several cases. In this paper, we propose a new eviction policy based on predicted files lifetimes. It is comparable to FIFO in terms of metadata overhead and simplicity (thus scalability), while giving a hit ratio comparable to LRU (or even 10% better for some tested traces). We also propose a naive multi-tier heterogeneous storage simulator implementation to evaluate such policies
Infrared Image Processing to Guide the Identification of Damage and Dissipative Mechanisms in 3D Layer-to-Layer Woven Composites
No full textInternational audienceThis article discusses techniques that aim at facilitating the identification of dissipative mechanisms activated in woven composites under cyclic loadings. The focus is put on the post-processing of thermal measurements acquired during heat build-up experiments, as these are usually used to identify the dissipation sources. The importance of motion compensation pre-processing is demonstrated as it is shown that the latter enhances the quality of the evaluated thermoelastic and dissipation fields. Two specific post-processing techniques are presented in this article. The first one analyzes temperature or thermoelastic fields and searches to detect thermal events associated with the creation of cracks. The second one is based on a Fourier decomposition of thermal fields and aims at highlighting an increased contribution of friction as a dissipation source