Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Interfacial fracture in soft solids -how geometry and viscoplasticity make crack fronts unstable
No full textInternational audiencePolyvinylbutyral (PVB) is a polymer with sizeable viscoelastic dissipation at room temperature. It is often used in laminated glass to impart shock resistance to glazings. We have investigated adhesion rupture in glass/PVB interfaces in the through crack tensile test (TCT) geometry, representative of laminated glass rupture. We find that even though, in the high velocity range, interfacial rupture apparently follows the linear viscoelastic predictions, at low velocity a much richer behaviour appears: the system bifurcates, the front undulates, and at still lower velocities, it stops. Such instabilities cannot be explained by linear viscoelastic fracture. Prompted by the measured tensile response of PVB at high strain rates/low temperatures, we have explored steady state viscoplastic fracture, using a generic numerical model. The results show that the TCT geometry enhances the viscoplastic response in the rupture process. They also demonstrate that with viscoplasticity, the rupture energy decreases with velocity, a characteristic which indeed accounts for the observed crack front instabilities. We further discuss the implications of these findings for a better understanding of adhesion and rupture in soft matter and their connection to viscoplasticity
Scattering of transient waves by an interface with time-modulated jump conditions
No full textInternational audienceTime modulation of the physical parameters offers interesting new possibilities for wave control. Examples include amplification of waves, harmonic generation and non-reciprocity, without resorting to non-linear mechanisms. Most of the recent studies focus on the time-modulation of the bulk physical properties. However, as the temporal modulation of these properties is difficult to achieve experimentally, we will concentrate here on the special case of an interface with time-varying jump conditions, which is simpler to implement. This work is focused on wave propagation in a one-dimensional medium containing one modulated interface. Properties of the scattered waves are investigated theoretically: energy balance, generation of harmonics, impedance matching and non-reciprocity. A fourth-order numerical method is also developed to simulate transient scattering. Numerical experiments are conducted to validate the numerical scheme and to illustrate the theoretical findings
A Survey on Versatile Embedded Machine Learning Hardware Acceleration
No full textInternational audienceThis survey investigates recent developments in versatile embedded ML hardware acceleration. Various architectural approaches for efficient implementation of ML algorithms on resource-constrained devices are analyzed, focusing on three key aspects: performance optimization, embedded system considerations (throughput, latency, energy efficiency) and multi-application support. Nevertheless, it does not take into account attacks and defenses of ML architectures themselves. The survey then explores different hardware acceleration strategies, from custom RISC-V instructions to specialized PE, PiM architectures and co-design approaches. Notable innovations include flexible bit-precision support, reconfigurable PE, and optimal memory management techniques for reducing weights and (hyper)-parameters movements overhead. Subsequently, these architectures are evaluated based on the aforementioned key aspects. Our analysis shows that relevant and robust embedded ML acceleration requires careful consideration of the trade-offs between computational capability, power consumption, and architecture flexibility, depending on the application
Experimental and Numerical Study of an Oscillating Foil Near the Free Surface
No full textInternational audiencePrevious studies have explored hydrofoils as a means to enhance ship propulsion in waves, focusing on both passive and active motion control (X. Wu et al., 2020). However, some aspects as free-surface effects on thrust performance remains less understood for 3D configurations, including with regular waves at the surface (J. Deng et al., 2022 ; N. Petikidis et al. 2023). This work presents experimental and numerical results to study the performances of an actively pitching and heaving NACA0015 near the free surface. The experiments are carried out in the Ifremer current and wave tank at Boulogne sur Mer for several heave and pitch amplitudes. The influence of the phase lag between the pitch and heave motions is also investigated, as well as the hydrofoil performances in regular waves. The experiments consists of forces measurements and there are compared with numerical predictions of the hydrodynamic forces using the potential flow code PUFFIn in its latest unsteady version, based on the boundary element method (P. Perali et al., 2024). Overall, the mean numerical forces exhibit similar trends to the experimental averages, validating the use of the potential flow code to optimize foil kinematics or geometry
Former aux enjeux environnementaux au Maghreb : entre les discours des décideurs et la réalité dans les écoles d'ingénieurs
No full textInternational audienceIn engineering education, there is a strong circulation of ideas on pedagogical practices, in an international level. Based on a survey of actors of engineering education in Algeria, Morocco and Tunisia, this article aims to understand how the programs are evolving in response to the environmental challenges faced in Maghreb. It presents the ongoing dynamics in curriculum production by examining how teachers in engineering schools translate various directives and influences into the construction of curricula.Le monde de la formation des ingénieurs est marqué par une forte circulation des idées au niveau international concernant les pratiques pédagogiques. Reposant sur une enquête auprès des acteurs de la formation des ingénieurs en Algérie, au Maroc et en Tunisie, cet article propose de saisir en quoi ces formations se transforment face aux défis environnementaux que connaît le Maghreb. Il présente les dynamiques en cours dans la production curriculaire à travers les manières qu’ont les enseignants en écoles d’ingénieurs de traduire les injonctions et les influences diverses dans la construction des curricula
Benchmark entre plusieurs méthodes hybrides rapides pour modéliser et calculer la SER de cibles métalliques en milieu maritime
No full textInternational audienceCet article propose d'étudier plusieurs méthodes hybrides pour calculer la surface équivalente radar (SER) de cibles métalliques dans un environnement maritime. Elles combinent, d'une part, l'équation d'onde parabolique (EOP) résolue dans un domaine d'ondelettes pour modéliser la propagation des ondes jusqu'à la cible, puis d'autre part des méthodes basées sur des équations intégrales (EI) pour évaluer l'onde diffusé par celle-ci. Dans ce cadre, nous comparons une méthode hybride EOP avec l'optique physique (OP), l'une avec la théorie physique de la diffraction (TPD) et une dernière basée sur l'hybridation avec la méthode des moments (MdM) en ondelettes. Des expériences numériques dans la bande VHF sont réalisées pour valider les performances et comparer les différents modèles hybrides
Low-complexity fully adaptive equalization algorithm for UWA communication systems
No full textInternational audienceThis paper presents an improved turbo equalization algorithm for underwater acoustic communications, designed to tackle severe time variations with reduced computational complexity. Our approach uses an enhanced optimization criterion that leverages soft information exchanged between the channel decoder and the equalizer to refine the equalizer's adaptation process. In contrast to previous works, our proposal has low computational complexity by selectively choosing the most relevant terms to be computed by the equalizer, and also by considering the sparsity of the Underwater Acoustic (UWA) channel. Our simulations, both on synthetic and real underwater channels, demonstrate that our method outperforms conventional equalization techniques. Compared to the fullcomplexity adaptive algorithm, our proposal shows better Bit Error Rate (BER) performance and lowers the computational complexity by about 90% for the 16-QAM constellation.</div
Secure by design Systems on Chip
No full textInternational audienceToday, systems-on-a-chip can be found in all industrial sectors. This is due to the combination of high performance and low energy footprint. These two aspects have been highly valued, particularly in the context of embedded systems, and have concentrated the bulk of development efforts, unfortunately to the detriment of security. Indeed, Security has only been considered relatively recently, through retrofitting. However, there have been some remarkable pioneering efforts, such as ARM's TrustZone technology, to make systems more secure. TrustZone separates processes into two worlds: a “secure” world and a ‘rich’ world with limited privileges, which cannot access the resources of the “secure” world. The main weakness of these approaches is the lack of security continuity when moving from software to hardware and back, making these solutions ill-suited to the integration of potentially unreliable devices into systems-on-a-chip, i.e. most embedded systems. In concrete terms, the pressure of time-to-market, but also of reduced development costs, motivates the re-use of non-certified, potentially even “black-box” hardware Ips; this puts the whole system at risk.The challenge is therefore to propose a low-cost – a key factor for adoption - approach to securing systems by design. Among the solutions available in the literature is the “TrustSoc” proposal. TrustSoc provides an end-to-end security framework that prevents process/node spoofing (for example, rich-world and safe-world processes are identified and separated), while providing a fine-grained rights management policy.The software component of the solution is based on seL4, a formally verified, high-performance operating system microkernel, which enables resources to be allocated while ensuring their proper use. As part of TrustSoc, it guarantees non-usurpation of identity from the requestor application when writing to the bus, hence proper rights analysis.At the hardware level, TrustSoc relies on the insertion of rights controllers at the I/O level of both slave (e.g. memory) and master (e.g. processor, IP block) devices. These controllers are dedicated to rights checking, and implement various policies (filtering, penalty). This solution has been proven portable to both ARM and RISC-V processors, both of which are widely used in embedded applications. The whole system supports a devops approach, for tailoring/debugging/verification purposes, with the Qemu simulator acting as an execution engine. Our reference implementation testing on an AMD-Xilinx Zynq-7000 SoC-FPGA showed the penalty of the proposed isolation strategy to be small, up to 1% in LUT and 0.7% Flip Flop utilizations, making it an efficient security solution
Explicit T-coercivity for the Stokes problem: a coercive finite element discretization: Explicit T -coercivity for Stokes
No full textUsing the T -coercivity theory as advocated in Chesnel-Ciarlet [Numer. Math., 2013], we propose a new variational formulation of the Stokes problem which does not involve nonlocal operators. With this new formulation, unstable finite element pairs are stabilized. In addition, the numerical scheme is easy to implement, and a better approximation of the velocity and the pressure is observed numerically when the viscosity is smal
QM-ARC: QoS-aware Multi-tier Adaptive Cache Replacement Strategy
No full textInternational audienceDistributed data-centric systems, such as Named Data Networking, utilize in-network caching to reduce application latency by buffering relevant data in high-speed memory. However, the significant increase in data traffic makes expanding memory capacity prohibitively expensive. To address this challenge, integrating technologies like non-volatile memory and high-speed solid-state drives with dynamic random-access memory can form a cost-effective multi-tier cache system. Additionally, most existing caching policies focus on categorizing data based on recency and frequency, overlooking the varying Quality-of-Service (QoS) requirements of applications and customers—a concept supported by Service Level Agreements in various service delivery models, particularly in Cloud computing. One of the most prominent algorithms in caching policy literature is the Adaptive Replacement Cache (ARC), that uses recency and frequency lists but does not account for QoS. In this paper, we propose a QoS-aware Multi-tier Adaptive Replacement Cache (QM-ARC) policy. QM-ARC extends ARC by incorporating QoS-based priorities between data applications and customers using a penalty concept borrowed from service-level management practices. QM-ARC is generic, applicable to any number of cache tiers, and can accommodate various penalty functions. Furthermore, we introduce a complementary feature for QM-ARC that employs Q-learning to dynamically adjust the sizes of the two ARC lists. Our solution, evaluated using both synthetic and real-world traces, demonstrates significant improvements in QoS compared to state-of-the-art methods by better considering priority levels. Results show that QM-ARC reduces penalties by up to 45% and increases the hit rate for high priority data by up to 84%, without negatively impacting the overall hit rate, which also increases by up to 61