Archive ouverte HAL-LAAS
Not a member yet
12189 research outputs found
Sort by
Obsolescence and Availability: The Hidden Challenges of Complex Systems
No full textInternational audienceThe obsolescence challenges associated with rapid technological advances have a significant impact on the overall performance of complex, longlife systems, particularly in sectors such as transportation and heavy industry. This paper addresses these challenges by modeling how obsolescence impacts system availability, reliability, and maintainability considering different categories of item: Make/Buy Repairable and Make/Buy Consumable. Each category encompasses distinct failure, fault and replacement scenarios, which are analysed using flow diagrams. The proposed models provide a systematic understanding of obsolescence mechanisms, including inventory management strategies, repair or replacement paths, and supplier interactions. They also address the progressive decline in availability caused by the obsolescence of components, tools, documentation and technical soft skills. The paper contributes a structured modeling approach based on flow diagrams that links obsolescence scenarios to maintenance decisions, helping anticipate and mitigate performance degradation. By clarifying the decision-making processes inherent in each scenario, the study improves the understanding of the impact of obsolescence on system performance. The article ends with findings and research on predicting system availability under obsolescence
Mixtures Closest to a Given Measure: A Semidefinite Programming Approach
No full textInternational audienceMixture models, such as Gaussian mixture models (GMMs), are widely used in machine learning to represent complex data distributions. A key challenge, especially in high-dimensional settings, is to determine the mixture order and estimate the mixture parameters. We study the problem of approximating a target measure, available only through finitely many of its moments, by a mixture of distributions from a parametric family (e.g., Gaussian, exponential, Poisson), with approximation quality measured by the 2-Wasserstein or the total variation distance. Unlike many existing approaches, the parameter set is not assumed to be finite; it is modeled as a compact basic semi-algebraic set. We introduce a hierarchy of semidefinite relaxations with asymptotic convergence to the desired optimal value. In addition, when a certain rank condition is satisfied, the convergence is even finite and recovery of an optimal mixing measure is obtained. We also present an application to clustering, where our framework serves either as a stand-alone method or as a preprocessing step that yields both the number of clusters and strong initial parameter estimates, thereby accelerating convergence of standard (local) clustering algorithms
A New Path-Oriented Optimization Procedure for Consistent Person Reidentification in a Camera Network
No full textInternational audienceCommon person re-identification (re-ID) approaches focus on finding person signature similarities extracted inpairs of cameras without explicitly maintaining consistency of the re-ID results across the camera network. In contrast,the so-called NCR [2] approach considers all person detections in the networked cameras as a weighted graph, andthen partitions the graph into clusters aiming at maximizing the sum of the cluster profits. NCR clustering uses a pa-rameter K ∈ [0, 1] to offset similarity scores, making them negative. The partitioning is highly sensitive to the tuningof this value, which significantly impacts it. To counter this drawback, a new D(Directed)-NCR approach is proposed inthis paper. It exploits a directed graph structure that explicitly incorporates temporal and topological information, andalso employs a clustering approach to estimate the number of paths associated with different identities in the net-work. Evaluations on the well-known public datasets HDA and RPIfield demonstrate that D-NCR outperforms the NCRapproach
Increasing the throughput of Direct-to-Satellite Narrowband IoT networks
No full textInternational audienceThe design of a Direct-to-Satellite (DtS) network infrastructure based on the availability of Low Earth Orbit (LEO) satellite constellations has recently emerged as a key enabler of global Internet of Things (IoT) connectivity. In this context, Narrowband IoT (NB-IoT) communications can be leveraged as a standards-based DtS solution that can grant reliable data collection in areas not reachable by traditional terrestrial IoT networks. However, the large coverage area of satellites and the limited connection time increase the probability of collisions during the NB-IoT Random Access (RA) procedure. This is especially true when many concurrent User Equipments (UEs) attempt to access the network simultaneously during the RA procedure. In this sense, the goal of this contribution is to show that the first message exchange between UEs and NB-IoT equipped LEO satellites (i.e., an uplink Msg1 frame followed by a downlink Msg2 frame) can be exploited by UEs to decide whether to try transmitting data into an uplink Msg3 frame or not. Herein, 2 early collision detection methods able to improve the RA success rate in satellite NB-IoT systems are presented. In detail, for both methods UEs estimate their relative position to the satellite and apply a corresponding time shift when transmitting Msg1. By comparing the Time Advance (TA) value received in Msg2 with their expected TA, UEs can determine whether to proceed with Msg3. The Closest First Method (CFM) requires no changes to the satellite and only uses the TA value of the first received Msg1. Instead, the Non-collided First Method (NFM) assumes the satellite can identify individual Msg1 transmissions and responds only to non-collided ones. Some preliminary simulation results show that both methods increase the number of successful accesses compared to the standard approach, even in the presence of position estimation errors
Time evolution of controlled many-body quantum systems with matrix product operators
No full textWe present a method for describing the time evolution of many-body controlled quantum systems using matrix product operators (MPOs). Existing techniques for solving the time-dependent Schrödinger equation (TDSE) with an MPO Hamiltonian often rely on time discretization. In contrast, our approach uses the Magnus expansion and Chebyshev polynomials to model the time evolution, and the MPO representation to efficiently encode the system's dynamics. This results in a scalable method that can be used efficiently for many-body controlled quantum systems. We apply this technique to quantum optimal control, specifically for a gate synthesis problem, demonstrating that it can be used for large-scale optimization problems that are otherwise impractical to formulate in a dense matrix representation
Bias Induced Ambipolar Transport in Organic Heterojunction Sensors
No full textInternational audienceAbstract Interface engineering in organic heterostructures is an important approach to tuning the characteristics of organic electronic devices and improving their performances in applications, such as gas sensing. Herein, organic heterostructures containing, a polyporphine (pZnP‐1), perfluorinated copper phthalocyanine (Cu(F 16 Pc)), and lutetium bis‐phthalocyanine (LuPc 2 ) are synthesized by a combination of electrochemical and PVD methods for investigation of charge transport and ammonia (NH 3 ) sensing application. pZnP‐1 is synthesized by controlled oxidative electropolymerization and reveals a rough surface, which influences the electrical nature of its interface with the phthalocyanine. The electrical properties of the heterojunction devices reveal distinct interfacial and bulk charge transport properties, which are modulated by the thickness of pZnP‐1 and the external electric field. Indeed, the heterojunction device containing a thin film of pZnP‐1 displays n‐type behavior at low bias and p‐type nature at higher bias; i.e., an ambipolar behavior, in which ambipolarity is triggered by the external electric field. On the other hand, the heterojunction device having a thick film of pZnP‐1 exhibits p‐type behavior at all the studied biases. Investigation of NH 3 sensing properties of the heterojunction devices highlights the advantages of introducing pZnP‐1 in the heterostructures, which enhances the sensitivity, stability, repeatability, and humidity tolerance of the sensors
Architectures et algorithmes intelligents pour les systèmes IoT autonomes
No full textNational audienceQuality of Service (QoS) is a critical factor that significantly impacts the performance of IoT services and applications. However, with the rapid growth of IoT devices in various domains such as smart homes, smart cities, and Industry 4.0, ensuring QoS has become increasingly challenging. IoT services hosted on infrastructure demand diverse network resources, and the emergence of paradigms like autonomous IoT systems and Zero-Touch Networks has further intensified the need for self-sufficient infrastructures capable of meeting QoS requirements without human intervention. While 5G networks introduced Network Slicing to partition physical infrastructure into logical slices tailored to different IoT service needs, this approach shifted the challenge toward efficiently placing the maximum number of slices while maintaining autonomy. In this thesis, we propose using Deep Reinforcement Learning (DRL) to provide customized QoS for IoT devices in autonomous IoT systems. First, we develop DRL agents that autonomously place slices within IoT infrastructure, continuously learning and improving through interaction and reward-based feedback. This results in IoT systems that not only function independently but also enhance their efficiency over time. Next, we address scenarios where Network Slicing is not feasible and introduce a DRL-based agent that ensures QoS at the application level by assisting IoT devices in selecting the optimal edge or cloud servers for task execution. The agent dynamically considers network conditions such as latency, bandwidth, and computational power to make efficient offloading decisions. Collectively, these contributions offer interesting solutions for achieving fully autonomous IoT systems.La qualité de service (QoS) est un élément essentiel au bon fonctionnement des services et applications IoT. Toutefois, avec l'augmentation massive du nombre d'objets connectés dans des domaines comme les maisons intelligentes, les villes intelligentes ou l'industrie 4.0, garantir cette QoS devient de plus en plus complexe. Ces services, lorsqu'ils sont hébergés sur une infrastructure, requièrent diverses ressources réseau. L'émergence de concepts tels que les systèmes IoT autonomes et le Zero-Touch Network accentue encore cette problématique, car l'objectif est désormais de disposer d'infrastructures capables de satisfaire automatiquement les besoins en QoS des services IoT, sans intervention humaine. Les réseaux 5G ont introduit le Network Slicing pour segmenter l'infrastructure physique en réseaux logiques appelés slices, chacun étant conçu pour répondre aux exigences spécifiques des services IoT qu'ils hébergent. Cependant, ce mécanisme a déplacé la problématique vers l'optimisation du placement de ces slices en maximisant leur nombre tout en garantissant une gestion autonome. Dans cette thèse, nous avons proposé l'utilisation de l'apprentissage profond par renforcement (DRL) pour offrir une QoS sur-mesure aux objets IoT dans un environnement autonome. Nous avons d'abord développé des agents DRL capables de placer de manière autonome des slices sur une infrastructure IoT, en s'améliorant continuellement grâce à l'apprentissage par interaction et un mécanisme de récompense. Ensuite, nous avons étudié le cas où le Network Slicing ne serait pas applicable et avons conçu un agent DRL capable d'assurer la QoS au niveau applicatif. Cet agent assiste les objets IoT dans le choix optimal des serveurs edge ou cloud pour l'exécution de leurs tâches, en tenant compte des conditions dynamiques du réseau telles que la latence, la bande passante et la puissance de calcul des serveurs. Ensemble, ces contributions apportent des solutions intéressantes vers la réalisation de véritables systèmes IoT autonomes
A new Constraint Programming model for the Multiple Constant Multiplication
No full textInternational audienceThe Multiple Constant Multiplication (MCM) problem arises in many applications such as, for example, digital signal processing. Given a set T of target constants, the goal of MCM is to find the most efficient way for multiplying an input number with each constant in T , where multiplications are realized through bit-shifts and additions, and where intermediate results may be shared to produce different target constants. Different metrics may be considered for evaluating the cost of a solution, and a classical objective function is to minimize the number of adders. State-of-the-art methods, based on Integer Linear Programming (ILP), suffer from numerous performance and scalability bottlenecks. In this work, we propose for the first time a Constraint Programming (CP) model for minimizing the number of adders for the MCM. Compared to the state-of-the-art ILP approach, CP does not suffer from the curse of linearization, hence permits significantly simpler formulations of the mathematical model. In order to evaluate our CP model, we focus on a widely used benchmark extracted from a collection of digital filter designs and compare ourselves with state-of-the-art ILP and SAT models. We show that our CP approach is less efficient on some easy instances, but more efficient on hard instances. We also introduce a pseudo-polynomial time algorithm which is able to solve some instances, and show that using this algorithm during a preprocessing step improves the solution process
Extraction de la surface de couplage équivalente d’un condensateur monté sur un circuit imprimé par balayage en champ proche
No full textNational audienceThis paper presents an approach to extract the equivalent coupling surface of a capacitor mounted on a PCB. This is the first step of developing a methodology to extract the insertion loss of an EMI filter in situ. The approach utilizes near-field scan technique based on H-field probe coupled to a capacitor on a PCB. The value of mutual inductance between the probe and capacitor is then extracted by post processing. Finally, the equivalent coupling surface is extracted by optimization using GEMSEO tools.Cet article présente une approche permettant d’extraire la surface de couplage équivalente d’un condensateur monté sur un circuit imprimé (PCB). Il s’agit de la première étape du développement d’une méthodologie destinée à extraire l’atténuation d’insertion d’un filtre EMI in situ. L’approche repose sur une technique de balayage en champ proche utilisant une sonde de champ magnétique (H-field) couplée à un condensateur sur un PCB. La valeur de l’inductance mutuelle entre la sonde et le condensateur est ensuite extraite par post-traitement. Enfin, la surface de couplage équivalente est déterminée par optimisation à l’aide des outils GEMSEO
Dynamic structure of the cytoplasm
No full textThe cytoplasm is a dense and complex milieu in which a plethora of biochemical reactions occur. Its structure is not understood so far, albeit being central to cellular functioning. In this review, we highlight a novel perspective in which the physical properties of the cytoplasm are regulated in space and time and actively contribute to cellular function. Furthermore, we underscore recent findings that the dynamic formation of local assemblies within the cytoplasm, such as condensates and polysomes, serves as a key regulator of mesoscale cytoplasmic dynamics.</div