WorldCat Identities

Laboratoire en sciences et techniques de l'information, de la communication et de la connaissance

Overview
Works: 223 works in 235 publications in 2 languages and 252 library holdings
Roles: Other, 981
Publication Timeline
.
Most widely held works by de la communication et de la connaissance Laboratoire en sciences et techniques de l'information
Personnalisation de l'apprentissage dans un environnement en ligne par recommandation de ressources pédagogiques à l'aide d'une approche de filtrage hybride dans un contexte d'enseignement supérieur au Maroc by Mohammed Baidada( )

1 edition published in 2020 in French and held by 2 WorldCat member libraries worldwide

The evolution of information technologies has impacted the field of education through the introduction of the use of digital technology in educational processes which makes it possible to assist and, in particular, to personalize learning. Much research has been carried out on the personalization and use of recommendation systems by taking up certain approaches applied in online commerce. Our research work falls within this context and aims to test the impact of hybridizing recommendation approaches by combining content-based filtering and collaborative filtering. These two methods are based respectively on the individual and social characteristics of the learner. The globally convincing results of our study and of the two experiments that accompanied it, allowed us to propose several recommendations and an application scenario in the form of an approach based on a blended learning mode combining face-to-face and distance learning. This approach ensures personalization of learning through the implementation of a layered architecture: services, recommendations and data. The various recommendations have been contextualized in the field of higher education in general, and particularly in the Moroccan private and public education system. The experiment on the interest of hybrid recommendation systems in online education took place in the health context of COVID-19
Performance and energy efficiency of new generation video decoding standards on low power consumption-based multi-cores architectures by Mohammed Bey Ahmed Khernache( )

1 edition published in 2021 in English and held by 2 WorldCat member libraries worldwide

The computational demands of video applications are constantly evolving. This evolution has puta great pressure on the battery life-time of mobile platforms. As a result, dedicated hardware components are integrated into system-on-chip (SoC) to optimize the power consumption of hardware video decoding (HW). However, on the one hand, these decoders (HW) are not flexible and have a long time-to-market. On the other hand, heterogeneous multi-core general purpose processors (GPPs) are powerful enough to perform software video decoding (SW) in real time and are flexible. However, they consume a large amount of energy. In this context, we investigate how to optimize SW video decoding on GPPs to get as close as possible to the energy efficiency of HW decoding. The thesis project consists in proposing strategies to optimize the energy consumption of video decoding on heterogeneous mobile platforms. The thesis is part of the FUI-23 EFIGI project. The first phase of the thesis is the characterization of the performance and energy consumption of video decoding implemented in HW and SW. For this purpose, we proposed a methodology based on measurement. It intervenes at two levels: (i) operating system, and (ii) application. The methodology consists in studying the impact of parameters, triggered at these levels, on the performance and energy consumption of HW and SW video decoding. The second phase of the thesis is the optimization of the energy consumption of SW video decoding. The proposed solution is decomposed into three phases: (1) frame complexity modeling, (2) classification by assigning frames to high or low performance cores according to their complexity, and (3) GPP frequency scaling using feedback control with DVFS
Introduction d'aléas dans les architectures matérielles pour une contribution à la sécurisation de chiffreurs AES dans un contexte IoT by Ghita Harcha( )

1 edition published in 2021 in French and held by 2 WorldCat member libraries worldwide

Nowadays, information and data exchange are key elements of our economies. Furthermore, one must add to this the explosion and rapid diffusion of the so-called Internet of Things (IoT, Internet of Things) on every aspect of our lives, professional and personal. These are area and energy-constrained embedded communicating systems, which are already widely deployed. However, IoT nodes present many vulnerabilities and therefore are among the targets of malicious attacks. Therefore, these devices rely more and more on encryption systems. Unfortunately, their implementations themselves can be prone to defaults. In this PhD thesis, we are interested in securing an AES encryption architecture against side-channel attacks, and in particular on power analysis attacks. As the field of IoT is targeted, low-cost AES architectures are targeted and the goal is to minimize the area, throughput, latency, and energy overhead. The proposed approach consists in adding a shuffling module to an AES component. In this context, several architectural solutions have been studied: the number of permutations and the types of permuted information. The robustness of the different architectures against different state-of-theart attacks is evaluated. The additional costs induced by the shuffling module are quantified and the effects of the synthesis options are studied. The results show that no byte of the encryption key was revealed on our most secure (and therefore the most complex and expensive) architectural model is revealed after one million power traces measured on the FPGA. This safety contribution is not for free, which is contained with regard to the solutions proposed on the state of the art: Regarding throughput, the factor overhead is 2.3; and the maximum frequency overhead is about 11%; the hardware cost is approximately equivalent to 3.9 times compared to our reference design
Modélisation du processus de saltation en milieu extrême : application au Raz Blanchard by Gaspard Minster( Book )

2 editions published in 2019 in French and held by 2 WorldCat member libraries worldwide

Understanding the mechanism of saltation in areas of high tidal currents such the Alderney Race, is needed to develop exploitation of hydrokinetics energy. In these areas, sediment transport is characterized by inertial regime. An experimental study in a channel and a numerical simulation are used. The principle consists to let spherical particles roll in different flows on a defined bed roughness. After having characterized the flow, we recorde the particles movement with a Fastcam in order to identify height and length of jumps, impact and rebound velocity and angle. A first law between relative roughness and length/height ratio is established, as well as a second law linking the Shields parameter and the height of the jumps. It is then possible, with the knowledge of theShields parameter, the bed roughness and the diameter of the sediments, to predict the mean jumps height and length. The experimental results are in agreement with the laws described in the literature, and complete them for the inertial regime. These proposed laws are then compared and completed by a numerical model of periodical saltation (Berzi et al. 2016). Finally, they are applied on the study area of the Alderney Race to carry out simple maps. These maps characterize the jumps in different places of the Alderney Race
Étude de la prise de décision chez les pilotes d'aviation commerciale : relation entre le contrôle cognitif et la charge mentale by Amine Laouar-Zouyed( )

1 edition published in 2021 in French and held by 2 WorldCat member libraries worldwide

This thesis aims to study the degradation factors of pilots' decision-making in dynamic situations, with the purpose of defining the principles for a decision-making assistance in future cockpits. The analysis of aviation decision-making processes allowed to focus the problematic on the question of cognitive control. We considered the Contextual Control Model (Hollnagel, 1993), which have four cognitive control modes - strategic, tactical, opportunistic and scrambled, from the more proactive to the less - and we explored the links between these modes, the situation's mental stress (as defined by DIN EN ISO 10075-1:2017) and the operators' mental strain. This led to the hypothesis that we could determine the cognitive control mode instantiated by assessing the mental stress and measuring the strain via physiological (cardiac pace) or neurophysiological measures (prefrontal cortex oxygenation). With the help of a Cognitive Work Analysis carried on piloting activity, we built and conducted two experimentations in order to explore the aforementioned links: one in laboratory with university students confronted to the MATB-II software, the other in a flight simulator with expert pilots who faced realistic scenarios. The experimental results confirmed the links between control modes and mental stress showing a preferential adoption of the tactical mode when the mental stress is low. We also observed a link between control modes and mental strain. Indeed, the prefrontal cortex blood oxygenation rate is lower for the tactical mode than for the other modes. Finally, these two experimentations showed that the tactical mode was the most efficient, associated with lower mental strain and adequate performances. The results of this study are discussed with regards to the perspectives about the development of adaptive assistances
La gestion du risque d'abordage dans le domaine du transport maritime : proposition d'un modèle générique tridimensionnel de la sécurité by Benoît Langard( )

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

This thesis belongs in the general theoretical area of the safety of complex systems. Its field of application is the merchant navy. Its object is twofold: 1. To investigate the management of collision risks by a systemic approach which takes into account both factors leading to failure and those leading to success; and, 2. To explore the links between three concepts of safety : Safety culture, Safety Management Systems and Resilience. Safety culture is a subset of the overall organizational culture. It consists of a set of values, beliefs, attitudes, norms and practices concerning safety. Shared by members of an organization, this directly influences behaviour at work. The first step in the research was to study the marine transportation system and collisions between merchant vessels listed by the United Kingdom's Marine Accident Investigation Branch (MAIB), focussing on two areas: the safety barriers system and the main failures leading to accidents. Qualitative and quantitative methods were then applied to gain an understanding of the safety culture of a shipping company with a good safety record. Analysis of this company's management system showed it to be effective: it was built on a strong reporting system and seemed to be participatory. A survey of the safety climate was conducted with several crews, using a questionnaire inspired by Håvold's questionnaire. This showed that a large majority of the company's personnel had a positive safety culture, although the scores were significantly influenced by the duty and the hierarchical level of individual respondents. Finally, the activities of bridge watchkeeping officers were studied to identify diachronic and synchronic mechanisms implemented to manage the collision risk. The results of these studies enable a discussion of the relationship between Safety Culture, Resilience and Safety Management System. A proposal for a generic three-dimensional safety model shows that Safety Culture seems to act as the mediator of the other two concepts. Safety Culture would also allow the system to maintain a significant portion of “managed safety”, complementing a strong element of “prescribed safety” flowing from regulation
Les rôles : médiateurs dynamiques entre modèles système et modèles de simulation by Jean-Philippe Schneider( Book )

2 editions published in 2015 in French and held by 2 WorldCat member libraries worldwide

Les systèmes actuels tendent à être intégrés les uns avec les autres. Mais cette intégration n'est pas forcément prévue à l'origine du système. Cette tendance créée des systèmes de systèmes. Un système de système de systèmes est un système constitué de systèmes qui sont gérés par des équipes indépendantes, qui sont fonctionnellement indépendants, qui collaborent, qui évoluent et qui sont géographiquement distribués. La communication entre les différentes équipes facilite la conception d'un système de systèmes. Cette communication peut être réalisée par l'utilisation de modèles et de simulation. Cependant, la modélisation du système de systèmes et la modélisation des simulations ne reposent pas sur les mêmes langages. Pour assurer la cohérence des modèles, il faut pouvoir créer les modèles de simulation à partir des modèles système. Cependant, il faut tenir compte des contraintes liées aux propriétés des systèmes de systèmes. Il faut être capable de manipuler des modèles systèmes réalisés dans des langages différents, de réaliser des simulations de natures différentes et suivre les évolutions des langages de modélisation et des outils de simulation. Pour répondre à ces problématiques, nous avons défini l'environnement Role4AII pour la manipulation de modèles systèmes réalisés dans des langages hétérogènes. Role4AII est basé sur la notion de rôles. Les rôles permettent de créer des simulations en accédant aux informations contenues dans des éléments de modèles indépendamment de leur type. Role4AII est capable de prendre en entrée des modèles sérialisés par différents outils grâce à l'utilisation de parsers combinateurs. Ces derniers apportent modularité et extensibilité aux fonctionnalités d'import. L'environnement Role4AII a été utilisé sur un exemple de système de systèmes : l'observatoire sous-marin MeDON
Résolution de problèmes réalistes de tournées à flotte hétérogène en milieu urbain : vers un solveur adaptatif mêlant recherche opérationnelle et apprentissage automatique by Flavien Lucas( )

1 edition published in 2020 in French and held by 2 WorldCat member libraries worldwide

In the field of operations research, solving vehicle routing problem has been an emblematic challenge since the 1960s. Initially developed on simple problems, many variants have improved the realism of the studied instances, especially in urban areas where congestion depends on the vehicle used. However, there are still few methods consistent with all the existing variants, which are becoming less effective as the number of clients increases. It is in this context that we have developed a robust and adaptive solver, obtaining good solutions in a very short period of time on a high number of instances as difficult as varied (several thousand customers, heterogeneous vehicles, urban environment, etc.). This new solver makes it possible to generate solutions from multiple instances, characterized by a set of variables, which allows a better understanding of what constitutes a good solution. In a context where interdisciplinary work combining operational research and machine learning is becoming more and more frequent, we were interested in the use of decision trees to map areas containing a priori all good quality solutions. The combination between decision trees and this new solver restricts the search for good solutions to promising areas, in order to increase the probability of obtaining good results and to reduce the duration of this exploration. Experimental results indicate that this interdisciplinary work is promising to generate a set of efficient hybrid methods
Distributions alpha-stable pour la caractérisation de phénomènes aléatoires observés par des capteurs placés dans un environnement maritime by Anthony Fiche( Book )

2 editions published in 2012 in French and held by 2 WorldCat member libraries worldwide

The purpose of this thesis is to characterize random signals, observed in air and underwater context, by using a statistical approach. In signal processing, the hypothesis of Gaussian model is often used for a statistical study. However, the Gaussian model is not valid when the probability density function of data are characterized by heavy-tailed and skewness phenomena. A family of laws can fit these phenomena: the [alpha]-stable distributions. Firstly, the [alpha]-stable distribution have been used to estimate generated and real data, extracted from a mono-beam echo-sounder, for seabed sediments classification. The classification is made by using the theory of belief functions, which can take into account the imprecision and uncertainty of data and theirs estimations. The results have been compared to a Bayesian approach. Secondly, in a context a marine surveillance, a statistical study from the [alpha]-stable distribution has been made to characterize undesirable echo reflected by a sea surface, called sea clutter, where the sea surface is considered in a bistatic configuration. The sea surface has been firstly generated by the Elfouhaily sea spectrum and the Radar Cross Section (RCS) of the sea surface has been computed by the Physical Optics (PO). The Weibull and distributions have been used and the results compared to the [alpha]-stable model. The validity of each model has been evaluated by a Kolmogorov-Smirnov test
RL-based Energy Management for Autonomous Cyber Physical Systems by Yohann Rioual( )

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

The energy management of a cyber physical system is a difficult task because of the complexity of hardware architectures and the use of OS. In addition, these systems are deployed in changing environments where their energy harvesting capacity varies. Over time, their energy consumption is modified due to the ageing of the components. Consumption models designed in the laboratory cannot take into account all possible deployment scenarios and system aging. One approach that is developing is the use of reinforcement learning in which we no longer know the system's consumption model; but thanks to this approach, the system is still able to adapt its operation during its deployment according to the evolution of its environment. Several approaches exist in reinforcement learning. The first part of this thesis is devoted to proposing guidelines to help for selecting the most appropriate approach for a given application and target. The second part of this thesis focuses on the design of the reward we give to our system that will influence its behaviour in its environment. Two reward functions able to adjust the system's performance according to the energy available are presented. The third and last part of this thesis explores the use of several agents to independently control the different modules of our system. This approach allows a more precise control of energy consumption, reducing memory usage compared to a single agent approach
Subutai : Distributed synchronization primitives for legacy and novel parallel applications by Rodrigo Cadore Cataldo( )

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

Parallel applications are essential for efficiently using the computational power of a MultiProcessor System-on-Chip (MPSoC). Unfortunately, these applications do not scale effortlessly with the number of cores because of synchronization operations that take away valuable computational time and restrict the parallelization gains. The existing solutions either restrict the application to a subset of synchronization primitives, require refactoring the source code of it, or both. We introduce Subutai, a hardware/software architecture designed to distribute the synchronization mechanisms over the Network-on-Chip. Subutai is comprised of novel hardware specialized in accelerating synchronization operations, a small private memory for recording events, an operating system driver, and a user space custom library that supports legacy and novel parallel applications. We target the POSIX Threads (PThreads) library as it is widely used as a synchronization library, and internally by other libraries such as OpenMP and Threading Building Blocks. We also provide extensions to Subutai intended to further accelerate parallel applications in two scenarios: (i) multiple applications running in a highly-contended scheduling scenario; (ii) remove the access serialization to condition variables in PThreads. Experimental results with four applications from the PARSEC benchmark running on a 64-core MPSoC show an average application speedup of 1.57× compared with the legacy software solutions. The same applications are further sped up to 5% using our proposed Critical Section-aware scheduling policy compared to a baseline Round-Robin scheduler without any changes in the application source code
Accélérateurs matériels sécurisés pour la cryptographie post-quantique by Timo Zijlstra( )

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

Shor's quantum algorithm can be used to efficiently solve the integer factorisation problem and the discrete logarithm in certain groups. The security of the most commonly used public key cryptographic protocols relies on the conjectured hardness of exactly these mathematical problems. A sufficiently large quantum computer could therefore pose a threat to the confidentiality and authenticity of secure digital communication. Post quantum cryptography relies on mathematical problems that are computationally hard for quantum computers, such as Learning with Errors (LWE) and its variants RLWE and MLWE. In this thesis, we present and compare FPGA implementations of LWE, RLWE and MLWE based public key encryption algorithms. We discuss various trade-offs between security, computation time and hardware cost. The implementations are parallelized in order to obtain maximal speed-up. We show that MLWE has the best performance in terms of computation time and area utilization, and can be parallelized more efficiently than RLWE. We also discuss hardware security and propose countermeasures against side channel attacks for RLWE. We consider countermeasures from the state of the art, such as masking and blinding, and propose improvements to these algorithms. Moreover, we propose new countermeasures based on redundant number representation and the random shuffling of operations. All countermeasures are implemented on FPGA to compare their cost and computation time overhead. Our proposed protection based on redundant number representation is particularly flexible, in the sens that it can be implemented for various degrees of protection at various costs
Hybridation multi-sources pour améliorer les fonctions de détection, pistage, localisation et positionnement dans des environnements difficiles by Ngoc-Tan Truong( Book )

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

Services related to GNSS positioning systems have been significantly developed over the last ten years and continue to develop in popularity and efficiency. Other services have been deployed in various fields and sectors of activity. For maritime applications, it is estimated that about 87% of the merchant navy already uses the global satellite system for navigation and positioning. Today, with the advent of the new Galileo and Beidou systems and the modernization of the GPS and GLONASS systems, new satellites and many new frequencies and signals are opening the door to countless new applications. Currently, some applications require better accuracy and exploitation of navigation systems, especially in harsh environments. Indeed, many factors affect the final performance of a GNSS receiver. Some of them are related to the inherent quality of the receiver and its antenna (sensitivity, quality of signal processing, resistance to interference and jamming... etc.), while others are related to the receiver's receiving environment. Sources of interference in the receiving environment (multipath, diffraction... etc.), including the dynamics of moving objects, have important effects on the performance of the receiver. It is in this context that all the work carried out is integrated. Thus, the work in this thesis aims at developing new algorithms with multi-source hybridization to improve the functions of detection, tracking, localization and positioning in difficult environments. The use of multiple signals has advantages as well as disadvantages, as they can contain poor quality signals that have a negative impact on position accuracy. Therefore, we have studied and presented different methods to reduce and/or eliminate outliers: iterative ReWeighted Least Squares (IRWLS) and Robust Extended Kalman Filter (Robust-EKF: REKF). [...] It should be noted that GNSS is not the only technology used to determine the position and velocity of the receiver. In the case where GNSS signals are blocked and/or unusable, we have combined and merged GNSS data with INS data (Inertial Navigation System), which represents a good alternative to determine the position of the receiver. The work carried out within the framework of this thesis is conducted in the context of the e-PANEMA project, supported under the Future Investment Program (FIP) operated by ADEME. The work was carried out in partnership with the other members of the consortium (SAFRAN SE, Diades Marine and ENSM)
Elaboration et évaluation de recommandations ergonomiques pour le guidage de l'apprenant en EVAH : application à l'apprentissage de procédure dans le domaine biomédical by Charlotte Hoareau( Book )

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

Virtual environments are tools that are becoming more frequently used in the field of training, however their design is not systematically guided by ergonomic considerations. This thesis presents the development and evaluation of ergonomic recommendations to guide the learner during VET (Virtual Environment for Training) training. After evaluating the usefulness of STA-R Virtuel, a VET for laboratory technicians, two experiments were conducted to determine optimal ways of cueing the learner during their training in the execution of coagulation tests. The first experiment analyzed the impact on learning of a hierarchical organization of auditory instructions. The second evaluated different ways of presenting a visual guide and their impact on learned proficiency in the procedure. The results obtained at the end of these two experiments were then used to develop ergonomic recommendations for the design of auditory and visual cues to enhance learner performance. Finally, these recommendations were implemented in the VET. The improved version was compared with the reference virtual environment in order to assess the effectiveness of the recommendations. The results show that the inclusion of the proposed ergonomic recommendations improves the learners' performance
Un cadre de définition de la sémantique basée MoC des modèles de systèmes dans le contexte de l'intégration d'outils by Papa Issa Diallo( Book )

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

L'utilisation des systèmes embarqués (EmS) connait un essor conséquent dans plusieurs domaines actuels tels que la téléphonie, l'industrie automobile et l'avionique. Dans ces différents domaines, la croissance des besoins en termes de fonctionnalités a pour conséquence l'augmentation de la taille et de la complexité des systèmes conçus. Dans ce contexte, les chaînes de conception des systèmes deviennent de plus en plus complexes et requièrent l'utilisation d'outils provenant de différents domaines d'ingénieries. L'intégration des paradigmes hétérogènes associés aux outils posent beaucoup de problèmes de fiabilité à l'échange des modèles entre outils d'une même chaîne de conception. Par exemple, dans le cadre des EmS, les outils d'ingénierie dirigés par les modèles (IDM) ne sont pas acceptés par les communautés de recherches pour la conception formelle d'EmS qui requièrent des bases solides et formelles de définition des sémantiques d'exécution pour réaliser les activités d'analyses, de validation et de synthèse des systèmes embarqués. En effet, les outils IDM dédiés aux EmS ne sont à ce jour pas encore suffisamment matures concernant l'expression et la prise en compte de la sémantique d'exécution formelle mettant explicitement en avant les modèles de concurrence des systèmes. Par ailleurs, la théorie du calcul est identifiée comme le domaine permettant de décrire de manière formelle les modèles de concurrences qui sont utilisés pour la description de systèmes embarqués. La motivation de cette thèse est de mettre en œuvre cette théorie du calcul pour réduire l'écart existant entre différents outils de conception qui possèdent des sémantiques d'exécution de modèles différentes dans une chaîne de conception. La thèse propose une méthodologie d'identification et de comparaison des sémantiques d'exécution de modèles qui se base sur la théorie des Modèles de Calcul (MoCs) et leur classification existante, ainsi qu'un langage de capture des sémantiques basées MoC. Ces dernières sont utilisées pour enrichir les modèles et préserver leur sémantique entre les outils d'une chaîne de conception. Pour illustrer l'utilisation de l'approche, nous avons défini un flot de conception permettant de connecter trois outils impliqués dans diverses activités du processus “Design & Implementation” (Spécification, Analyse, Exploration de l'espace des choix de Conception). La chaîne d'outils présentée adresse la connexion de l'outil UML Modeler (IBM Rhapsody) (pour la spécification et l'analyse), Forsyde (cadre de simulation multi-MoC et de synthèse) et Spear (pour l'exploration de l'espace des choix de Conception et l'analyse). La chaîne est appliquée sur un modèle de Radar simplifié fourni comme cas d'utilisation dans le cadre du projet iFEST
Découverte et exploitation de la hiérarchie des tâches pour apprendre des séquences de politiques motrices par un robot stratégique et interactif by Nicolas Duminy( )

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

Il y a actuellement des efforts pour faire opérer des robots dans des environnements complexes, non bornés, évoluant en permanence, au milieu ou même en coopération avec des humains. Leurs tâches peuvent être de types variés, hiérarchiques, et peuvent subir des changements radicaux ou même être créées après le déploiement du robot. Ainsi, ces robots doivent être capable d'apprendre en continu de nouvelles compétences, dans un espace non-borné, stochastique et à haute dimensionnalité. Ce type d'environnement ne peut pas être exploré en totalité, le robot va devoir organiser son exploration et décider ce qui est le plus important à apprendre ainsi que la méthode d'apprentissage. Ceci devient encore plus difficile lorsque le robot est face à des tâches à complexités variables, demandant soit une action simple ou une séquence d'actions pour être réalisées. Nous avons développé une infrastructure algorithmique d'apprentissage stratégique intrinsèquement motivé, appelée Socially Guided Intrinsic Motivation for Sequences of Actions through Hierarchical Tasks (SGIM-SAHT), apprenant la relation entre ses actions et leurs conséquences sur l'environnement. Elle organise son apprentissage, en décidant activement sur quelle tâche se concentrer, et quelle stratégie employer entre autonomes et interactives. Afin d'apprendre des tâches hiérarchiques, une architecture algorithmique appelée procédures fut développée pour découvrir et exploiter la hiérarchie des tâches, afin de combiner des compétences en fonction des tâches. L'utilisation de séquences d'actions a permis à cette architecture d'apprentissage d'adapter la complexité de ses actions à celle de la tâche étudiée
Pilotage automatique des bateaux volants : algorithmes dynamiques et multicritères by Mathilde Tréhin( )

1 edition published in 2020 in French and held by 2 WorldCat member libraries worldwide

Over the years, the sailboats designed for ocean races have strongly evolved. Thanks to the arrival of the hydrofoils, they became “flying machines”. However, keep these machines flying is a constant challenge. In fact, as observed in the previous America's Cup, a team member is totally dedicated to the hydrofoil control. But skippers, in solo offshore races, can't spend all their time adjusting the hydrofoil settings since they must focus on sail trim, refine tactical elements and so on. A solution to maintain the flight is to automate the hydrofoil control. In this study, we detail the design of a flight controller for sailboat. It consists of a stabilization loop and a control loop to manage the flight level. The main actuators for this control system are the flaps of the rudders. In fact, they significantly modify the boat's attitude with limited efforts (easily deployed and controlled, low energy consumer). To enable a quasi-permanent use of the control system, the sailboat needs a reliable energy supply during the whole race. However, energy resources are limited on board. They have few renewable energy producers (solar panel, wind turbine, etc.), but they can meet the boat needs only under specific weather conditions. Sailboats have a limited quantity of fuel which can recharge a battery set. But, they try to gain in performance by minimizing the weight which means reducing the fuel volume. Soon, the flight's autopilot will need to manage more actuators to control these appendages. It is going to increase energy requirements. So, taking into account this criterion in the control laws and optimize its use is becoming mandatory. Thus, we propose a new control method aware of the power consumption to manage the actuators. Finally, to test these algorithms, some tools to analyze control performance are proposed and the instrumentation of a test platform is detailed. This platform enables to confront the control laws with real navigation conditions
Contribution des méthodes d'apprentissage à la distribution de tâches dans un cluster robotique by Paul Gautier( )

1 edition published in 2021 in French and held by 2 WorldCat member libraries worldwide

The multi-robot systems coordination is a complex task requiring the execution of computationaly intensive algorithms. Therefore, the robot's computing resources become a major concern since they condition the system's ability to interact with the environment. While the use of the robotic cloud removes this limitation, its use remains conditional and intractable in many cases. Another solution consists of organizing the multi-robot system into a robotic cluster where the unused resources are pooled in order to parallelize the highly computational tasks. However, these pooled resources turn out to be limited and fluctuating, requiring the implementation of a dynamic allocation process. These thesis works are devoted to the contributions of reinforcement learning in the administration of a robotic cluster operating in a dynamic and uncertain environment. In a new context called MRpTA, two axis are studied: The first deals with the transfer of tasks within a cluster where classical approaches are compared to a DQN solution. The second axis addresses, in a search and rescue context, the question of the distribution of tasks within a cluster. The use of reinforcement learning drastically improves the system performance
Cooperative MIMO techniques for outdoor optical wireless communication systems by Mohamed Abaza( Book )

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

Free-space optical (FSO) communication has been the subject of ongoing research activities and commercial attention in the past few years. Such attention is driven by the promise of high data rate, license-free operation, and cheap and ecological friendly means of communications alternative to congested radio frequency communications. In most previous work considering multiple transmitters, uncorrelated channel conditions have been considered. An uncorrelated channel requires sufficient spacing between transmitters. However, this can be difficult and may not be always feasible in some applications. Thereby, this thesis studies repetition codes (RCs) and orthogonal space-time block codes performance in correlated log-normal FSO channels using intensity modulation and direct detection (IM/DD). Furthermore, the effect of different weather conditions on the average bit error rate (ABER) performance of the FSO links is studied. Multiple-input multiple-output (MIMO) FSO communication systems using space shift keying (SSK) modulation have been also analyzed. Obtained results show that SSK is a potential technique for spectral efficiencies equal or greater than 4 bits/s/Hz as compared to RCs with multiple pulse amplitude modulations. The performance analysis of a multi-hop decode and forward relays for FSO communication system using IM/DD is also considered in this thesis. It is shown that multi-hop is an efficient technique to mitigate atmospheric turbulence and different weather attenuation effects and geometric losses in FSO communication systems. Our simulation results show that multiple-input single-output (MISO) multi-hop FSO systems are superior to direct link and MISO systems over links exhibiting high attenuation. Meeting the growing demand for higher data rates communication networks, a system with full-duplex (FD) relays is considered. For such a system, the outage probability and the ABER performance are analyzed under different turbulence conditions, misalignment error and path loss effects. FD relays are compared with the direct link and half-duplex relays. Obtained results show that FD relays have the lowest ABER and the outage probability as compared to the two other systems. Finally, the obtained results in this thesis are very promising towards the next generation of FSO systems
Cybersecurity of industrial connected equipment : modeling, detection and temporal performance in the presence of intrusions of cyber-physical systems of the factory 4.0 by Salwa Alem( )

1 edition published in 2021 in English and held by 2 WorldCat member libraries worldwide

Over the past decade, the industry has been the victim of several attacks. This phenomenon of cybercrime is eased with the emergence of I4.0. This fourth industrial revolution is characterized by the convergence of Information Technology and Operation Technology worlds', the huge generated data, the use of Cloud as new storage means and the limits of the security mechanisms. All these factors increase the risk of cyber-attacks in industry. The latter can be protected by several security mechanisms including IDS. IDS give visibility of a system activities which in turn allow a timely detection and response to suspicious events. In literature, exist two IDS approaches types: signatures-based and anomaly-based IDS. This last one is more efficient to detect advanced attacks and it is composed of two other IDS types: specification-based IDS and behavioral-based IDS. In industrial intrusion detection field, the main problematic is how to distinguish an industrial process dysfunction from a real intrusion. In this thesis, the proposed approach deals with this specific point. Consequently, hybridization of two types of anomaly-based IDS remains the most efficient method to distinguish a process dysfunction from a real cyber-attack in order to perform a high intrusion detection precision. Therefore, this thesis proposes a hybrid detection approach composed of a behavioral-based IDS is based on the network traffic analysis' using supervised neural network algorithm which learns the normal behavior of the production line and considers any activity that deviates from this reference behavior as intrusion. A specification-based IDS is based on ISA95 standard allowing the detection of the process anomalies and helps to measure the previously detected attacks impact. And a decision-making system (DMS) which pairs the two previous IDS detection results to determine the anomaly nature. The results of this approach are promising with more accurate detection with fewer false positives
 
moreShow More Titles
fewerShow Fewer Titles
Audience Level
0
Audience Level
1
  General Special  
Audience level: 0.00 (from 0.00 for Subutai : ... to 0.00 for Subutai : ...)

Alternative Names
Centre national de la recherche scientifique (France). Laboratoire en sciences et techniques de l'information, de la communication et de la connaissance

Centre national de la recherche scientifique (France). Unité mixte de recherche (6285)

École nationale d’ingénieurs de Brest. Laboratoire en sciences et techniques de l'information, de la communication et de la connaissance

École nationale supérieure de techniques avancées Bretagne. Laboratoire en sciences et techniques de l'information, de la communication et de la connaissance

Ecole nationale supérieure des télécommunications de Bretagne (Brest). Laboratoire en sciences et techniques de l'information, de la communication et de la connaissance

Lab-sticc

Labsticc

Université de Bretagne occidentale. Laboratoire en sciences et techniques de l'information, de la communication et de la connaissance

Université de Bretagne-Sud. Laboratoire en sciences et techniques de l'information, de la communication et de la connaissance

Languages
French (17)

English (10)