Résumé: To further enhance decision making in autonomous vehicles field, grant more safety, comfort, reduce traffic, and accidents, learning approaches were adopted, mainly reinforcement learning. However, possibility in upgrading these algorithms is still available due to many limitations including : convergence rate, stability, handling multiple dynamic environments, raw performance, robustness, and complexity of algorithms. To tackle these problems, we propose a novel extension of the well-known deep Q network called “alternative bidirectional Q network” that aims mainly to enhance stability and performance with improving exploration and Q values update policies, to overcome the literature gap that generally focuses on only one policy to handle decision making in multiple scenarios (avoiding obstacles, goal-oriented environments, etc.). In “alternative bidirectional Q network,” data about previous, current, and upcoming states are used to update the Q values where the actions are selected according to the relation between these data to handle several scenarios: highways, merges, roundabouts, and parking. This concept provides reinforcement learning agents with more balance between exploration and exploitation and enhances stability during learning. A gym simulator was adopted for training and testing the proposed algorithm’s outcome, while various state-of-the-art algorithms were used as benchmark models. The performance of the proposed extension was evaluated using several metrics being: loss, accuracy, speed, and reward values, where the results of comparison showed the superiority of the novel extension in all scenarios for most of the exploited metrics. The experiment results were confirmed using the complexity and the robustness aspects.

Résumé: The complexity of taking decisions for an autonomous vehicle (AV) to avoid road accident fatalities, provide safety, comfort, and reduce traffic raises the need for improvements in the field of decision making. To solve these challenges, many algorithms and techniques were applied, and the most common ones were reinforcement learning (RL) algorithms combined with deep learning techniques. Therefore, in this paper we proposed a novel extension of the popular “SARSA” (State-Action-Reward-State-Action) RL technique called “Harmonic SK Deep SARSA” that takes advantage of the stability which SARSA algorithm provides and uses the notion of similar and cumulative states saved in an alternative memory to enhance the stability of the algorithm and achieve remarkable performance that SARSA could not accomplish due to its on policy nature. Through the investigation of our novel extension the adaptability of the algorithm to unexpected situations during learning and to unforeseen changes in the environment was proved while reducing the computational load in the learning process and increasing the convergence rate that plays a key role in upgrading decision making application that require numerous real time consecutive decisions, including autonomous vehicles, industrial robots, gaming, aerial navigation... The novel algorithm was tested in a gym environment simulator called “Highway-env” with multiple highway situations (multiple lanes configurations, highway with dynamic number of lanes (from 4-lane to 2-lane, from 4-lane to 6-lane), merge) with numerous dynamic obstacles. For the purpose of comparison, we used a benchmark of cutting edge algorithms known for their prominent performance. The experimental results showed that the proposed algorithm outperformed the comparison algorithms in learning stability and performance that were validated by the following metrics: average loss value per episode, average accuracy per episode, maximum speed value reached per episode, average speed per episode, and the total reward per episode.

Résumé: The challenge to guarantee healthy aging has become a major social concern. Due to the cognitive deficits related to age, hectic daily activities, and multiple medications prescriptions, the elders often tend to forget their pills intakes. This has a colossal impact on their health and life. Moreover, the recent pandemic of COVID-19 has accentuated the importance to provide independent and autonomous living for the elderly. This paper presents a pill management system based on IoT intended for aged individuals. The proposed system is a smart pill dispenser associated with a mobile application. The main actors of the system are the patient, the doctor, the pharmacist, and the patient’s caregiver and/or relative, each having restricted access to the system via specific credentials. The prescription is directly edited on the mobile application by the doctor and the scheduling and filling of the pillbox is done wirelessly by the pharmacist. The reminding of medications intakes in this system is done gradually to help the patient adhere to his/her prescription. First, it notifies the patient about his/her scheduled pill by a mobile notification then via the pill dispenser using LCD, LED, and buzzer. The implemented system also allows the doctor and caregiver to keep a tab of the patient’s intakes. Furthermore, the pill dispenser is featured with a locking mechanism to ensure medication dosage control and prevent drugs abuse. Experiments show that the proposed system is appreciated by the elderly and encourages them to take their pills successfully and safely without causing any disturbance.

Résumé: Nowadays, we are dealing with panic and unpleasant situations in which, we are constrained to make crucial decisions in a limited delay, due to the mixed emotions that may affect our decision, especially FEAR, this kind of emotion occurs when unwanted or uncontrollable events are present in the environment. These recent years, fear modelling has been well researched and since this emotion is usually associated with the fact that one or more fundamental desires are at stake Unluckily, most of these models miss that FEAR does not always occur similarly in all agents. This paper proposes a new conceptual architecture with a new component by extending BDI logic with the emotion of FEAR, so that the new Emotional-BDI agents may better cope with extremely dynamic unpleasant situations in their surroundings. We also address how we verify the emotional properties by employing a model checker NuSMV. The proposed architecture confirms that NuSMV can be applied to verify the emotional specifications we can program agents that are capable of reasoning over emotions, our experimental results indicate the viability and efficiency of our model.

Résumé: The combination of NFC with mobile connectivity could achieve many benefits and push innovation in the transport field. Ticketing systems provide travel card for a fixed duration which is usable more than one travel, literature abounds papers talking about the ticketing area but none of them address the expiry of the validity of travel cards for the final travel; because some passengers access with valid travel cards which become invalid during the last travel without their knowledge, making them in illegal situation. To overcome this problem, this paper proposes, in addition to emulating principally the travel card on NFC mobile phone, a callback mechanism in order to make utilization of M-Card a pleasurable experience for passengers. We started from uml2 to describe the architecture and design system. A system prototype under Android is achieved for transportation ticket implementing its main processe

Résumé: Pour développer un système embarqué avec un rapport coût / performance raisonnable, les concepteurs s’orientent vers la conception conjointe matériel/logiciel (M/L) ou codesign. Cette méthodologie comporte une suite d’étapes dont les résultats de la première est une spécification exécutable et la dernière est un prototype virtuel du système. Le partitionnement M/L, une étape cruciale du processus de Co design, a pour finalité de diviser le comportement d’un système en deux sous-ensembles de fonctions, l’un matériel et l’autre logiciel. Le problème de partitionnement M/L étant NP-difficile, la littérature foisonne d’approches complémentaires pour remédier approximativement à ce problème. Dans ce papier, nous proposons un nouveau algorithme hybride et fiable de partitionnement M/L basé sur le principe de l’algorithme Late Acceptance Hill Climbing(LAHC) et de l’algorithme génétique (AG). LAHC est connue pour sa recherche locale efficace, conjuguée à la potentialité importante de recherche globale de l’AG. L’algorithme hybride, a été appliqué et a donné de meilleurs résultats, comparés à ceux de l’état de l’art.

Résumé: Aujourd’hui, les architectures orientées services et les services Web sont importants et nécessaires pour le développement des systèmes d’information interopérables. Cependant, ces technologies ne permettent pas de prendre en charge l’aspect sémantique. Dans cet article, nous présentons d’abord une classification et une comparaison des travaux les plus pertinents dans le domaine de l’interopérabilité sémantique via les services Web. Ensuite, nous proposons une architecture et une approche orientées services pour la mise en œuvre du mécanisme d’interopérabilité, suivie d’une validation à l’aide d’une étude de cas. L’architecture fournie est structurée en trois couches. La première est une architecture orientée services alors que la deuxième concerne l’architecture permettant de capturer la sémantique des services obtenus de la première couche. Elle est basée sur la notion d’ontologies et de contexte, et constitue le modèle de référence pour annoter les descriptions des services Web. La dernière couche offre une architecture pour l’interopérabilité sémantique exploitant les deux premières couches et dont le noyau est une infrastructure de médiation. L’approche proposée est caractérisée par sa simplicité et sa flexibilité, elle exploite les normes existantes des services Web en représentant explicitement leurs hétérogénéités sémantiques, ce qui permet de les résoudre automatiquement.

Résumé: Le fossé séparant la production des systèmes et l’évolution technologique n’a cessé de croître ces derniers temps, conduisant à une réutilisation de composants préconçus et pré-vérifiés appelés IP (Intellectual Property). L’essor de ces derniers ne va pas sans rencontrer quelques difficultés. Nous citons entre autres un manque de standards pour la réalisation des IPs, rendant une intégration voire une communication difficile à établir entre les mêmes composants d’une application, un intéressement seulement à certains types d’IPs matériel ou logiciel et une approche de développement généralement incomplète. Dans ce papier, nous présentons une démarche complète de développement des IPs en partant d’un métamodèle en UML prenant en charge les deux descriptions matérielles et logicielles en utilisant C++ comme langage de programmation et SystemC comme langage de description matérielle. Notre approche a été expérimentée sur une unité arithmétique et logique paramétrée avec SystemC et UML. Les outils Cofluent et Quartus II nous ont servi à obtenir les différents modèles. Les résultats sont encourage

Résumé: System-on-Chip (SoC) designs are increasingly becoming more complex. One of the major constraints is the time to market New design methods are necessary, and the tendency is with the integration of the software and hardware parts on the same chip. Efficient on-chip communication architectures are critical for achieving desired performance in these systems Thus, the development of codesign’s modern methods and the appearance of hardware description languages (HDL) based on C/C++ such as SystemC or specC allowing to employ the same language to describe the software and the hardware, and returning of this fact easier and more effective Co-simulation. These methods would be able to generate an optimal solution starting from a functional specification by reducing the time and the cost of the design. Thus, one of the main objectives of this paper is the development of a SystemC platform for multiprocessors architectural exploration at the compromise level (TLM) by using SystemC/TLM. It must lead to partition system into hw/sw and also to validate it by simulation or to move easily modules from hardware to software (or vice versa) during the architectural exploration. Except for the software task priorities that could be modified, we only need to recompile and simulate

Résumé: Cybersecurity is a compelling need for all organizations, compounded by rapid technological change and an evolving cyberattack landscape. Many areas have been explored to address the changing security requirements, and one of them is based on blockchain technology. In recent years, blockchain technology has become a top priority for many organizations. Its range of applications (smart contracts, secure transaction recording, peer-to-peer, etc.) has grown rapidly due to its inherent robustness to hackers. In this chapter, we introduce cybersecurity and present its applications in blockchain, as well as provide existing vulnerabilities and potential attacks in the state of the art to perceive the use of cybersecurity in blockchain. We also analyse the main cybersecurity challenges and how they might be addressed through advances in blockchain technology.

Résumé: System-on-Chip (SoC) designs are increasingly becoming more complex. One of the major constraints is the time to market . New design methods are necessary, and the tendency is with the integration of the software and hardware parts on the same chip. Efficient on-chip communication architectures are critical for achieving desired performance in these systems Thus, the development of codesign's modern methods and the appearance of hardware description languages (HDL) based on C/C++ such as SystemC or specC allowing to employ the same language to describe the software and the hardware, and returning of this fact easier and more effective Co-simulation. These methods would be able to generate an optimal solution starting from a functional specification by reducing the time and the cost of the design. Thus, one of the main objectives of this paper is the development of a SystemC platform for multiprocessors architectural exploration at the compromise level (TLM) by using SystemC/TLM. It must lead to partition system into hw/sw and also to validate it by simulation or to move easily modules from hardware to software (or vice versa) during the architectural exploration. Except for the software task priorities that could be modified, we only need to recompile and simulate.