Tesis sobre el tema "Network Monitoring Objects"

The goal with this thesis work has been to identify what a telecommunications company should monitor and to find a network monitoring system that can monitor these identified objects on two different platforms: Windows and Linux. The network monitoring system has been implemented in a telecommunications company’s environment and this thesis presents how the system monitors their environment. The subject for this thesis work is within network monitoring. The problem formulation has been answered by conducting a literature study and by testing network monitoring systems’ features in a lab environment. The sources used in the literature study consists of scientific articles and other articles found on the web. The lab environment consisted of virtual machines that runs Linux or Windows as an operating system.   The purpose of the work was to enlighten Cellip in what objects they should monitor and to help the company to monitor them by implementing a network monitoring system. Cellip is a telecommunications company that provides IP-telephony services through Session Initiation Protocol. The limits of this thesis work are based on what their environment supports in terms of monitoring. Cellip’s environment consists of Linux and Windows servers, Cisco switches and firewalls, and Sonus Session Border Controllers.   In summary, the result of this thesis gives the reader information about what a telecommunications company with a similar environment to Cellip should monitor, what three systems that can monitor these objects, which of the three systems that has most automatized features and finally how the chosen system Datadog monitors and presents the objects. Some of the objects that are important to monitor is: memory, disk storage, latency, packet loss. In conclusion, this thesis presents a monitoring baseline for telecommunication companies with a similar environment to Cellip.

The body of work hereby presented aims at better defining the specific mechanisms underlying cognitive control and flexibility, and to investigate the neural substrates that might support these dynamics. More specifically, the anterior cingulate (ACC), dorsolateral prefrontal (dlPFC) and frontopolar (FPC) cortices have been proposed to play a fundamental role in monitoring and detecting the presence of environmental contingencies that require the recruitment of cognitive control (such as competition between responses in the presence of conflicting information), implementing cognitive control, and supporting higher-order cognitive processing, respectively. This thesis investigates the effects of damage to these regions, and of interference with their activity, on these processes. It also argues for the importance of dissociating possible separate cognitive control components that might differently contribute to behavioural adjustments (such as caution and attention/task-relevant processing), and provides one of the first attempts to quantify them within the parameters of a mathematical model of choice response-time, the Linear Ballistic Accumulator (LBA). The results confirm the crucial role of the dlPFC in modulating behavioural adjustments, as both damage and interference with this region’s activity significantly affect measures of conflict-induced behavioural adaptation. It is hypothesized that dlPFC might drive behavioural adjustments by encoding recent conflict history and/or supporting the automatization of a newly advantageous behavioural strategy during the early stages after a change in conflict levels. When a task does not involve competition between a habit and instructed behaviour, lesions or interference with ACC’s activity do not appear to affect behaviour in a manner that is consistent with the classic conflict-monitoring framework. It is suggested that its role might be better described as a more general monitoring and confirmatory mechanism that evaluates both actual and potential outcomes of an action, in order to proactively guide adjustments away from contextually disadvantageous responses. Finally, lesions to the FPC do not affect abstract-rule integration, but do impair the early stages of acquisition of a new abstract rule, when a previously rewarded rule stops being rewarded, and specifically when acquisition is dependent on self-initiated exploration. This suggests a role for FPC in the evaluation of multiple concurrent options in order to aid the development of new behavioural strategies.

This thesis deals with the monitoring of the sewerage system in the city of Brno. In the research part, there is made an overview of the most used measuring equipment for the monitoring of the sewerage system – level meters, flowmeters, automatic samplers. The practical part is devoted to the monitoring of the sewerage system in Brno for partial evaluation of the benefits of the project „Reconstruction and completion of sewerage in Brno“, realized in 2012 – 2014. Measuring campaign has taken place from June to October 2017. In the course of the work, the instalation of measuring equipment and the collection of measured data was done. After finishing measuring campaign, the data were evaluated from different aspects – the number of overflows, the quantity of overflowing water, the dilution ratio and the quantity of discharged pollution. Then, the hydraulic and construction technical assement of the pipelines of sewer network and overflow chambers and assements of the quantity of the rainwater were made. The results of the evaluation were compared with the results of the previous monitoring campaingns and with the data obtained before the project has been realized. Based on this, the benefit of the project was determined.

Les réseaux « Internet des Objets » se composent de plusieurs millions d’objets qui possèdent une adresse IP et qui peuvent connecter sur Internet. En général, ces objets sont supposé d’être autonomes et peuvent résoudre des tâches; mesurer, traiter et fournir des informations pour les systèmes connectés et pour les utilisateurs. Aussi, ces réseaux sont vulnérables (c.-à-d. : les éléments peuvent être mobiles et la topologie du réseau peut changer dynamiquement), les changements peuvent influencer le (bon) fonctionnement du réseau. De plus, ils peuvent être alimentés par des batteries de durée de vie limitée, ce que nécessite la réduction de leur consommation.Ce travail de thèse aborde un sujet important dans le domaine de l’Internet des Objets, qui consiste à savoir comment assurer la robustesse et le fonctionnement tolérant aux pannes du réseau pour répondre aux exigences des missions critiques. Avec le large déploiement des services IdO, ce problème est deventé ou de détection de pannes et de sécurité industriel où l’état des objets communicants doit être constamment vérifié pour le rétablissement rapide en cas de problème particulièrement crucial pour les applications telles que le monitorage intelligent de sames de communication inattendus. On cherche alors de minimiser le coût du monitorage et l’utilisation de l’énergie, et aussi les charges additionnelles sur les réseaux.Nous avons proposé un algorithme qui vise à réaliser un placement distribué des moniteurs avec une complexité minimale pour le calcul. L’algorithme proposé fonctionne avec RPL. L’objectif principal est d’augmenter la robustesse dans les réseaux IdO ciblant les applications critiques en temps réel via le monitorage des liaisons dans les DODAGs construits par RPL. Dans notre première contribution, le problème est modélisé comme un problème de couverture minimale des sommets (VCP) sur le DODAG. Nous avons développé un algorithme à temps polynomial qui transforme le DODAG en une décomposition arborescente (Nice-Tree Decomposition) avec une largeur arborescente (treewidth) d’unité. Cette stratégie profite de la spécificité des DODAG et a abouti à une réduction significative de la complexité de la résolution du VCP sur les DODAG. Elle peut être résolue en temps polynomial.La deuxième proposition est un modèle approché pour l’optimisation de l’ordonnancement du rôle de monitorage des nœuds dans les réseaux IdO, afin de maximiser la durée de vie des dispositifs embarqués à ressources limitées, tout en minimisant le coût global du monitorage de réseau. Le monitorage de réseau est très coûteux, en particulier pour les réseaux à ressources limitées tels que l’IdO. Par conséquent, le monitorage doit être économe en énergie et avec des frais généraux minimaux sur la performance normale du réseau. Notre travail correspondant contient une proposition d’un modèle mathématique en trois phases pour assurer l’exigence d’une couverture des moniteurs tout en minimisant la consommation d’énergie de monitorage et les frais de communication.Notre modèle proposé décompose le problème abordé en trois problèmes d’optimisation bien connus, il s’agit du problème de couverture de sommets, problème d’affectation généralisé multi-objectives et problème de voyageur de commerce.Dans cette troisième partie, une approche exacte est proposée pour résoudre le problème décrit dans (Contribution 2). Comme nous avons vu, la décomposition en trois phases ne donne pas la solution exacte. Nous avons donc proposé une formulation exacte du problème qui consiste en un problème de l'affectation minimum des tâches de surveillance avec un fonctionnement de surveillance cyclique. Pour cela, nous avons formulé un programme en nombres entiers binaires. L'ordonnancement optimal garantit la couverture du graphe pour la surveillance avec une consommation d'énergie minimale
By connecting billions of things to the Internet, IoT created a plethora of applications that touch every aspect of human life. Time-sensitive, mission-critical services, require robust connectivity and strict reliability constraints. On the other hand, the IoT relies mainly on Low-power Lossy Networks, which are unreliable by nature due to their limited resources, hard duty cycles, dynamic topologies, and uncertain radio connectivity. Faults in LLNs are common rather than rare events, therefore, maintaining continuous availability of devices and reliability of communication, are critical factors to guarantee a constant, reliable flow of application data.After a comprehensive literature review, and up to our knowledge, it is clear that there is a call for a new approach to monitoring the unreliable nodes and links in an optimized, energy-efficient, proactive manner, and complete interoperability with IoT protocols. To target this research gap, our contributions address the correct assignment (placement) of the monitoring nodes. This problem is known as the minimum assignment problem, which is NP-hard. We target scalable monitoring by mapping the assignment problem into the well-studied MVC problem, also NP-hard. We proposed an algorithm to convert the DODAG into a nice-tree decomposition with its parameter (treewidth) restricted to the value one. As a result of these propositions, the monitor placement becomes only Fixed-Parameter Tractable, and can also be polynomial-time solvable.To prolong network longevity, the monitoring role should be distributed and balanced between the entire set of nodes. To that end, assuming periodical functioning, we propose in a second contribution to schedule between several subsets of nodes; each is covering the entire network. A three-phase centralized computation of the scheduling was proposed. The proposition decomposes the monitoring problem and maps it into three well-known sub-problems, for which approximation algorithms already exist in the literature. Thus, the computational complexity can be reduced.However, the one major limitation of the proposed three-phase decomposition is that it is not an exact solution. We provide the exact solution to the minimum monitor assignment problem with a duty-cycled monitoring approach, by formulating a Binary Integer Program (BIP). Experimentation is designed using network instances of different topologies and sizes. Results demonstrate the effectiveness of the proposed model in realizing full monitoring coverage with minimum energy consumption and communication overhead while balancing the monitoring role between nodes.The final contribution targeted the dynamic distributed monitoring placement and scheduling. The dynamic feature of the model ensures real-time adaptation of the monitoring schedule to the frequent instabilities of networks, and the distributed feature aims at reducing the communication overhead

Questo elaborato si propone di risolvere il problema della ricerca di un parcheggio. Sono tante le grandi città che oggi investono in tecnologia, nel tentativo di fornire servizi intelligenti ai propri cittadini. Questa pratica si è notevolmente sviluppata negli ultimi anni, tanto che è stato coniato il termine Smart City. Spesso, in alcune aree urbane ed in determinati orari, il cittadino per trovare un parcheggio libero impiega molto tempo ed è costretto ad effettuare una ricerca nelle zone limitrofe, sperando che questa porti ad un esito positivo, nel più breve tempo possibile. Ciò non accadrebbe se ci fosse un servizio capace di informare l’utente sull’ubicazione del parcheggio libero più vicino rispetto alla propria posizione. La tecnologia che manca è un sistema di visione robusto per il monitoraggio degli spazi di parcheggio. Si è proceduto, quindi, con una ricerca su come il problema sia stato già affrontato in letteratura. Da tale ricerca è emerso che le attuali tecnologie consentono solo di configurare il sistema e le telecamere in modo da analizzare determinate aree delle immagini prodotte. Ciò consentirebbe soltanto la realizzazione di questo servizio per pochi parcheggi, e non si applicherebbe ad un numero più elevato quale quello relativo alle aree parcheggio di un’intera città. Tali considerazioni ci hanno portato ad individuare una tecnologia capace di localizzare un parcheggio, estrapolando le informazioni da una base di conoscenza. Tutto ciò è realizzabile tramite l’Object Detection. Uno dei sistemi di Object Detection più all’avanguardia è Yolo, un sistema real-time che si occupa del rilevamento di oggetti. Questo sistema utilizza una Convolutional Neural Network per effettuare la previsione di una Bounding Box, che delimita l’oggetto all’interno dell’immagine ed effettua la classificazione dell’oggetto presente all’interno di tale Box. Per poter addestrare questo tipo di sistema è stato utilizzato un Dataset presente in letteratura, PKLot.

While it is generally accepted that environmental data are critical to understanding environmental phenomena, there are yet improvements to be made in their consistent collection, curation, and sharing. This thesis describes two research efforts to improve two different aspects of hydrologic data collection and management. First described is a recipe for the design, development, and deployment of a low-cost environmental data logging and transmission system for environmental sensors and its connection to an open source data-sharing network. The hardware is built using several low-cost, open-source, mass-produced components. The system automatically ingests data into HydroServer, a standards-based server in the open source Hydrologic Information System (HIS) created by the Consortium of Universities for the Advancement of Hydrologic Sciences Inc (CUAHSI). A recipe for building the system is provided along with several test deployment results. Second, a connection between HydroServer and HydroShare is described. While the CUAHSI HIS system is intended to empower the hydrologic sciences community with better data storage and distribution, it lacks support for the kind of “Web 2.0” collaboration and social-networking capabilities that are increasing scientific discovery in other fields. The design, development, and testing of a software system that integrates CUAHSI HIS with the HydroShare social hydrology architecture is presented. The resulting system supports efficient archive, discovery, and retrieval of data, extensive creator and science metadata, assignment of a persistent digital identifier such as a Digital Object Identifier (DOI), scientific discussion and collaboration around the data and other basic social-networking features. In this system, HydroShare provides functionality for social interaction and collaboration while the existing HIS provides the distributed data management and web services framework. The system is expected to enable scientists, for the first time, to access and share both national- and research lab-scale hydrologic time series in a standards-based web services architecture combined with a social network developed specifically for the hydrologic sciences.These two research projects address and provide a solution for significant challenges in the automatic collection, curation, and feature-rich sharing of hydrologic data.

Les réseaux de capteurs sans fil (aussi appelés LLNs en anglais) sont des réseaux contraints composés de nœuds ayant de faibles ressources (mémoire, CPU, batterie). Ils sont de nature très hétérogène et utilisés dans des contextes variés comme la domotique ou les villes intelligentes. Pour se connecter nativement à l’Internet, un LLN utilise une passerelle, qui a une vue précise du trafic transitant entre Internet et le LLN du fait de sa position. Le but de cette thèse est d’exposer comment des fonctionnalités peuvent être ajoutées à une passerelle d’un LLN dans le but d’optimiser l’utilisation des ressources limitées des nœuds contraints et d’améliorer la connaissance de leur état de fonctionnement. La première contribution est un estimateur non intrusif utilisant le trafic passant par la passerelle pour inférer l’utilisation de la radio des nœuds contraints. La seconde contribution adapte la durée de vie d’informations mises en cache (afin d’utiliser les ressources en cache au lieu de solliciter le réseau) en fonction du compromis entre le coût et l’efficacité. Enfin, la troisième contribution est Makesense, un framework permettant de documenter, d’exécuter et d’analyser une expérience pour réseaux de capteurs sans fil de façon reproductible à partir d’une description unique
Low-Power and Lossy Network (LLN)s are constrained networks composed by nodes with little resources (memory, CPU, battery). Those networks are typically used to provide real-time measurement of their environment in various contexts such as home automation or smart cities. LLNs connect to other networks by using a gateway that can host various enhancing features due to its key location between constrained and unconstrained devices. This thesis shows three contributions aiming to improve the reliability and performance of a LLN by using its gateway. The first contribution introduce a non-intrusive estimator of a node radio usage by observing its network traffic passing through the gateway. The second contribution offers to determine the validity time of an information within a cache placed at the gateway to reduce the load on LLNs nodes by doing a trade-off between energy cost and efficiency. Finally, we present Makesense, an open source framework for reproducible experiments that can document, execute and analyze a complete LLN experiment on simulation or real nodes from a unique description

Les réseaux de capteurs permettent une surveillance multi-paramètres de zones d'étendue limitée grâce à la coopération d'un ensemble de récepteurs déployés in-situ qui gèrent l'acquisition, le traitement et le transfert de données. Afin de coupler le concept de réseaux de capteurs et un positionnement précis des récepteurs à l'aide de puces GPS mono-fréquence, l'Institut National de l'Information Géographique et Forestière (IGN) a mis au point le Geocube.La première partie de ce travail a été consacrée au développement d'une stratégie d'acquisition, de transfert et de traitement des données GPS des Geocubes pour permettre un positionnement précis et en temps-réel des récepteurs au sein du réseau. Un traitement utilisant les observations de phase GPS ainsi qu'un filtrage de Kalman a été adopté. Cependant, les séries temporelles de positions brutes sont entachées de l'effet des multitrajets. Diverses méthodes d'atténuation de ce phénomène sont alors proposées. Au final, une précision infra-centimétrique à millimétrique est atteinte.La seconde partie de cette thèse a été consacrée à l'application de réseaux de Geocubes pour l'étude d'objets géophysiques. Deux sites d'étude ont été sélectionnés: le glissement de terrain de Super-Sauze dans la vallée de l'Ubaye et le glacier d'Argentière dans le massif du Mont-Blanc. La dynamique des objets d'intérêt peut alors être étudiée à une échelle infra-journalière grâce à la précision et à la grande résolution temporelle du positionnement des Geocubes. De plus, la densité des réseaux de mesure et leur facilité d'installation permet d'instrumenter la grande majorité des points où un besoin de surveillance est identifié
Wireless Sensor Networks (WSN) allow a multi-parameters monitoring of small extend areas thanks to cooperative data acquisition, transfer and processing. In order to combine WSN with a precise positioning of the receivers within the network using single frequency GPS modules, the Geocube has been developed by the French National Institute of Geographic and Forest Information (IGN-France). The first part of this work focused on GPS data management and processing to allow the relative positioning of the Geocubes within a local network. To this end, a processing method customized for Geocube data and WSN environment was developed. It is based on the use of GPS carrier phase double differences and a Kalman filtering. Due to the basic GPS antenna used into the Geocube to minimize its price and its size, multipath affect position time series. Various strategies are proposed for multipath mitigation, and finally a sub-centimeter to millimeter level accuracy is reached for relative positioning depending on measurement conditions.The second part of this work was devoted to the use of Geocube networks for geophysical structures monitoring. Two test sites were selected: the Super-Sauze landslide (Ubaye valley, Alpes de Haute-Provence, France) and the Argentière glacier (Mont-Blanc massif, Haute-Savoie, France). The dynamics of the studied areas was investigated at a sub-daily time scale thanks to the high accuracy and the high time resolution of positioning time series derived from Geocubes. In addition, positioning data were acquired quite everywhere a deformation measurement was needed thanks to the low-cost of Geocubes and their easy set up

Railways are a main element of a sustainable transport policy in several countries as they are considered a safe, efficient and green mode of transportation. Owing to these advantages, there is a cumulative request for the railway industry to increase the performance, the capacity and the availability in addition to safely transport goods and people at higher speeds. To meet the demand, large adjustment of the infrastructure and improvement of maintenance process are required.  Inspection activities are essential in establishing the required maintenance, and it is periodically required to reduce unexpected failures and to prevent dangerous consequences.  Maintenance of railway catenary systems is a critical task for warranting the safety of electrical railway operation.Usually, the catenary inspection is performed manually by trained personnel. However, as in all human-based inspections characterized by slowness and lack of objectivity, might have a number of crucial disadvantages and potentially lead to dangerous consequences. With the rapid progress of artificial intelligence, it is appropriate for computer vision detection approaches to replace the traditional manual methods during inspections.  In this thesis, a strategy for monitoring the health of catenary wires is developed, which include the various steps needed to detect anomalies in this component. Moreover, a solution for detecting different types of wires in the railway catenary system was implemented, in which a deep learning framework is developed by combining the Convolutional Neural Network (CNN) and the Region Proposal Network (RPN).

L’objectif de cette de thèse est d’étudier à la fois des solutions matérielles et logicielles pour améliorer les conditions de travail des sapeurs-pompiers. Il s’agit de développer un système intelligent basé sur l’internet des objets pour surveiller l'état de santé des pompiers et aider à les localiser lors des interventions. Dans la première partie de la thèse, nous avons étudié et proposé plusieurs approches permettant de réduire la consommation d’énergie du système afin de maximiser sa durée de vie. La première approche présente un modèle de prédiction basé sur la corrélation temporelle entre les mesures collectées par le même capteur. Il permet de réduire la quantité de données collectées et transmises au centre de contrôle. Ce modèle est exécuté à la fois par le capteur et le centre et qui s'auto-adapte en fonction de l’écart constaté entre les mesures réelles collectées et les mesures prédites. Une deuxième version de cette approche a été étudiée pour prendre en considération la perte de message et la synchronisation entre le capteur et le centre de contrôle. D’un autre côté et pour réduire davantage la consommation d’énergie, nous avons couplé l’approche de prédiction avec un algorithme de collecte de données adaptatif permettant de réduire l’activité du capteur et le taux d’échantillonnage. Toutes ces approches ont été testées via des simulations et de l’implémentation réelle. Les résultats obtenus montrent l’efficacité de ces approches en termes de réduction de la consommation d’énergie tout en gardant l’intégrité de données. La deuxième partie de cette thèse est dédiée au traitement des données issues des interventions des sapeurs-pompiers. Nous avons étudié plusieurs méthodes de clustérisation permettant un prétraitement de données avant l’extraction des connaissances. D’un autre côté, nous avons appliqué des méthodes d'apprentissage profond sur un grand ensemble de données concernant 200.000 interventions qui ont eu lieu pendant une période de 6 ans dans le département du Doubs, en France. Le but de cette partie était de prédire le nombre d’interventions futures en fonction de variables explicatives externes, pour aider les pompiers à bien gérer leurs ressources
In this thesis, we present a multilevel scheme consisting of both hardware and software solutions to improve the daily operational life of firefighters. As a core part of this scheme, we design and develop a smart system of wearable IoT devices used for state assessment and localization of firefighters during interventions. To ensure a maximum lifetime for this system, we propose multiple data-driven energy management techniques for resource constraint IoT devices. The first one is an algorithm that reduces the amount of data transmitted between the sensor and the destination (Sink). This latter exploits the temporal correlation of collected sensor measurements to build a simple yet robust model that can forecast future observations. Then, we coupled this approach with a mechanism that can identify lost packets, force synchronization, and reconstruct missing data. Furthermore, knowing that the sensing activity does also require a significant amount of energy, we extended the previous algorithm and added an additional adaptive sampling layer. Finally, we also proposed a decentralized data reduction approach for cluster-based sensor networks. All the previous algorithms have been tested and validated in terms of energy efficiency using custom-built simulators and through implementation on real sensor devices. The results were promising as we were able to demonstrate that our proposals can significantly improve the lifetime of the network. The last part of this thesis focusses on building data-centric decision-making tools to improve the efficiency of interventions. Since sensor data clustering is an important pre-processing phase and a stepstone towards knowledge extraction, we review recent clustering techniques for massive data management in IoT and compared them using real data for a gas leak detection sensor network. Furthermore, with our hands on a large dataset containing information on 200,000 interventions that happened during a period of 6 years in the region of Doubs, France. We study the possibility of using Machine Learning to predict the number of future interventions and help firefighters better manage their mobile resources according to the frequency of events

Dans le domaine de la fabrication, la détection d’anomalies telles que les défauts et les défaillances mécaniques permet de lancer des tâches de maintenance prédictive, qui visent à prévoir les défauts, les erreurs et les défaillances futurs et à permettre des actions de maintenance. Avec la tendance de l’industrie 4.0, les tâches de maintenance prédictive bénéficient de technologies avancées telles que les systèmes cyberphysiques (CPS), l’Internet des objets (IoT) et l’informatique dématérialisée (cloud computing). Ces technologies avancées permettent la collecte et le traitement de données de capteurs qui contiennent des mesures de signaux physiques de machines, tels que la température, la tension et les vibrations. Cependant, en raison de la nature hétérogène des données industrielles, les connaissances extraites des données industrielles sont parfois présentées dans une structure complexe. Des méthodes formelles de représentation des connaissances sont donc nécessaires pour faciliter la compréhension et l’exploitation des connaissances. En outre, comme les CPSs sont de plus en plus axées sur la connaissance, une représentation uniforme de la connaissance des ressources physiques et des capacités de raisonnement pour les tâches analytiques est nécessaire pour automatiser les processus de prise de décision dans les CPSs. Ces problèmes constituent des obstacles pour les opérateurs de machines qui doivent effectuer des opérations de maintenance appropriées. Pour relever les défis susmentionnés, nous proposons dans cette thèse une nouvelle approche sémantique pour faciliter les tâches de maintenance prédictive dans les processus de fabrication. En particulier, nous proposons quatre contributions principales: i) un cadre ontologique à trois niveaux qui est l’élément central d’un système de maintenance prédictive basé sur la connaissance; ii) une nouvelle approche sémantique hybride pour automatiser les tâches de prédiction des pannes de machines, qui est basée sur l’utilisation combinée de chroniques (un type plus descriptif de modèles séquentiels) et de technologies sémantiques; iii) a new approach that uses clustering methods with Semantic Web Rule Language (SWRL) rules to assess failures according to their criticality levels; iv) une nouvelle approche d’affinement de la base de règles qui utilise des mesures de qualité des règles comme références pour affiner une base de règles dans un système de maintenance prédictive basé sur la connaissance. Ces approches ont été validées sur des ensembles de données réelles et synthétiques
In the manufacturing domain, the detection of anomalies such as mechanical faults and failures enables the launching of predictive maintenance tasks, which aim to predict future faults, errors, and failures and also enable maintenance actions. With the trend of Industry 4.0, predictive maintenance tasks are benefiting from advanced technologies such as Cyber-Physical Systems (CPS), the Internet of Things (IoT), and Cloud Computing. These advanced technologies enable the collection and processing of sensor data that contain measurements of physical signals of machinery, such as temperature, voltage, and vibration. However, due to the heterogeneous nature of industrial data, sometimes the knowledge extracted from industrial data is presented in a complex structure. Therefore formal knowledge representation methods are required to facilitate the understanding and exploitation of the knowledge. Furthermore, as the CPSs are becoming more and more knowledge-intensive, uniform knowledge representation of physical resources and reasoning capabilities for analytic tasks are needed to automate the decision-making processes in CPSs. These issues bring obstacles to machine operators to perform appropriate maintenance actions. To address the aforementioned challenges, in this thesis, we propose a novel semantic approach to facilitate predictive maintenance tasks in manufacturing processes. In particular, we propose four main contributions: i) a three-layered ontological framework that is the core component of a knowledge-based predictive maintenance system; ii) a novel hybrid semantic approach to automate machinery failure prediction tasks, which is based on the combined use of chronicles (a more descriptive type of sequential patterns) and semantic technologies; iii) a new approach that uses clustering methods with Semantic Web Rule Language (SWRL) rules to assess failures according to their criticality levels; iv) a novel rule base refinement approach that uses rule quality measures as references to refine a rule base within a knowledge-based predictive maintenance system. These approaches have been validated on both real-world and synthetic data sets

Les réseaux de capteurs sans fil qui conviennent pour vaste domaine d’applications, constituent une solution prometteuse qui répond à toute exigence de surveillance continue. L’autonomie énergétique des nœuds constitue un facteur de vulnérabilité qui influe directement leur longévité et la capacité du réseau à assurer longuement la couverture d’une zone géographique d’intérêt. La gestion de consommation énergétique représente la seule approche pour accroître la durée de vie de ces réseaux et leur conférer une autonomie raisonnable. Des solutions logicielles proposées à travers les protocoles MAC, apportent des améliorations significatives à la minimisation de la dépense énergétique des nœuds. Elles permettent de réduire les périodes d’écoute du canal qui, représente l’opération la plus coûteuse en termes d’énergie dans le fonctionnement des nœuds de capteurs sans fil. Néanmoins, se limiter à ces solutions n’est pas suffisant pour garantir une longévité acceptable. La seule méthode pour optimiser la conservation d’énergie dans les RCSFs est de mettre chaque nœud constamment en mode faible puissance et d’utiliser un mécanisme de télé-réveil à travers des signaux de réveil. Cela implique, l’utilisation de circuits de réveil de faible consommation qui assurent la surveillance de canal et qui déclenchent le réveil des nœuds uniquement à chaque fois qu’événement d’intérêt se produit. Dans ce contexte, une quantité importante de travaux ont proposés l’utilisation d’un mécanisme d’adressage (adresses MAC ou d’autres informations binaires), pour permettre aux nœuds non concernés de retourner rapidement dans son état de sommeil. Cette démarche est intéressante, mais implique toutefois une dépense énergétique non négligeable, liée à la réception et au traitement des informations d’adresse au niveau de tous les nœuds. La solution la plus efficace énergétiquement serait l’utilisation d’une autre forme d’adresse. Cette thèse s’inscrit dans le contexte de minimisation de la consommation énergétique des RCSFs par la mise en œuvre d’un adressage qui permet aux nœuds de recevoir et de traiter les signaux de réveil, sans allumer leur module de communication principal. Il s’agit pour nous de supprimer la dépense énergétique liée à l’allumage du module RF et à la réception de paquets d’adresse, en se tournant vers l’exploitation de la durée des signaux de réveils. Notre solution se repose sur les caractéristiques matérielles du microcontrôleur (IRQ, Timer/Counter) des nœuds de capteurs. Elle permet de réduire les complexités liées aux conditionnements des signaux de réveils. Notre solution est implémentée sur un réseau de petite taille. Elle est évaluée expérimentalement et ses performances énergétiques sont comparées à celles d’un schéma classique de télé-réveil sans mécanisme d’adressage et à celles d’un schéma classique basé sur le duty-cycling
Wireless sensor networks that are suitable for a wide range of applications, represent a promising solution that meets any requirement for continuous monitoring. The energy autonomy of sensor nodes constitutes a vulnerability factor that directly affects their longevity and the capacity of the network to ensure long coverage of the geographical area of interest. Energy consumption management is the only way to increase the lifespan of these networks and to give them a reasonable autonomy. Software solutions proposed through MAC protocols, bring significant improvements to the minimization of the energy expenditure of sensor nodes. They reduce the idle-listening periods which represents the most expensive operation in terms of energy, in the operation of the wireless sensor nodes. However, Focusing lonely on these solutions is not enough to guarantee acceptable longevity. The only way to optimize energy conservation in the WSN is to constantly put each node in low power mode and use a wakeup mechanism through wake-up signals. This involves the use of low-power wake-up circuits that provide channel monitoring, and trigger node wake-up only whenever event of interest occurs. In this context, a significant amount of work has proposed the use of an addressing mechanism (MAC addresses or other binary informations), to allow non-concerned nodes to quickly return to their sleep state. This approach is interesting, but involves a significant energy expenditure, related to address information’s reception and processing at all nodes. The most energy efficient solution would be the use of another type of address. This thesis is part of the context of minimizing the energy consumption of the WSN, using an addressing system that allows sensor nodes to receive and process the wake-up signals, without turning on their main communication module. It is to eliminate the energy expenditure related to the RF module’s activation and the reception of address packets, by exploiting wakeup signals duration. Our solution is based on the hardware characteristics of the microcontroller (IRQ, Timer/Counter) of sensor nodes. It reduces the complexities related to wakeup signals conditioning. Our solution is implemented on a small network. Its evaluations were done experimentally and its energy performance is compared to a conventional wake-up mechanism without addressing,and a conventional scheme based on duty-cycling

博士
臺灣大學
資訊管理學研究所
98
There are two important challenges in WSNs design. One is to construct an efficient WSN for applications to guarantee desired quality of service (QoS). The other challenge is to prolong the lifetime of WSNs. From application viewpoint, the abilities of environment surveillance, object intrusion detection, and object tracking have to support the QoS. Besides, it is difficult to recharge or replace the battery for numerous sensors in the most scenarios. Therefore, how to prolong the lifetime of WSNs also becomes a key issue. 　　In this dissertation, we focus on the network planning problem to support object monitoring and object tracking services from various perspectives. We develop five algorithms to solve optimization problems based on Lagrangean relaxation method, simulation techniques, and heuristic approaches. In addition, we develop one prediction-based algorithm based on modified Viterbi algorithm to solve object tracking problem. We present each topic briefly as follows: 　　For boundary monitoring problem, we propose two algorithms, BMAFS and BMAMS, to support boundary monitoring services. The BMAFS is to construct boundary monitoring for grouping capabilities, and it tries to find the maximum k groups of sensors for boundary monitoring of the sensor field to prolong the system lifetime. In the test problems, the experiment results show that the proposed algorithm achieves optimality in the boundary monitoring for grouping capabilities. The BMAMS is to address the problem of boundary node relocation, and it can move previously deployed sensors to cover uncovered check points due to failure of other nodes or battery exhaustion of other nodes. The mechanism can further prolong the system lifetime. The experiment results show that the proposed BMAMS gets effectiveness in the boundary monitoring services for mobile and grouping capabilities. 　　For in-depth defense problem, we propose two algorithms, LDA and NLDA, to support in-depth defense services. The LDA is to construct layered defense for wireless sensor networks of grouping capabilities. It tries to find the maximum k groups of sensors for layered defense of the monitoring region to prolong the system lifetime. The experiment results show that the proposed LDA gets efficiency in the layered defense for grouping capabilities. The NLDA is to construct non-layered defense of supporting different types of intruders for grouping capabilities, and it tries to find the maximum k groups of sensors for non-layered defense subject to the constraints of defense rate, early warning rate, battery capacity, intruder behaviors, and defender strategies. The NLDA can prolong the system lifetime and provide lead time alarms. The experiment results show that the proposed NLDA gets applicability and effectiveness in the non-layered defense services of supporting different types of intruders for grouping capabilities. 　　For object tracking problem, we propose two algorithms, TOTA and POTA, to support object tracking services. The TOTA is to construct an object tracking tree for object tracking. Such tree-based algorithm can achieve energy-efficient object tracking for given arbitrary topology of sensor networks. The experiment results show that the proposed TOTA gets a near optimization in the energy-efficient object tracking. Furthermore, the algorithm is efficient and scalable in terms of the running time. The POTA is to construct a dynamic prediction-based algorithm for object tracking. Such the POTA can minimize the number of nodes participating in the tracking activities, minimize out of tracking probability, and maximize the accuracy of object predicted position. The POTA can prolong the system lifetime. 　　The experiment results show that all six algorithms can support object monitoring and tracking services efficiently.

碩士
國立臺灣科技大學
工程技術研究所
86
The main purpose of this study is to develop a network-based controlling and moni-toring system of workshop using distributed object environment. In a CIM factory, manycontrollers were integrated to achieve the manufacturing automation by variety of commu-nicating method. In this study, distributed object environment is used in analysis and des-ign of the controlling and monitoring system of workshop. Before implementation, the UML, DCOM notation and OOA/OOD method were employed to model the system. In this study, the controlling and monitoring system of the Vision System and the Remote Sensing System of Automation & Control Center in NTUST was implemented by using Microsoft DCOM. Every Object in the distributed object environment has a trans-parent characteristic of operation. It can be easily controlled and executed parallelly under simplified communication mechanism. The process objects provide automated control ofmanufacturing process. Other objects operate the equipment objects, created by wrapping the opening interface of equipment. The data objects support data persistence, data syn-chronization and data query functions of controlling and monitoring system. With transac-tion server, the data objects can support atomic transaction and rollback functions for ob-ject status. From the practical operation of these machines, it was shown that the distributed object in the method has the following advantages: reusability, scalability and reliability. Besides, many homepage editors can be used to design the User Interface (UI) for system monitoring and controlling. The user will easily access the manufacturing information through any WWW browser on the Internet.

碩士
國立東華大學
資訊工程學系
97
Sensor networks make people’s life more convenient, and target tracking is an important application of sensor networks, which can be used in household safety and army detecting. Because of the progress of location techniques and popularity of GPS, sensors are likely capable of detecting location of objects within sensing area in the future. If sensors have mobility, it can increase the coverage and scalability of sensor networks. Based on the assumption mentioned above, we design a complete method and protocol for moving object monitoring and responding messages needed for mission transferring. Every sensor operates in accordance with state diagram by using distribution process The state is sorted into seven main categories: sleep, ready, track, patrol, search, restructure, report. We use communicating information to achieve each goal and catch the timing for switching mode. In order to detect the object immediately, we set up a sensor to monitor the boundary of monitoring area. Once an object enters the area, the sensor will begins to track. When handoff is triggered, the sensor implement will search for another replacement. If object disappears, sensor implement searching to ensure object could be detected continuously and report back the detecting result in compliance with application need. We use three tracing methods, which are target following method, distance closing method and least movement method. In the end we verify the quantity of the objects, speed and differences of the methods in the density of the sensor networks, and from the proportion of every information find out the state distribution that sensor networks enter most frequently in the tracking procedure.

The technologies associated with smart healthcare are a reality nowadays, however in the physical therapy area there is still lack of patient monitoring during the physical rehabilitation and common usage of walking aids by the patients affected by lower limb impairments. Currently there are fewer systems that provide the patient monitoring during the rehabilitation process by physiotherapists, which may lead to less adequate diagnostic techniques for the patient's physical condition. The dissertation presents a solution to this problem by relying on smart equipment used in physical rehabilitation, more precisely a crutch. By embedding multiple smart sensors on crutches, the physiotherapist will be provided appropriate information regarding the interaction between the patient and the walking aids through a mobile application, developed for Android systems, which will receive data from the sensors via Bluetooth. All the data collected will be stored in a local database located on the physiotherapist’s mobile device and also on a remote server, giving the possibility of having a full offline application. This system allows for any session previously done to be consulted, which results in the possibility of visualizing historical values and comparing them with different sessions, allowing the physiotherapist to analyze the evolution of the patients.
As tecnologias associadas à saúde são uma realidade na atualidade, porém na área de fisioterapia ainda há falta de monitorização dos pacientes durante a fisioterapia e o uso de objetos que auxiliam o movimento pelos pacientes afetados por deficiências nos membros inferiores. Atualmente, existem poucos sistemas que proporcionam a monitorização do paciente durante o processo de reabilitação por fisioterapeutas, o que pode levar a técnicas de diagnóstico menos adequadas para a condição física do paciente. A dissertação apresenta uma solução para este problema, contando com equipamentos inteligentes utilizados em fisioterapia, mais precisamente uma muleta. Ao incorporar vários sensores inteligentes em muletas, o fisioterapeuta receberá informações adequadas sobre a interação entre o paciente e as muletas através de uma aplicação móvel, desenvolvida para sistemas Android, que receberá dados dos sensores via Bluetooth. Todos os dados recebidos serão armazenados numa base de dados local localizada no dispositivo móvel do fisioterapeuta e também num servidor remoto para fins de sincronização, dando a possibilidade de ter um uma aplicação completamente offline.