Les réseaux de capteurs sans fil (WSN) jouent un rôle important dans des applications telles que la surveillance de l'environnement, le suivi de sources et le suivi médical, ...etc. Dans les WSN, les capteurs ont la capacité d'effectuer l'échantillonnage des données, des calculs distribués et de fusionner des données. Pour effectuer ces tâches complexes, la synchronisation des horloges et la localisation sont fondamentales et essentielles. Les WSN ont été largement étudiés ces dernières années et la littérature scientifique rapporte de nombreux résultats qui les rendent applicables pour de nombreuses applications. Pour d'autres, la recherche doit encore trouver des solutions à certains des défis posés par la limitation énergétique, la dynamicité et la faible puissance de calcul. Dans le but de contribuer à la recherche sur les WSN, cette thèse propose de nouveaux algorithmes pour la synchronisation d'horloge et la localisation. La synchronisation d'horloge est nécessaire afin que les effectuent de manière efficace la fusion de données. En appliquant l'algorithme de synchronisation d'horloge, les capteurs établissent un consensus temporel et travaillent donc au même rythme. Compte tenu de la dynamicité, des faibles capacités de calcul et de la parcimonie des WSN, un nouvel algorithme de synchronisation décentralisée à impulsions couplées est proposé pour améliorer la précision de la synchronisation. L'avantage de ce type d'algorithme est que les capteurs échangent des impulsions au lieu de paquets, de sorte que non seulement la communication est efficace, mais aussi robuste à toute défaillance des capteurs dans le réseau. La localisation de capteurs a été largement étudiée dans la littérature scientifique. Cependant, la qualité et la précision de la localisation peuvent encore être améliore. Cette thèse applique l'algorithme LSCR (Régression de régions corrélées à signes dominants) au problème de localisation. Avec LSCR, on évalue des régions de confiance avec des niveaux de confiance prescrits, qui fournissent non seulement on emplacement mais aussi la confiance en cet emplacement. Dans cette thèse, plusieurs approches de localisation sont implémentées et comparées. Le résultat de la simulation montre que, sous hypothèses modérées, LSCR obtient des résultats compétitifs par rapport à d'autres méthodes. / Wireless sensor networks (WSNs) play an important role in applications such as environmental monitoring, source tracking, and health care,... In WSN, sensors have the ability to perform data sampling, distributed computing and information fusion. To perform such complex tasks, clock synchronization and localization are two fundamental and essential algorithms. WSNs have been widely studied in the past years, and the scientific literature reports many outcomes that make them applicable for some applications. For some others, research still needs to find solutions to some of the challenges posed by battery limitation, dynamicity, and low computing clock rate. With the aim of contributing to the research on WSN, this thesis proposes new algorithms for both clock synchronization and localization. For clock synchronization, sensors converge their local physical clock to perform data fusion. By applying the clock synchronization algorithm, sensors converge the time difference and therefore work at the same rate. In view of dynamicity, low computing and sparsity of WSN, a new pulse-coupled decentralized synchronization algorithm is proposed to improve the precision of the synchronization. The benefit of this kind of algorithm is that sensors only exchange zero-bit pulse instead of packets, so not only the communication is efficient but also robust to any failure of the sensors in the network. Localization of sensors has been widely studied. However, the quality and the accuracy of the localization still have a large room to improve. This thesis apply Leave-out Sign-dominant Correlated Regions (LSCR) algorithm to localization problem. With LSCR, one evaluates the accurate estimates of confidence regions with prescribed confidence levels, which provide not only the location but also the confidence of the estimation. In this thesis, several localization approaches are implemented and compared. The simulation result shows under mild assumptions, LSCR obtains competitive results compared to other methods.

The Liquid Salt Cooled Reactor (LSCR) provides several potential benefits compared to pressurized water-cooled reactor systems. These include low operating pressure of the liquid salt coolant, the high burnup tolerance of the fuel, and the high operating temperatures which leads to increases in efficiency. However, due to inherently low heavy metal loading, the fuel cycle design presents specific challenges. In order to study options for optimizing the fuel design and fuel cycle, SCALE6.1 was used to create simplified models of the reactor and look at various parameters. The primary parameters of interest included packing factor and fuel enrichment. An economic analysis was performed on these results by developing a simple fuel cycle cost (FCC) model that could be used to compare the different options from an economic standpoint. The lithium enrichment of the FLiBe coolant was also investigated. The main focus was to understand the practical limitations associated with the Li-7 enrichment and whether it could be used for beneficial purposes. The main idea was to determine whether a lower-than-equilibrium enrichment could be used at reactor start up so that the Li-6 isotope acts as a burnable absorber. The results for the lithium enrichment study showed that the enrichment converges over time, but the amount of time required to reach steady state is much too long and the FLiBe coolant could not be utilized for reactivity control as a burnable absorber. The results found through this research provide reasonable guidelines for expected costs and narrow down the types of configurations that should be considered as fuel design options for the LSCR. Additionally, knowledge was gained on methods for modeling the system not only accurately but also efficiently to reduce the required computing power and time.