Le présent travail est une contribution à la K-théorie bivariante des C*-algèbres au sens de Kasparov, et en particulier à sa version équivariante. Un rôle clé dans cette théorie est joué par l'élément"gamma" de Kasparov, une sorte de classe fondamentale équivariante d'un groupe localement compact. On s'intéresse à la représenter par desK-cycles (modules de Fredholm) possédant de bonnes propriétés.Dans cette thèse on donne une nouvelle construction de tels K-cyclespour les groupes hyperboliques au sens de Gromov. Les modules de Fredholm obtenus sont finiment sommables, i.e. ils possèdent une propriété de régularité particulièrement forte. On donne aussi une majoration de leur degré minimal de sommabilité.On s'inspire des travaux de V. Lafforgue: les K-cycles considérés sontsimilaires à ceux utilisés par Lafforgue dans sa démonstration de la Conjecture de Baum-Connes à coefficients pour les groupes hyperboliques. Leur construction est basée sur les idées de Mineyev sur les "bicombings homologiques" des groupes hyperboliques et procède par récurrence sur les squelettes d'un complexe de Rips associé au groupe.Une preuve non-constructive de la sommabilité finie d'un élément "gamma"a été obtenue par Emerson et Nica pour les groupes hyperboliques decaractéristique d'Euler-Poincaré zéro. Des constructions explicites deK-cycles représentant l'élément "gamma" d'un groupe hyperbolique ont étédonnées par Kasparov-Skandalis et V. Lafforgue, mais on ne sait passi leurs modules sont finiment sommables. En général, on ne peut pasespérer trouver des éléments "gamma" finiment sommables pour d'autresclasses de groupes discrets. / This work is a contribution to the bivariant K-theory of C*-algebras in the sense of Kasparov and in particular to its equivariant version. In this theory, a key role is played by Kasparov’s “gamma”-element, a kind of equivariant fundamental equivariant class for a locally compact group. It is of interest to find particularly well behaved K-cycles (Fredholm modules) representing this class.We present a new construction of K-cycles representing a "gamma"-element for hyperbolic groups in the sens of Gromov. The Fredholm modules obtained are finitely summable i.e. they possess particularly strong regularity properties. We also obtain an upper bound of their minimal degree of summability.Our approach is inspired by the work of V. Lafforgue: the K-cycles under consideration are similar to those used by Lafforgue in his demonstration of Baum-Connes conjecture with coefficients for hyperbolic groups. Their construction is based on Mineyev’s ideas on homological bicombings and proceeds by induction over the skeleta of a Rips complex associated to the group.A non-constructive proof of the finite summablity of a “gamma” element was obtained by Emerson and Nica for the hyperbolic groups of Euler-Poincaré characteristic zero. Explicit constructions of K-cycles representing the “gamma”-element of hyperbolic groups were given by Kasparov-Skandalis and V. Lafforgue, but it is not known whether their modules are finitely summable. In general one cannot hope to find finitely summable “gamma” elements for other classes of discrete groups.

No description available.

L’objectif de ces travaux était d’étudier la faisabilité d’un module radio ultra large bande en gamme millimétrique destiné à des applications de réseaux de capteurs ultra faible consommation pour des applications domotiques, industrielles ou environnementales. Aujourd'hui, l'usage du canal 60 GHz radio est attrayante pour les applications de réseaux de capteurs car elle bénéficie d'une ressource spectrale large (7GHz entre 57GHz et 64GHz), d’une miniaturisation des modules radio, de l'interférence limitée avec les autres systèmes de communication. Après avoir comparé plusieurs technologies de couche physique nous avons opté pour un module radio en technique impulsionnelle ultra large bande transposée en gamme millimétrique autour de 60GHz, qui présente une faible consommation d'énergie, une faible complexité de l'architecture radio, une faible sensibilité à la non-linéarité de l'émetteur, une robustesse aux effets de propagation multiple et une résolution temporelle élevée pour des applications de localisation. Nous avons ensuite montré que pour des réseaux de moyenne densité et à faible débit numérique les approches S-MAC et Zigbee sont prometteuses demandant à être encore améliorées dans la mesure où il n’existe pas à ce jour de normes établies pour les liaisons radio entre nœuds de réseau de capteurs. De plus, afin de ne pas complexifier l’architecture globale du module radio et de diminuer la consommation nous avons choisi une architecture ULB transposée en gamme millimétrique avec une modulation de type OOK et une solution de démodulation basée sur la détection non cohérente de l'énergie Pour valider le principe, nous avons conçu deux circuits utilisant la technologie AsGa 0.1um pHEMT de chez OMMIC. L’un est un émetteur qui fonctionne à 30GHz et l’autre fonctionne à 60GHz. Dans la dernière partie, nous présentons la conception de l’émetteur du nœud de capteur en technologie CMOS 65nm SOI composé d’un oscillateur à 30 GHz contrôlé par un générateur d’impulsion sub nanoseconde, d’un doubleur de fréquence et d’un amplificateur dont les alimentations sont déclenchées par un autre générateur d’impulsion. La synchronisation, la détermination du seuil et la démodulation des données sont implémentées sous un FPGA afin de les valider et de déterminer la consommation d’énergie. / He objective of this work was to study an ultra wideband (UWB) millimeter-wave module with low power consumption for wireless sensor networks (WSNs). WSNs provide distributed information collection and transmission which are useful for many industrial or environmental applications. Nowadays the use of 60GHz radio channel is attractive for WSNs applications since it benefits of a wide spectral resource (7 GHz allocated between 57GHz and 64GHz), a possibility for the miniaturization of the radio modules, limited interference with other communicating systems, as well as access to worldwide allocated non regulatory frequency bands. After comparing several physical layer technologies we chosed an impulse radio millimeter-wave UWB architecture transposed to 60GHz, the advantages of which are low power consumption, low complexity architecture, low sensitivity to the nonlinearity of the transmitter, robustness to multiple propagation effects and high time resolution for localization applications. We then showed that the approaches (S-MAC and Zigbee) are promising for the MAC layer of the WSNs,but need to be further improved since currently there are no established standards. In order to simplify the overall architecture of the radio module and reduce the power consumption, we have chosen an UWB millimeter-wave architecture using OOK modulation and noncoherent demodulation based on the energy detection. To validate the principle, we have designed two circuits using 0.1um GaAs pHEMT from the OMMIC. The first one is a 30 GHz transmitter and the second one is at 60GHz. In the end a new simple 60 GHz UWB transmitter using 65nm CMOS SOI technology was presented. This transmitter is composed of a negative differential resistance (NDR) oscillator driven by an UWB pulse generator in conjunction with a frequency doubler and a medium power amplifier (MPA), the supply of which is triggered by another UWB pulse generator. The synchronization, the determination of threshold and the demodulation of data are implemented on an FPGA to validate and determine the power consumption.

The goal of this report is the development of a PSpice based modeling platform for the evaluation of power IGBT modules to be used in HVDC and FACTS applications. The use of simulation tools is of great value in the process of developing new power electronic devices and new converter topologies. By means of this proper model platform, new design ideas and better understanding of devices behavior and related physical phenomena of the modules can be easily estimated and the outcome is reducing the demand of extensive laboratory testing. Particularly important is the choice of a proper model capable of fast simulation times and adequate accuracy, while a challenging issue is to guarantee the convergence of such models given the hard nonlinearities and multiple cross-references involved. The choice of a Spice based simulation platform consisting in a circuitously based model, allows us to evaluate the characteristics of each module as a function of parameters like a dc-voltage, load current, stray inductance and gate driving.

Multi-phase Buck is widely used in Voltage Regulator Modules design because of its low cost and simplicity. But this topology also has a lot of drawbacks. One of the most fundamental drawback is that it has narrow duty cycles when it operates at high switching frequency with low output voltage (for example 1V). Narrow duty cycles yield high switching loss which limits the switching frequency of Buck; making it difficult to design a Buck based VRM that can achieve high efficiency at a high switching frequency. In this thesis three new non-isolated full bridge topologies will be introduced to solve the aforementioned problems of Buck. One is a new non-isolated full bridge topology, this new topology use a transformer to extend the duty cycle and it capable to achieve zero voltage switching. Experimental results demonstrate that it has significant advantages over multi-phase Buck. In some applications when huge output current is required, several converters are paralleled to supply the current that is not an optimal solution. Two two-phase non-isolated full bridge topologies are proposed to solve this problem. They double the output power of one-phase non-isolated full bridge, and achieve higher efficiency with fewer switches compared with parallel two non-isolated full bridge converters. Non-isolated VRM usually is used for personal computers, VRM for servers is called power pod, and usually isolation is required for power pod due to safety considerations. Server usually require much more power than personal computers, their power consumption is around several KW. To provide the power for the server a few power modules will need to be paralleled, this kind solution is expensive and make current sharing complex. In this thesis two new two-phase isolated full bridge topologies are proposed. They are capable to operate at soft switching mode. And they double the output power compared with conventional full bridge converter. Compared with parallel two full bridge converters, they can achieve higher efficiency with fewer switches. / Thesis (Ph.D, Electrical & Computer Engineering) -- Queen's University, 2008-02-28 09:53:50.23

No description available.

Yokonuma-type Hecke algebras are a class of Hecke algebras built from a Type A construction. In this thesis, I construct the irreducible representations for a class of generic Yokonuma-type Hecke algebras which specialize to group algebras of the complex reflection groups and to endomorphism rings of certain permutation characters of finite general linear groups.

A test system was commissioned to characterize commercial thermoelectric modules to be used in the Pizza Oven Waste Energy Recovery (POWER) system for Pizza Pizza restaurants. The objective of this testing was to obtain and classify the thermoelectric material parameters of the Bi2Te3 within commercial thermoelectric generator (TEG) modules. These parameters consisted of the Seebeck coefficient, the thermal conductivity and the electrical resistivity. Together they provide the normalized figure of merit for the thermoelectric material which is a performance indicator for energy efficiency at a given operating temperature. From this research, a two phase methodology was developed that was able to extract the desired values from these modules. Material quality and device composition was first assessed with tools such as SEM and EDS. During this phase, dimensional and elemental data was gathered and a finite element model was constructed to ensure the validity of the primary selected test method for this research which was the Harman technique. The results obtained with this method were all three of the aforementioned thermoelectric parameters as well as a direct measurement of the figure of merit. Thermal and electrical losses for the TEG1B-12610-5.1 module were characterized from room temperature to 200°C using this process. It was determined that the thermal losses were more dominant and could be approximated using a function of T4 to within 1% of their calculated values. This process can be applied to any model of TEG to forecast these losses. To assist with future research, a secondary test method known as the Parallel Thermal Conductance technique was researched and a proposed model of it was designed for use in temperatures up to 300°C. Due to the relatively short test time of the Harman Technique, it was also used to effectively bin incoming groups of TEGs used in the POWER system so that they could be placed strategically in different areas of heat flow based on their measured performance. An increase of 13.2% was observed in the electrical output of the system after the binning had occurred. / Thesis / Master of Applied Science (MASc)

<p> A ring R is said to be QF-1 if every finitely generated faithful R-module has the double centralizer property (or is balanced). A necessary and sufficient condition for an artinian ring to be QF-1 is given. The class of QF-1 rings properly contains the class of QF rings and this is shown by an example. Several constructions of modules which are not balanced are collected. Finally, the structure of artinian local QF-1 rings which are finitely generated over their centers is gotten. This is a generalization of theorems of Floyd, and, Fuller and Dickson.</p> / Thesis / Master of Science (MSc)

No description available.