Nowadays, people expectations for ubiquitous connectivity is continuously growing. Cities are now moving towards the smart city paradigm. Electricity companies aims to become part of smart grids. Internet is no longer exclusive for humans, we now assume the Internet of everything. We consider that Wireless Mesh Networks (WMNs) have a set of valuable features that will make it an important part of such environments. WMNs can also be use in less favored areas thanks to their low-cost deployment. This is socially relevant since it facilitates the digital divide reduction and could help to improve the population quality of life. Research and industry have been working these years in open or proprietary mesh solutions. Standardization efforts and real deployments establish a solid starting point.We expect that WMNs will be a supporting part for an unlimited number of new applications from a variety of fields: community networking, intelligent transportation systems, health systems, public safety, disaster management, advanced metering, etc. For all these cases, the growing needs of users for real-time and multimedia information is currently evident. On this basis, this thesis proposes a set of contributions to improve the performance of an application service of such type and to promote the better use of two critical resources (memory and energy) of WMNs.For the offered service, this work focuses on a Video on Demand (VoD) system. One of the requirements of this system is the high capacity support. This is mainly achieved by distributing the video contents among various distribution points which in turn consist of several video servers. Each client request that arrives to such video server cluster must be handled by a specific server in a way that the load is balanced. For such task, this thesis proposes a mechanism to appropriately select a specific video server such that the transfer time at the cluster could be minimized.On the other hand, mesh routers that creates the mesh backbone are equipped with multiple interfaces from different technologies and channel types. An important resource is the amount of memory intended for buffers. The quality of service perceived by the users are largely affected by the size of such buffers. This is because important network performance parameters such as packet loss probability, delay, and channel utilization are highly affected by the buffer sizes. An efficient use of memory for buffering, in addition to facilitate the mesh devices scalability, also prevents the problems associated with excessively large buffers. Most of the current works associate the buffer sizing problem with the dynamics of TCP congestion control mechanism. Since this work focuses on real time services, in which the use of TCP is unfeasible, this thesis proposes a dynamic buffer sizing mechanism mainly dedicated for such real time flows. The approach is based on the maximum entropy principle and allows that each device be able to dynamically self-configure its buffers to achieve more efficient memory utilization. The proper performance of the proposal has been extensively evaluated in wired and wireless interfaces. Classical infrastructure-based wireless and multi-hop mesh interfaces have been considered. Finally, when the WMN is built by the interconnection of user hand-helds, energy is a limited and scarce resource, and therefore any approach to optimize its use is valuable. For this case, this thesis proposes a topology control mechanism based on centrality metrics. The main idea is that, instead of having all the devices executing routing functionalities, just a subset of nodes are selected for this task. We evaluate different centralities, form both centralized and distributed perspectives. In addition to the common random mobility models we include the analysis of the proposal with a socially-aware mobility model that generates networks with a community structure. / Actualmente las expectativas de las personas de una conectividad ubicua están creciendo. Las ciudades están trabajando para alcanzar el paradigma de ciudades inteligentes. Internet ha dejado de ser exclusivo de las personas y ahora se asume el Internet de todo. Las redes inalámbricas de malla (WMNs) poseen un valioso conjunto de características que las harán parte importante de tales entornos. Las WMNs pueden utilizarse en zonas menos favorecidas debido a su despliegue económico. Esto es socialmente relevante ya que facilita la reducción de la brecha digital y puede ayudar a mejorar la calidad de vida de la población. Los esfuerzos de estandarización y los despliegues de redes reales establecen un punto de partida sólido.Se espera entonces, que las WMNs den soporte a un número importante de nuevas aplicaciones y servicios, de una variedad de campos: redes comunitarias, sistemas de transporte inteligente, sistemas de salud y seguridad, operaciones de rescate y de emergencia, etc. En todos estos casos, es evidente la necesidad de disponer de información multimedia y en tiempo real. En base a estos precedentes, esta tesis propone un conjunto de contribuciones para mejorar el funcionamiento de un servicio de este tipo y promover un uso eficiente de dos recursos críticos (memoria y energía) de las WMNs.Para el servicio ofrecido, este trabajo se centra en un sistema de video bajo demanda. Uno de los requisitos de estos sistemas es el de soportar capacidades elevadas. Esto se consigue principalmente distribuyendo los contenidos de video entre diferentes puntos de distribución, los cuales a su vez están formados por varios servidores. Cada solicitud de un cliente que llega a dicho conjunto de servidores debe ser manejada por un servidor específico, de tal forma que la carga sea balanceada. Para esta tarea, esta tesis propone un mecanismo que selecciona apropiadamente un servidor de tal manera que el tiempo de transferencia del sistema sea minimizado.Por su parte, los enrutadores de malla que crean la red troncal están equipados con múltiples interfaces de diferentes tecnologías y tipos de canal. Un recurso muy importante para éstos es la memoria destinada a sus colas. La calidad de servicio percibida por los usuarios está altamente influenciada por el tamaño de las colas. Esto porque parámetros importantes del rendimiento de la red como la probabilidad de pérdida de paquetes, el retardo, y la utilización del canal se ven afectados por dicho tamaño. Un uso eficiente de tal memoria, a más de facilitar la escalabilidad de los equipos, también evita los problemas asociados a colas muy largas. La mayoría de los trabajos actuales asocian el problema de dimensionamiento de las colas con la dinámica del mecanismo de control de congestión de TCP. Debido a que este trabajo se enfoca en servicios en tiempo real, en los cuales no es factible usar TCP, esta tesis propone un mecanismo de dimensionamiento dinámico de colas dedicado principalmente a flujos en tiempo real. La propuesta está basada en el principio de máxima entropía y permite que los dispositivos sean capaces de auto-configurar sus colas y así lograr un uso más eficiente de la memoria. Finalmente, cuando la WMN se construye a través de la interconexión de los dispositivos portátiles, la energía es un recurso limitado y escaso, y cualquier propuesta para optimizar su uso es muy valorada. Para esto, esta tesis propone un mecanismo de control de topología basado en métricas de centralidad. La idea principal es que en lugar de que todos los dispositivos realicen funciones de enrutamiento, solo un subconjunto de nodos es seleccionado para esta tarea. Se evalúan diferentes métricas, desde una perspectiva centralizada y otra distribuida. A más de los modelos aleatorios clásicos de movilidad, se incluye el análisis de la propuesta con modelos de movilidad basados en información social que toman en cuenta el comportamiento humano y generan redes con una clara estructura de comunidades

This thesis is focused on the conducted electromagnetic Interference generated at Integrated Circuit (IC) Level. Recently, several electrical models have been proposed in order to describe EMI at IC level, but they have several limitations. The first drawback is that these electrical models do not take into account the wear-out mechanisms on the EMI behaviour. The wear-out mechanisms are due to the dielectric degradation when an electric stress is applied to the oxide gate. Due to this degradation, the MOSFET characteristics are shifted. Therefore, the evaluation of wear-out mechanisms allow the designers, during the IC design, to predict the circuit behaviour along its lifetime. However, the impact of these wear-out mechanisms on the conducted EMI drift at IC level has not been deeply investigated. Hence, one of the aims of this thesis will be focused on the impact of wear-out mechanisms in signal integrity and conducted EMI at IC level. Moreover, current integrated circuits have a high operation frequency. Thus, the electromagnetic noise induced on those devices presents a higher harmonic content. For this reason, the electronics industry requires electrical models to predict high frequency conducted emissions. In this sense, the other aim of this thesis will be focused on expanding the current EMI models beyond 1 GHz. The IC behaviour may be affected by temperature, as well as conducted emission levels. Therefore, the proposed electrical model will take into account the impact of temperature. The experimental results have been obtained with three integrated circuits, two of them are specific test chip designed by Freescale Semiconductor, Inc., and the third IC is a commercial circuit of Maxim Integrated Circuits. This document is structured in four chapters. Chapter 1 describes the main wear-out mechanisms and the electromagnetic compatibility at IC level. The different EMI produced at IC are explained. Also, it describes aging methods to characterize the impact of wear-out mechanisms on MOS devices. Furthermore, the EMI characterization methods are explained and different EMC electrical models are described. To confirm the accuracy of the EMC models, the ¿Feature Selective Validation¿ (FSV) technique has been used. On this chapter, the FSV method and its application on computational electromagnetism is detailed. The chapter ends with the state of the art on wear-out mechanisms and EMI at IC level. Chapter 2 analyzes the IC reliability. The IC aging of the MOSFET I-V curve characteristics is studied, for further EMI characterization of the impact of wear-out mechanisms. The experimental results are presented at the end of Chapter 2. Chapter 3 presents an electrical model to characterize the conducted emissions of an IC up to 3 GHz. This electrical model considers the impact of temperature. The proposed model is validated with experimental results and verified with the FSV method. Chapter 4 summarizes the conclusions of the thesis and the main contributions. In addition, a list of the publications derived from this thesis is included. Finally, the chapter presents the lines for future research. / Esta tesis se centra en el estudio de las interferencias electromagnéticas (“Electromagnetic Interferences” o EMI) conducidas generadas a nivel de circuito integrado (CI). En la actualidad, existen modelos eléctricos para describir las EMI conducidas a nivel de CI, pero presentan ciertas limitaciones. La primera de ellas es que estos modelos no tienen en cuenta el impacto de los mecanismos de degradación sobre las EMI. Los mecanismos de degradación aparecen por el deterioro del dieléctrico debido al estrés eléctrico aplicado en el óxido de puerta. Estos mecanismos producen la variación de las características eléctricas de los dispositivos MOS. El estudio de estos efectos permite predecir, durante la etapa inicial del diseño, su impacto durante el tiempo de vida de los CI. Sin embargo, hasta la fecha, no se han llevado a cabo estudios del efecto de los mecanismos de degradación en las EMI conducidas a nivel de CI. Por lo tanto, uno de los primeros objetivos de la tesis será caracterizar el impacto de los mecanismos de degradación en la integridad de la señal y en las EMI conducidas a nivel de CI. Asimismo, los CI tienen una frecuencia de funcionamiento cada vez mayor, de modo que el ruido electromagnético generado por estos dispositivos tiene un contenido harmónico de más alta frecuencia. Es por esto que conviene tener modelos eléctricos que permitan modelizar las EMI de alta frecuencia. El segundo objetivo de la tesis consiste en modelizar las EMI conducidas más allá de la frecuencia de 1 GHz ya que los modelos actuales son válidos hasta esta frecuencia. La temperatura de funcionamiento del CI puede afectar al comportamiento del mismo, así como a los niveles de las emisiones conducidas. Por lo tanto será de interés que el modelo propuesto tenga en cuenta el impacto de la temperatura, ya que los modelos actuales únicamente son válidos para una temperatura de funcionamiento. La validación experimental se ha llevado a cabo sobre tres circuitos integrados, dos de ellos diseñados específicamente para este estudio por la empresa Freescale Semiconductor, Inc. y el tercer CI es un circuito comercial de Maxim Integrated Circuits. Este documento se compone de cuatro capítulos. El capítulo 1 empieza con la descripción de los principales mecanismos de degradación y de la compatibilidad electromagnética a nivel de circuito integrado. Se detallan las diferentes interferencias electromagnéticas que pueden producirse a nivel de circuito integrado. Se procede con la descripción de los métodos acelerados de envejecimiento para caracterizar el impacto de los mecanismos de degradación en los dispositivos MOS. Se continúa con una explicación de los métodos para caracterizar las EMI y la presentación de diferentes modelos EMC para su modelización. Para la validación de los estos modelos EMC se hace uso del método “Feature Selective Validation” (FSV). En este capítulo se da explicación al método FSV y su aplicación en el electromagnetismo computacional. Para finalizar el capítulo, se describe el estado actual de la investigación en el campo de los mecanismos de degradación y de las EMI a nivel de CI. En el capítulo 2 se analiza la fiabilidad de los CI. Se estudia el impacto de los mecanismos de degradación en el comportamiento de los transistores, para posteriormente estudiar el impacto de estos mecanismos en las EMI. El capítulo 2 se complementa con los resultados experimentales obtenidos en el laboratorio. El capítulo 3 se centra en la caracterización y el modelado de las EMI en los circuitos integrados. Se propone un modelo eléctrico para caracterizar las interferencias electromagnéticas conducidas hasta los 3 GHz y el impacto de la temperatura en las emisiones conducidas. El modelo propuesto es comprobado con medidas experimentales y verificado con el método FSV. Por último, el capítulo 4 resume las conclusiones de la tesis y las principales contribuciones. Además, en este capítulo se presenta las líneas de investigación futuras. Esta tesis se ha desarrollado dentro de una de las líneas de investigación del Grupo de Electrónica Industrial de Terrassa (“Terrassa Industrial Electronics Group” - TIEG), dentro del marco del proyecto de investigación TEC2009-09994, TEC2010-18550 y AGAUR 2009 SGR 142.

This thesis presents a body of work on the modeling of and performance predictions for carbon nanotube field-effect transistors (CNFET) and graphene field-effect transistors (GFET). While conventional silicon-based CMOS is expected to reach its ultimate scaling limits during the next decade, these two novel technologies are promising candidates for future high-performance electronics. The main goal of this work is to investigate on the opportunities of using such carbon-based electronics for RF integrated circuits. This thesis addresses 1) the modeling of noise and process variability in CNFETs, 2) RF performance predictions for CNFETs, and 3) an accurate GFET compact model. This work proposes the first CNFET noise compact model. Noise is of primary importance for RF applications and its description significantly increases the insight gained from simulation studies. Furthermore, a CNFET variability model is presented, which handles tube synthesis and metal tube removal imperfections. These two model extensions have been added to the Stanford CNFET compact model and allow for the variability-aware RF performance assessment of the CNFET technology. In continuation, comprehensive RF performance projections for CNFETs are provided both on the device and circuit level. The overall set of ITRS RF-CMOS technology requirement FoMs is determined and shows that the CNFET performs excellently in terms of speed, gain, and minimum noise figure. Furthermore, for the first time FoMs are reported for the basic RF building blocks low-noise amplifier and oscillator. In addition, it is shown that CNFET downscaling yields significant performance improvements. Based on these analyses it is confirmed that the CNFET has the potential to outperform Si-CMOS in RF applications. A third key contribution of this thesis is the development of an accurate GFET compact model. Previous compact models simplify several physical aspects, which can cause erroneous simulation results. Here, an accurate yet simple mathematical description of the GFET’s current-voltage relation is proposed and implemented in Verilog-A. Comprehensive error analyses are done in order to highlight the advantages of the new approach. Furthermore, the model is verified against measurement results. The developed GFET model is an important step towards better understanding the characteristics and opportunities of graphene-based analog circuitry.

This thesis deals with the processing of different types of multidimensional SAR data for distinct applications. Instead of handling the original pixels of the image, which correspond to very local information and are strongly contaminated by speckle noise, a region-based and multiscale data abstraction is defined, the Binary Partition Tree (BPT). In this representation, each region stands for an homogeneous area of the data, grouping pixels with similar properties and making easier its interpretation and processing. The work presented in this thesis concerns the definition of the BPT structures for Polarimetric SAR (PolSAR) images and also for temporal series of SAR acquisitions. It covers the description of the corresponding data models and the algorithms for BPT construction and its exploitation. Particular attention has been paid to the speckle filtering application. The proposed technique has proven to achieve arbitrarily large regions over homogeneous areas while also preserving the spatial resolution and the small details of the original data. As a consequence, this approach has demonstrated an improvement in the performance of the target response estimation with respect to other speckle filtering techniques. Moreover, due to the flexibility and convenience of this representation, it has been employed for other applications as scene segmentation and classification. The processing of SAR time series has also been addressed, proposing different approaches for dealing with the temporal information of the data, resulting into distinct BPT abstractions. These representations have allowed the development of speckle filtering techniques in the spatial and temporal domains and also the improvement and the definition of additional methods for classification and temporal change detection and characterization.

Over the last few decades, transistor miniaturization has enabled a tremendous increase in the processing capability of commercial electronic devices, which, combined with the reduction of production costs, has tremendously fostered the usage of the Information and communications Technologies (ICTs) both in terms of number of users and required data rates. In turn, this has led to a tremendous increment in the energetic demand of the ICT sector, which is expected to further grow during the upcoming years, reaching unsustainable levels of greenhouse gas emissions as reported by the European Council. Additionally, the autonomy of battery operated devices is getting reduced year after year since battery technology has not evolved fast enough to cope with the increase of energy consumption associated to the growth of the node¿s processing capability. Energy harvesting, which is known as the process of collecting energy from the environment by different means (e.g., solar cells, piezoelectric generators, etc.), has become a potential technology to palliate both of these problems. However, when energy harvesting modules are placed in wireless communication devices (e.g., sensor nodes or hand-held devices), traditional transmission strategies are no longer applicable because the temporal variations of the node¿s energy availability must be carefully accounted for in the design. Apart from not considering energy harvesting, traditional transmission strategies assume that the transmission radiated power is the unique energy sink in the node. This is a reasonable assumption when the transmission range is large, but it no longer holds for low consumption devices such as sensor nodes that transmit to short distances. As a result, classical transmission strategies become suboptimal in short-range communications with low consumption devices and new strategies should be investigated. Consequently, in this dissertation we investigate and design transmission strategies for Wireless Energy Harvesting Nodes (WEHNs) by paying a special emphasis on the different sinks of energy consumption at the transmitter(s). First, we consider a finite battery WEHN operating in a point-to-point link through a static channel and derive the transmission strategy that minimizes the transmission completion time of a set of data packets that become available dynamically over time. The transmission strategy has to satisfy causality constrains in data transmission and energy consumption, which impose that the node cannot transmit data that is not yet available nor consume energy that has not yet been harvested. Second, we consider a WEHN that has an infinite backlog of data to be transmitted through a point-to-point link in a time-varying linear vector Gaussian channel and study the linear precoding strategy that maximizes the mutual information given an arbitrary distribution of the input symbols while satisfying the Energy Causality Constraints (ECCs) at the transmitter. Next, apart from the transmission radiated power, we take into account additional energy sinks in the power consumption model and analyze how these energy sinks affect to the transmission strategy that maximizes the mutual information achieved by a WEHN operating in a point-to-point link. Finally, we consider multiple transmitter and receiver pairs sharing a common channel and investigate a distributed power allocation strategy that aims at maximizing the network sum-rate by taking into account the energy availability in the different transmitters and a generalized power consumption model. / Durant les últimes dècades, la miniaturització del transistor i la reducció dels seus costos de fabricació han provocat un augment substancial del nombre de terminals de comunicacions i del tràfic de dades requerit per aquests dispositius. Així doncs, el consum energètic del sector de les Tecnologies de la Informació i Comunicacions ha incrementat notablement. A més a més, s’espera que aquest consum segueixi creixent durant els propers anys arribant a nivells insostenibles d’emissions de gasos d’efecte hivernacle segons ha informat el Consell Europeu. D’altra banda, la tecnologia de les bateries no ha evolucionat suficientment ràpid com per fer front a l’augment del consum energètic associat al creixement de la capacitat de processament dels dispositius. Això ha ocasionat que l’autonomia dels dispositius que operen amb bateries empitjori any rere any. Les energies renovables (per exemple, energia solar, cinètica, etc.) s’han convertit en una solució potencial per pal•liar aquests dos problemes. No obstant això, quan els dispositius de comunicació sense fils incorporen mòduls de captació d’energies renovables, les estratègies tradicionals de transmissió deixen de ser vàlides, ja que les variacions temporals de la disponibilitat d’energia en el dispositiu han de ser considerades en el disseny. A més a més, les estratègies de transmissió tradicionals assumeixen que la potència radiada és l’única font de consum energètic del node. Aquesta és una suposició raonable per distàncies de transmissió llargues, però deixa de ser vàlida quan es consideren dispositius de baix consum que transmeten en distàncies curtes. Com a resultat, les estratègies de transmissió clàssiques són subòptimes en comunicacions de curt abast amb dispositius de baix consum i per això, s’han d’investigar noves estratègies. En conseqüència, en aquesta tesi doctoral s’investiguen i es dissenyen noves estratègies de transmissió per nodes sense fils que operen amb energies renovables (WEHN) posant un èmfasi especial en les diferents fonts de consum d’energia en el transmissor. En primer lloc, la tesi investiga l’estratègia de transmissió en un enllaç¸ punt a punt a través d’un canal estàtic que minimitza el temps de transmissió d’un conjunt de paquets de dades que s’adquireixen al llarg del temps. L’estratègia de transmissió ha de satisfer les limitacions per causalitat en la transmissió de dades i en el consum d’energia les quals imposen que el node no pot transmetre dades que no han estat encara obtingudes o utilitzar energia que encara no ha estat adquirida. En segon lloc, es considera un WEHN que sempre disposa de dades per a transmetre a través d’un enllaç¸ punt a punt en un canal lineal Gaussià amb variacions temporals. En aquest escenari i, també, donada una distribució arbitrària dels símbols d’entrada, s’estudia l’estratègia de precodificació lineal que maximitza la informació mútua alhora que satisfà la causalitat d’energia en el transmissor. A continuació, a part de la potència radiada en transmissió, s’inclouen en el model de consum energètic els costos d’activació per accés al canal i per portadora. Donat aquest model, s’analitza com aquestes fonts de consum addicionals afecten a l’estratègia de transmissió que maximitza la informació mútua d’un WEHN que opera en un enllaç punt a punt. Finalment, la tesi considera diversos parells transmissor i receptor que comparteixen un canal comú i investiga una estratègia d’assignació de potència distribuïda la qual té com a objectiu maximitzar la suma de les taxes de transmissió dels diferents nodes tenint en compte la disponibilitat energètica en cada transmissor que està basada en un model de consum energètic generalitzat.

Tesi per compendi de publicacions / From the perspective of increasingly data rate requirements in mobile communications, it is deemed necessary to do further research so that the future goals can be reached. To that end, the radio-based communications are resorting to multicarrier modulations and spatial diversity. Until today, the orthogonal frequency division multiplexing (OFDM) modulation is regarded as the dominant technology. On one hand, the OFDM modulation is able to accommodate multiantenna configurations in a very straightforward manner. On the other hand, the poor stopband attenuation exhibited by the OFDM modulation, highlights that a definitely tight synchronization is required. In addition, the cyclic prefix (CP) has to be sufficiently long to avoid inter-block interference, which may substantially reduce the spectral efficiency. In order to overcome the OFDM drawbacks, the filter bank multicarrier modulation based on OQAM (FBMC/OQAM) is introduced. This modulation does not need any CP and benefits from pulse shaping techniques. This aspect becomes crucial in cognitive radio networks and communication systems where nodes are unlikely to be synchronized. In principle, the poor frequency confinement exhibited by OFDM should tip the balance towards FBMC/OQAM. However, the perfect reconstruction property of FBMC/OQAM systems does not hold in presence of multipath fading. This means that the FBMC/OQAM modulation is affected by inter-symbol and inter-carrier interference, unless the channel is equalized to some extent. This observation highlights that the FBMC/OQAM extension to MIMO architectures becomes a big challenge due to the need to cope with both modulation- and multiantenna-induced interference. The goal of this thesis is to study how the FBMC/OQAM modulation scheme can benefit from the degrees of freedom provided by the spatial dimension. In this regard, the first attempt to put the research on track is based on designing signal processing techniques at reception. In this case the emphasis is on single-input-multiple-output (SIMO) architectures. Next, the possibility of pre-equalizing the channel at transmission is investigated. It is considered that multiple antennas are placed at the transmit side giving rise to a multiple-input-single-output (MISO) configuration. In this scenario, the research is not only focused on counteracting the channel but also on distributing the power among subcarriers. Finally, the joint transmitter and receiver design in multiple-input-multiple-output (MIMO) communication systems is covered. From the theory developed in this thesis, it is possible to conclude that the techniques originally devised in the OFDM context can be easily adapted to FBMC/OQAM systems if the channel frequency response is flat within the subchannels. However, metrics such as the peak to average power ratio or the sensitivity to the carrier frequency offset constraint the number of subcarriers, so that the frequency selectivity may be appreciable at the subcarrier level. Then, the flat fading assumption is not satisfied and the specificities of FBMC/OQAM systems have to be considered. In this situation, the proposed techniques allow FBMC/OQAM to remain competitive with OFDM. In addition, for some multiantenna configurations and propagation conditions FBMC/OQAM turns out to be the best choice. The simulation-based results together with the theoretical analysis conducted in this thesis contribute to make progress towards the application of FBMC/OQAM to MIMO channels. The signal processing techniques that are described in this dissertation allow designers to exploit the potentials of FBMC/OQAM and MIMO to improve the link reliability as well as the spectral efficiency.

Tesi per compendi de publicacions. DOI els articles publicats per IEEE: 10.1109/TSG.2012.2206060, 10.1109/TSP.2013.2248002, 10.1109/TSP.2014.2307835 / The modern power grid is facing major challenges in the transition to a low-carbon energy sector. The growing energy demand and environmental concerns require carefully revisiting how electricity is generated, transmitted, and consumed, with an eye to the integration of renewable energy sources. The envisioned smart grid is expected to address such issues by introducing advanced information, control, and communication technologies into the energy infrastructure. In this context, demand-side management (DSM) makes the end users responsible for improving the efficiency, reliability and sustainability of the power system: this opens up unprecedented possibilities for optimizing the energy usage and cost at different levels of the network. The design of DSM techniques has been extensively discussed in the literature in the last decade, although the performance of these methods has been scarcely investigated from the analytical point of view. In this thesis, we consider the demand-side of the electrical network as a multiuser system composed of coupled active consumers with DSM capabilities and we propose a general framework for analyzing and solving demand-side management problems. Since centralized solution methods are too demanding in most practical applications due to their inherent computational complexity and communication overhead, we focus on developing efficient distributed algorithms, with particular emphasis on crucial issues such as convergence speed, information exchange, scalability, and privacy. In this respect, we provide a rigorous theoretical analysis of the conditions ensuring the existence of optimal solutions and the convergence of the proposed algorithms. Among the plethora of DSM methods, energy consumption scheduling (ECS) programs allow to modify the user's demand profile by rescheduling flexible loads to off-peak hours. On the other hand, incorporating dispatchable distributed generation (DG) and distributed storage (DS) into the demand-side of the network has been shown to be equally successful in diminishing the peak-to-average ratio of the demand curve, plus overcoming the limitations in terms users' inconvenience introduced by ECS. Quite surprisingly, while the literature has mostly concentrated on ECS techniques, DSM approaches based on dispatchable DG and DS have not attracted the deserved attention despite their load-shaping potential and their capacity to facilitate the integration of renewable sources. In this dissertation, we fill this gap and devise accurate DSM models to study the impact of dispatchable DG and DS at the level of the end users and on the whole electricity infrastructure. With this objective in mind, we tackle several DSM scenarios, starting from a deterministic day-ahead optimization with local constraints and culminating with a stochastic day-ahead optimization combined with real-time adjustments under both local and global requirements. Each task is complemented by defining appropriate network and pricing models that enable the implementation of the DSM paradigm in realistic energy market environments. In this regard, we design both user-oriented and holistic-based DSM optimization frameworks, which are respectively applicable to competitive and externally regulated market scenarios. Numerical results are reported to corroborate the presented distributed schemes. On the one hand, the users' electricity expenditures are consistently reduced, which encourages their active and voluntary participation in the proposed DSM programs; on the other hand, this results in a lower generation costs and enhances the robustness of the whole grid. / La xarxa elèctrica moderna s'enfronta a enormes reptes en la transició cap a un sector energètic de baixa generació de carboni. La creixent demanda d'energia i les preocupacions ambientals requereixen revisar acuradament com es genera, transmet, i consumeix l'electricitat, amb l'objectiu de la integració de les fonts d'energia renovables. S'espera que el concepte de smart grid pugui abordar aquestes qüestions mitjançant la introducció d’informació avançada, control i tecnologies de la comunicació en la infraestructura energètica. En aquest context, el concepte de gestió de la demanda (DSM) fa que els usuaris finals siguin responsables de millorar l’eficiència, la fiabilitat i la sostenibilitat del sistema de potència obrint possibilitats sense precedents per a l’optimització de l’ús i el cost de l'energia en els diferents nivells de la xarxa. El disseny de tècniques de DSM s'ha debatut àmpliament en la literatura durant l’ultima dècada, tot i que el rendiment d'aquests mètodes ha estat poc investigat des del punt de vista analític. En aquesta tesi es considera la demanda de la xarxa elèctrica com un sistema multiusuari format per consumidors actius amb capacitats de DSM i es proposa un marc general per analitzar i resoldre problemes de gestió. Donat que els mètodes de solució centralitzats són excessivament exigents per a aplicacions pràctiques per la seva complexitat computacional i al inherent sobrecost de comunicació, ens centrem en el desenvolupament d'algorismes distribuïts, amb especial èmfasi en temes crucials com la velocitat de convergència, l'intercanvi d’informació, l'escalabilitat i la privacitat. En aquest sentit, oferim un rigorós anàlisi teòric de les condicions que garanteixen l’existència de solucions òptimes i la convergència dels algorismes proposats. Entre la gran quantitat de mètodes de DSM, els programes de programació del consum d'energia (ECS) permeten modificar el perfil de la demanda dels usuaris a través de la reprogramació de càrregues flexibles durant hores de baix consum. D'altra banda, la incorporació de generació distribuïda (DG) i d'emmagatzematge distribuït (DS) ha demostrat ser igualment eficaç disminuint la relació entre potència de pic i mitja de la corba de demanda, evitant els inconvenients introduïts pel ECS als usuaris. Sorprenentment, si bé que la literatura s'ha concentrat sobretot en les tècniques de ECS, les tècniques de DSM basades en DG i DS no han atret l’atenció merescuda malgrat el seu potencial de confirmació de la càrrega i la seva capacitat de facilitar la integració de les fonts renovables. En aquesta tesi, omplim aquest buit i elaborem models precisos de DSM per estudiar l'impacte de DG i DS a nivell dels usuaris finals i de tota la infraestructura elèctrica . Tenint present aquest objectiu, fem front a diversos escenaris de DSM, partint d'una optimització sobre les previsions amb un dia d’antelació (day-ahead). Es considera des del cas determinista amb restriccions locals fins al cas estocàstic combinat amb ajustos en temps real i amb restriccions locals i globals. Cada tasca es complementa amb la definició de models de xarxa i de tarifació apropiats que permetin la posada en pràctica del paradigma de DSM en entorns realistes del mercat energètic. En aquest sentit vam dissenyar marcs d’optimització de DSM globals i orientats als usuaris, que són respectivament aplicables a situacions de mercat competitives i regulades externament. Els resultats numèrics reportats corroboren els esquemes distribuïts presentats. D'una banda, les despeses d'electricitat dels usuaris es redueixen de forma consistent, el que fomenta la seva participació activa en els programes de DSM proposats; per una altra banda, aquesta optimització resulta en un cost de generació inferior i millora la robustesa de tota la xarxa. / La red eléctrica moderna se enfrenta a enormes retos en la transición hacia un sector energético de baja generación de carbono. La creciente demanda de energía y las preocupaciones ambientales requieren revisar cuidadosamente cómo se genera, transmite y consume la electricidad, con vista a la integración de las fuentes de energía renovable. Se espera que el concepto de smart grid pueda abordar estas cuestiones mediante la introducción de información avanzada, control y tecnologías de la comunicación en la infraestructura energética. En este contexto, el concepto de gestión de la demanda (DSM) hace que los usuarios finales sean responsables de mejorar la eficiencia, la fiabilidad y la sostenibilidad del sistema de potencia abriéndose posibilidades sin precedentes para la optimización del uso y el coste de la energía en los diferentes niveles de la red. El diseño de técnicas de DSM se ha debatido ampliamente en la literatura en la última década, aunque el rendimiento de estos métodos ha sido poco investigado desde el punto de vista analítico. En esta tesis se considera la demanda de la red eléctrica como un sistema multiusuario compuesto por consumidores activos con capacidades de DSM y se propone un marco general para analizar y resolver problemas de gestión de demanda. Dado que los métodos de solución centralizados son excesivamente exigentes para aplicaciones prácticas debido a su complejidad computacional y al inherente sobrecoste de comunicación, nos centramos en el desarrollo de algoritmos distribuidos, con especial énfasis en temas cruciales como la velocidad de convergencia, el intercambio de información, la escalabilidad y la privacidad. En este sentido, ofrecemos un riguroso análisis teórico de las condiciones que garantizan la existencia de soluciones óptimas y la convergencia de los algoritmos propuestos. Entre la gran cantidad de métodos de DSM, los programas de programación del consumo de energía (ECS) permiten modificar el perfil de la demanda de los usuarios a través de la reprogramación de cargas flexibles durante horas de bajo consumo. Por otro lado, la incorporación de generación distribuida (DG) y de almacenamiento distribuido (DS) ha demostrado ser igualmente eficaz disminuyendo la relación entre potencia de pico y media de la curva de demanda, evitando los inconvenientes introducidos por el ECS a los usuarios. Sorprendentemente, mientras que la literatura se ha concentrado sobre todo en las técnicas de ECS, los programas de DSM basados en DG y DS no han atraído la atención merecida a pesar de su potencial de conformación de la carga y su capacidad de facilitar la integración de las fuentes renovables. En esta tesis, llenamos este vacío y elaboramos modelos precisos de DSM para estudiar el impacto de DG y DS a nivel de los usuarios finales y de toda la infraestructura eléctrica. Teniendo presente este objetivo, hacemos frente a varios escenarios de DSM, a partir de una optimización sobre las previsiones con un día de antelación (day-ahead). Se considera desde el caso determinista con restricciones locales hasta el caso estocástico combinado con ajustes en tiempo real y con restricciones locales y globales. Cada tarea se complementa con la definición de modelos de red y de tarificación apropiados que permitan la puesta en práctica del paradigma de DSM en entornos realistas del mercado energético. En este sentido diseñamos marcos de optimización de DSM globales y orientados a los usuarios, que son respectivamente aplicables a situaciones de mercado competitivas y reguladas externamente. Los resultados numéricos reportados corroboran los esquemas distribuidos presentados. Por un lado, los gastos de electricidad de los usuarios se reducen de forma consistente, lo que fomenta su participación activa en los programas de DSM propuestos; por otra parte, esta optimización resulta en un coste de generación inferior y mejora la robustez de toda la red

The present thesis studies information rates for reliable transmission of information over fading channels in the realistic situation where the receiver has only imperfect channel-state knowledge. Of particular interest are analytical expressions of achievable transmission rates under imperfect and no CSI, that is, lower bounds on the mutual information and on the Shannon capacity. A well-known mutual information lower bound for Gaussian codebooks is obtained when conflating the additive (thermal) noise with the multiplicative noise due to the imperfections of the CSIR into a single effective noise term, and then assuming that this term is independent Gaussian. This so-called worst-case-noise approach allows to derive a strikingly simple and well-known lower bound on the mutual information of the channel. A first part of this thesis proposes a simple way to improve this worst-case-noise bound by means of a rate-splitting approach: by expressing the Gaussian input as a sum of several independent Gaussian inputs, and by assuming that the receiver performs successive decoding of the corresponding information streams, we show how to derive a larger mutual information lower bound. On channels with a single transmit antenna, the optimal allocation of transmit power across the different inputs is found to be approached as the number of inputs (so-called layers) tends to infinity, and the power assigned to each layer tends to zero. This infinite-layering limit gives rise to a mutual information bound expressible as an integral. On channels with multiple transmit antennas, an analogous result is derived. However, since multiple transmit antennas open up more possibilities for spatial multiplexing, the rate-splitting approach gives rise to a whole family of infinite-layering bounds. This family of bounds is closely studied for independent and identically zero-mean Gaussian distributed fading coefficients (so-called i.i.d. Rayleigh fading). Most notably, it is shown that for asymptotically perfect CSIR, any bound from the family is asymptotically tight at high signal-to-noise ratios (SNR). Specifically, this means that the difference between the mutual information and its lower bound tends to zero as the SNR tends to infinity, provided that the CSIR tends to be exact as the SNR tends to infinity. A second part of this thesis proposes a framework for the optimization of a class of utility functions in black-Rayleigh fading multiple-antenna channels with transmit-side antenna correlation, and no CSI at the receiver. A fraction of each fading block is reserved for transmitting a sequence of training symbols, while the remaining time instants are used for transmission of data. The receiver estimates the channel matrix based on the noisy training observation and then decodes the data signal using this channel estimate. For utilities that are symmetric functions of the eigenvalues of the matrix-valued effective SNR (such as, e.g., the worst-case-noise bound), the problems consisting in optimizing the pilot sequence and the linear precoder are cast into convex (or quasi-convex) problems for concave (or quasi-concave) utility functions. We also study an important subproblem of the joint optimization, which consists in computing jointly Pareto-optimal pilot sequences and precoders. By wrapping these optimization procedures into a cyclic iteration, we obtain an algorithm which converges to a local joint optimum for any utility. / Aquesta tesi estudia les taxes d'informació per la transmissió fiable d'informació en canals amb esvaïments sota la hipòtesi realista de que el receptor té un coneixement tan sols imperfecte de l'esvaïment aleatori. De particular interès són les expressions analítiques de les taxes de transmissió assolibles amb coneixement imperfecte i sense coneixement de l'estat del canal, és a dir, cotes inferiors de la informació mútua i de la capacitat de Shannon. Una cota inferior de la informació mútua per a codis gaussians ben coneguda s'obté combinant el soroll additiu (tèrmic) amb el terme de soroll multiplicatiu causat per les imperfeccions del coneixement de l'estat del canal en un únic soroll efectiu, i assumint que el soroll és gaussià i independent. Aquesta aproximació del pitjor soroll permet obtenir una expressió molt simple i ben coneguda de la informació mútua del canal. Una primera part d'aquesta tesi proposa un procediment senzill per a millorar aquesta cota associada al pitjor cas mitjançant una estratègia de repartiment de taxa: expressant l'entrada gaussiana del canal com a la suma de diverses entrades gaussianes independents i suposant que el receptor realitza una descodificació seqüencial dels fluxos d'informació, es mostra com obtenir una major cota inferior de la informació mútua del canal. En canals amb una única antena en transmissió, la distribució òptima de potència als diferents fluxos s'obté quan el seu nombre (capes) tendeix a infinit, i la potència associada a cada capa tendeix a zero. El límit associat a un nombre infinit de capes dóna lloc a una expressió integral de la cota de la informació mútua. En canals amb múltiples antenes s'obté un resultat similar. No obstant això, atès que la utilització de múltiples antenes proporciona més possibilitats de multiplexat espacial, el procediment dóna lloc a tota una família de cotes inferiors de la informació mútua associades a una combinació de capes infinita. S'estudia en detall aquesta família de cotes per al cas de coeficients d'esvaïments gaussians de mitjana zero, independents i idènticament distribuïts (conegut com esvaïment i.i.d. Rayleigh). S'obtenen diverses propietats de la família de cotes. És important destacar que per a coneixement asimptòtic perfecte del canal en recepció, qualsevol membre de la família de cotes és asimptòticament ajustat per alta relació senyal a soroll (SNR). En concret, la diferència entre la informació mútua i la seva cota inferior tendeix a zero quan la SNR tendeix a infinit sempre que el coneixement del canal tendeixi a ser exacte a mesura que la SNR tendeix a infinit. Una segona part d'aquesta tesi proposa un marc per a l'optimització d'una classe de funcions d'utilitat en canals amb múltiples antenes i esvaïments Rayleigh per blocs amb correlació en transmissió i sense informació sobre el canal a recepció. Una fracció temporal de cada bloc d'esvaïment es reserva per transmetre una seqüència de símbols d'entrenament mentre que la resta de mostres temporals s'utilitzen per transmetre informació. El receptor estima la matriu del canal partint de la seva observació sorollosa i descodifica la informació mitjançant la seva estimació del canal. Per a una classe de funcions d'utilitat que són funcions simètriques dels autovalors de la SNR matricial efectiva, els problemes consistents en optimitzar la seqüència pilot i el precodificador lineal són transformats en problemes convexos (o quasi-convexos) per a funcions d'utilitat còncaves (o quasi-còncaves). També s'estudia un subproblema important de l'optimització conjunta, que consisteix en el càlcul de les seqüències d'entrenament i dels precodificadors conjuntament Pareto-òptims. Integrant aquests procediments d'optimització en una iteració cíclica, s'obté un algoritme que convergeix a un òptim local conjunt per a qualsevol utilitat quasi-còncava

This work deals with the study of Back-2-Back converters, with special emphasis on its effects on the common mode. For the study performed, a review of the different known causes for bearing currents and the contribution to the common mode voltatge by different modulation methods, in both B2B and VSIs converters. Various original methods of modulation are studied that allow us significant reductions in the common mode provided, and therefore the life of the drives and connected loads. Finally, we study some original contributions using the common mode to take benefits when applying it does not cause us concern. / El present treball aborda l'estudi dels convertidors Back-2-Back, amb especial èmfasi en els seus efectes sobre el mode comú. Per l'estudi es realitza una revisió de les diferents causes conegudes dels corrents pels coixinets, així com del mode comú aportat pels diferents mètodes de modulació, tant en VSIs com en convertidors B2B. S'estudien diversos mètodes de modulació originals que permeten reduccions importants en el mode comú aportat, i per tant, en la vida dels accionaments i càrregues. Per finalitzar, s'estudien determinades aportacions també originals que permeten aprofitar el mode comú per a obtenir efectes beneficiosos quan l'aplicació del mateix no ens causa preocupació

In 1925, Lilienfeld patented the basic principle of field effect transistor (FET). Thirty-four years later, Kahng and Atalla invented the MOSFET. Since that time, it has become the most widely used type of transistor in Integrated Circuits (ICs) and then the most important device in the electronics industry. Progress in the field for at least the last 40 years has followed an exponential behavior in accordance with Moore¿s Law. That is, in order to achieve higher densities and performance at lower power consumption, MOS devices have been scaled down. But this aggressive scaling down of the physical dimensions of MOSFETs has required the introduction of a wide variety of innovative factors to ensure that they could still be properly manufactured. Transistors have expe- rienced an amazing journey in the last 10 years starting with strained channel CMOS transistors at 90nm, carrying on the introduction of the high-k/metal-gate silicon CMOS transistors at 45nm until the use of the multiple-gate transistor architectures at 22nm and at recently achieved 14nm technology node. But, what technology will be able to produce sub-10nm transistors? Different novel materials and devices are being investigated. As an extension and enhancement to current MOSFETs some promising devices are n-type III-V and p-type Germanium FETs, Nanowire and Tunnel FETs, Graphene FETs and Carbon Nanotube FETs. Also, non-conventional FETs and other charge-based information carrier devices and alternative information processing devices are being studied. This thesis is focused on carbon nanotube technology as a possible option for sub-10nm transistors. In recent years, carbon nanotubes (CNTs) have been attracting considerable attention in the field of nanotechnology. They are considered to be a promising substitute for silicon channel because of their small size, unusual geometry (1D structure), and extraordinary electronic properties, including excellent carrier mobility and quasi-ballistic transport. In the same way, carbon nanotube field-effect transistors (CNFETs) could be potential substitutes for MOSFETs. Ideal CNFETs (meaning all CNTs in the transistor behave as semiconductors, have the same diameter and doping level, and are aligned and well-positioned) are predicted to be 5x faster than silicon CMOS, while consuming the same power. However, nowadays CNFETs are also affected by manufacturing variability, and several significant challenges must be overcome before these benefits can be achieved. Certain CNFET manufacturing imperfections, such as CNT diameter and doping variations, mispositioned and misaligned CNTs, high metal-CNT contact resistance, the presence of metallic CNTs (m-CNTs), and CNT density variations, can affect CNFET performance and reliability and must be addressed. The main objective of this thesis is to analyze the impact of the current CNFET manufacturing challenges on multi-channel CNFET performance from the point of view of variability and reliability and at different levels, device and circuit level. Assuming that CNFETs are not ideal or non-homogeneous because of today CNFET manufacturing imperfections, we propose a methodology of analysis that based on a CNFET ideal compact model is able to simulate heterogeneous or non-ideal CNFETs; that is, transistors with different number of tubes that have different diameters, are not uniformly spaced, have different source/drain doping levels, and, most importantly, are made up not only of semiconducting CNTs but also metallic ones. This method will allow us to analyze how CNT-specific variations affect CNFET device characteristics and parameters and CNFET digital circuit performance. Furthermore, we also derive a CNFET failure model and propose an alternative technique based on fault-tolerant architectures to deal with the presence of m-CNTs, one of the main causes of failure in CNFET circuits.