Une des principales solutions technologiques liées à la réduction d'échelle de la technologie CMOS est aujourd'hui clairement orientée vers les transistors MOSFET faiblement dopés à multiples grilles. Ceux-ci proposent une meilleure immunité contre les effets canaux courts comparés aux transistors MOSFET bulk planaires (cf. ITRS 2011). Parmi les MOSFETs à multiples grilles, le transistor FinFET SOI est un candidat intéressant de par la similarité de son processus de fabrication avec la technologie des transistors planaires. En parallèle, il existe une réelle attente de la part des concepteurs et des fonderies à disposer de modèles compacts efficaces numériquement, précis et proches de la physique, insérés dans les " design tools " permettant alors d'étudier et d'élaborer des circuits ambitieux en technologie FinFET. Cette thèse porte sur l'élaboration d'un modèle compact orienté conception du transistor FinFET valide aux dimensions nanométriques. Ce modèle prend en compte les effets canaux courts, la modulation de longueur de canal, la dégradation de la mobilité, leseffets de mécanique quantique et les transcapacités. Une validation de ce modèle est réalisée par des comparaisons avec des simulations TCAD 3D. Le modèle compact est implémenté en langage Verilog-A afin de simuler des circuits innovants à base de transistors FinFET. Une modélisation niveau-porte est développée pour la simulation de circuits numériques complexes. Cette thèse présente également un modèle compact générique de transistors MOSFET SOI canaux long faiblement dopés à multiple grilles. La dépendance à la température est prise en compte. Selon un concept de transformation géométrique, notre modèle compact du transistor MOSFET double grille planaire est étendu pour s'appliquer à tout autre type de transistor MOSFET à multiple grille (MuGFET). Une validation expérimentale du modèle MuGFET sur un transistor triple grille est proposée. Cette thèse apporte enfin des solutions pour la modélisation des transistors MOSFET double grille sans jonction.

The move from Standard Definition (SD) to High Definition (HD) represents a six times increases in data, which needs to be processed. With expanding resolutions and evolving compression, there is a need for high performance with flexible architectures to allow for quick upgrade ability. The technology advances in image display resolutions, advanced compression techniques, and video intelligence. Software implementation of these systems can attain accuracy with tradeoffs among processing performance (to achieve specified frame rates, working on large image data sets), power and cost constraints. There is a need for new architectures to be in pace with the fast innovations in video and imaging. It contains dedicated hardware implementation of the pixel and frame rate processes on Field Programmable Gate Array (FPGA) to achieve the real-time performance. The following outlines the contributions of the dissertation. (1) We develop a target detection system by applying a novel running average mean threshold (RAMT) approach to globalize the threshold required for background subtraction. This approach adapts the threshold automatically to different environments (indoor and outdoor) and different targets (humans and vehicles). For low power consumption and better performance, we design the complete system on FPGA. (2) We introduce a safe distance factor and develop an algorithm for occlusion occurrence detection during target tracking. A novel mean-threshold is calculated by motion-position analysis. (3) A new strategy for gesture recognition is developed using Combinational Neural Networks (CNN) based on a tree structure. Analysis of the method is done on American Sign Language (ASL) gestures. We introduce novel point of interests approach to reduce the feature vector size and gradient threshold approach for accurate classification. (4) We design a gesture recognition system using a hardware/ software co-simulation neural network for high speed and low memory storage requirements provided by the FPGA. We develop an innovative maximum distant algorithm which uses only 0.39% of the image as the feature vector to train and test the system design. Database set gestures involved in different applications may vary. Therefore, it is highly essential to keep the feature vector as low as possible while maintaining the same accuracy and performance

With the advancement of CMOS process and fabrication, it has been a trend to maximize digital design while minimize analog correspondents in mixed-signal system designs. So is the case for PLL. PLL has always been a traditional mixed-signal system limited by analog part performance. Around 2000, there emerged ADPLL of which all the blocks besides oscillator are implemented in digital circuits. There have been successful examples in application of Bluetooth, and it is moving to improve results for application of WiMax and ad-hoc frequency hopping communication link. Based on the theoretic and measurement results of existing materials, ADPLL has shown advantages such as fast time-to-market, low area, low cost and better system integration; but it also showed disadvantages in frequency resolution and phase noise, etc. Also this new topic still opens questions in many researching points important to PLL such as tracking behavior and quantization effect. In this thesis, a non-linear phase domain model for all digital phase-locked loop (ADPLL) was established and validated. Based on that, we analyzed that ADPLL phase noise prediction derived from traditional linear quantization model became inaccurate in non-linear cases because its probability density of quantization error did not meet the premise assumption of linear model. The phenomena of bandwidth expansion and in-band phase noise decreasing peculiar to integer-N ADPLL were demonstrated and explained by matlab and verilog behavior level simulation test bench. The expression of threshold quantization step was defined and derived as the method to distinguish whether an integer-N ADPLL was in non-linear cases or not, and the results conformed to those of matlab simulation. A simplified approximation model for non-linear integer-N ADPLL with noise sources was established to predict in-band phase noise, and the trends of the results conformed to those of matlab simulation. Other basic analysis serving for the conclusions above covered: ADPLL loop dynamics, traditional linear theory and its quantitative limitations and numerical analysis of random number. Finally, a present measurement setup was demonstrated and the results were analyzed for future work.

Les réseaux de capteurs sans fil nécessitent des fonctions de calcul et de transmissionradio associées à chaque capteur. Les SoCs RF intégrant ces fonctions doivent avoir uneautonomie la plus grande possible et donc une très faible consommation. Aujourd'hui, leursperformances énergétiques pourraient être fortement améliorées par des systèmes d'alimentationinnovants. En effet, les circuits d'alimentation remplissent leur fonction classique de conversiond'énergie mais aussi des fonctions d'isolation des blocs RF et digitaux. Leurs performancess'évaluent donc en termes d'efficacité énergétique et de réponse transitoire mais aussi d'isolationentre blocs et de réjection de bruit.Ce travail de thèse concerne l'intégration du système de gestion et de distribution del’énergie aux différents blocs RF d’un émetteur/récepteur en élaborant une méthodologie « topdown» pour déterminer la sensibilité de chaque bloc à son alimentation et en construisant unearchitecture innovante et dynamique de gestion/distribution de l'énergie sur le SoC. Cetteméthodologie repose sur la disponibilité de régulateurs de tension présentant des performancesadaptées. Un deuxième volet du travail de thèse a donc été de réaliser un régulateur linéaire detype LDO à forte réjection sur une bande passante relativement large et bien adapté àl'alimentation de blocs RF très sensibles aux bruits de l'alimentation. / Wireless sensor networks require calculation functions and radiofrequencytransmission modules within each sensor. RF SoCs integrating these functions must have thebiggest battery life and so a very small consumption. Today, innovative power managementsystems could highly enhance the energy performances of this type of RF SoC. Indeed, thesepower systems perform energy conversion and also the isolation functions of RF and digitalblocks. Their features are thus estimated in terms of energy efficiency, transient response and alsoisolation between blocks and noise rejection.This thesis work concerns the integration of the power management systems and itsdistribution channels into different ultra-low-power SoCs. This was achieved mainly thanks to thedevelopment of a new “top-down” approach. This new methodology consists of determining thesensibility of every block to its power supply and of designing an innovative and dynamicarchitecture of power management circuits on the SoC. This study ends up in the implementationof a very efficient low dropout (LDO) regulator for noise-sensitive low-current RF blocks inmixed SoC applications. The fabricated prototype achieves a high power supply rejection for awide range of frequencies.

The objective of this thesis is to automate the design of round-robin arbiter logic. The resulting arbitration logic is more than 1.8X times faster than the fastest prior state-of-the-art arbitration logic the author could find reported in the literature. The generated arbiter implemented in a single chip is fast enough in 0.25ьm CMOS technology to achieve terabit switching with a single chip computer network switch. Moreover, this arbiter is applicable to crossbar (Xbar) arbitration logic. The generated Xbar, customized according to user specifications, provides multiple communication paths among masters and slaves. As the number of transistors on a single chip increases rapidly, there is a productivity gap between the number of transistors available in a chip and the number of transistors per hour a chip designer designs. One solution to reduce this productivity gap is to increase the use of Silicon Intellectual Property (SIP) cores. However, a SIP core should be customized before being used in a system different than the one for which it was designed. Thus, to reconfigure the SIP core, either an engineer must spend significant effort altering the core by hand or else an enhanced CAD tool can automatically customize the core according to customer specifications. In this thesis, we present SIP generator tools for arbiter and Xbar generation. First, we introduce a Round-robin Arbiter Generator (RAG). The RAG can generate a hierarchical Bus Arbiter (BA) which is faster than all known previous approaches. RAG can also generate a hierarchical Switch Arbiter (SA) which is faster than all known previous approaches. Using a 0.25ьm TSMC standard cell library from LEDA Systems, we show the arbitration time of a 32x32 SA and demonstrate that our SA meets the time constraint to achieve terabit throughput. Furthermore, using a novel token-passing hierarchical arbitration scheme, our 32x32 SA performs better than the Ping-Pong Arbiter and Programmable Priority Encoder by factors of 1.8X and 2.3X, respectively, with less power dissipation. Finally, we present an Xbar switch Generator (X-Gt) tool that automatically configures a crossbar for a multiprocessor System-on-a-Chip (SoC). An Xbar is generated in Register Transfer Level (RTL) Verilog HDL.

Les réseaux de capteurs sans fil nécessitent des fonctions de calcul et de transmissionradio associées à chaque capteur. Les SoCs RF intégrant ces fonctions doivent avoir uneautonomie la plus grande possible et donc une très faible consommation. Aujourd'hui, leursperformances énergétiques pourraient être fortement améliorées par des systèmes d'alimentationinnovants. En effet, les circuits d'alimentation remplissent leur fonction classique de conversiond'énergie mais aussi des fonctions d'isolation des blocs RF et digitaux. Leurs performancess'évaluent donc en termes d'efficacité énergétique et de réponse transitoire mais aussi d'isolationentre blocs et de réjection de bruit.Ce travail de thèse concerne l'intégration du système de gestion et de distribution del'énergie aux différents blocs RF d'un émetteur/récepteur en élaborant une méthodologie " topdown" pour déterminer la sensibilité de chaque bloc à son alimentation et en construisant unearchitecture innovante et dynamique de gestion/distribution de l'énergie sur le SoC. Cetteméthodologie repose sur la disponibilité de régulateurs de tension présentant des performancesadaptées. Un deuxième volet du travail de thèse a donc été de réaliser un régulateur linéaire detype LDO à forte réjection sur une bande passante relativement large et bien adapté àl'alimentation de blocs RF très sensibles aux bruits de l'alimentation.

During the design and verification of the Hyperstone S5 flash memory controller, we developed a highly effective way to use the SystemVerilog direct programming interface (DPI) to integrate an instruction set simulator (ISS) and a software debugger in logic simulation. The processor simulation was performed by the ISS, while all other hardware components were simulated in the logic simulator. The ISS integration allowed us to filter many of the bus accesses out of the logic simulation, accelerating runtime drastically. The software debugger integration freed both hardware and software engineers to work in their chosen development environments. Other benefits of this approach include testing and integrating code earlier in the design cycle and more easily reproducing, in simulation, problems found in FPGA prototypes.

No description available.

Like cell to the human body, transistors are the basic building blocks of any electronics circuits. Silicon has been the industries obvious choice for making transistors. Transistors with large size occupy large chip area, consume lots of power and the number of functionalities will be limited due to area constraints. Thus to make the devices smaller, smarter and faster, the transistors are aggressively scaled down in each generation. Moore's law states that the transistors count in any electronic circuits doubles every 18 months. Following this Moore's law, the transistor has already been scaled down to 14 nm. However there are limitations to how much further these transistors can be scaled down. Particularly below 10 nm, these silicon based transistors hit the fundamental limits like loss of gate control, high leakage and various other short channel effects. Thus it is not possible to favor the silicon transistors for future electronics applications. As a result, the research has shifted to new device concepts and device materials alternative to silicon. Carbon is the next abundant element found in the Earth and one of such carbon based nanomaterial is graphene. Graphene when extracted from Graphite, the same material used as the lid in pencil, have a tremendous potential to take future electronics devices to new heights in terms of size, cost and efficiency. Thus after its first experimental discovery of graphene in 2004, graphene has been the leading research area for both academics as well as industries. This dissertation is focused on the analysis and optimization of graphene based circuits for future electronics. The first part of this dissertation considers graphene based transistors for analog/radio frequency (RF) circuits. In this section, a dual gate Graphene Field Effect Transistor (GFET) is considered to build the case study circuits like voltage controlled oscillator (VCO) and low noise amplifier (LNA). The behavioral model of the transistor is modeled in different tools: well accepted EDA (electronic design automation) and a non-EDA based tool i.e. \simscape. This section of the dissertation addresses the application of non-EDA based concepts for the analysis of new device concepts, taking LC-VCO and LNA as a case study circuits. The non-EDA based approach is very handy for a new device material when the concept is not matured and the model files are not readily available from the fab. The results matches very well with that of the EDA tools. The second part of the section considers application of multiswarm optimization (MSO) in an EDA tool to explore the design space for the design of LC-VCO. The VCO provides an oscillation frequency at 2.85 GHz, with phase noise of less than -80 dBc/Hz and power dissipation less than 16 mW. The second part of this dissertation considers graphene nanotube field effect transistors (GNRFET) for the application of digital domain. As a case study, static random access memory (SRAM) hs been design and the results shows a very promising future for GNRFET based SRAM as compared to silicon based transistor SRAM. The power comparison between the two shows that GNRFET based SRAM are 93% more power efficient than the silicon transistor based SRAM at 45 nm. In summary, the dissertation is to expected to aid the state of the art in following ways: 1) A non-EDA based tool has been used to characterize the device and measure the circuit performance. The results well matches to that obtained from the EDA tools. This tool becomes very handy for new device concepts when the simulation needs to be fast and accuracy can be tradeoff with. 2)Since an analog domain lacks well-design design paradigm, as compared to digital domain, this dissertation considers case study circuits to design the circuits and apply optimization. 3) Performance comparison of GNRFET based SRAM to the conventional silicon based SRAM shows that with maturation of the fabrication technology, graphene can be very useful for digital circuits as well.

[ES] La electrofisiología extracelular es una técnica ampliamente usada en investigación neurocientífica, la cual estudia el funcionamiento del cerebro mediante la medición de campos eléctricos generados por la actividad neuronal. Esto se realiza a través de electrodos implantados en el cerebro y conectados a dispositivos electrónicos para amplificación y digitalización de las señales. De los muchos modelos animales usados en experimentación, las ratas y los ratones se encuentran entre las especies más comúnmente utilizadas. Actualmente, la experimentación electrofisiológica busca condiciones cada vez más complejas, limitadas por la tecnología de los dispositivos de adquisición. Dos aspectos son de particular interés: Realimentación de lazo cerrado y comportamiento en condiciones naturales. En esta tesis se presentan desarrollos con el objetivo de mejorar diferentes facetas de estos dos problemas. La realimentación en lazo cerrado se refiere a todas las técnicas en las que los estímulos son producidos en respuesta a un evento generado por el animal. La latencia debe ajustarse a las escalas temporales bajo estudio. Los sistemas modernos de adquisición presentan latencias en el orden de los 10ms. Sin embargo, para responder a eventos rápidos, como pueden ser los potenciales de acción, se requieren latencias por debajo de 1ms. Además, los algoritmos para detectar los eventos o generar los estímulos pueden ser complejos, integrando varias entradas de datos en tiempo real. Integrar el desarrollo de dichos algoritmos en las herramientas de adquisición forma parte del diseño experimental. Para estudiar comportamientos naturales, los animales deben ser capaces de moverse libremente en entornos emulando condiciones naturales. Experimentos de este tipo se ven dificultados por la naturaleza cableada de los sistemas de adquisición. Otras restricciones físicas, como el peso de los implantes o limitaciones en el consumo de energía, pueden también afectar a la duración de los experimentos, limitándola. La experimentación puede verse enriquecida cuando los datos electrofisiológicos se ven complementados con múltiples fuentes distintas. Por ejemplo, seguimiento de los animales o miscroscopía. Herramientas capaces de integrar datos independientemente de su origen abren la puerta a nuevas posibilidades. Los avances tecnológicos presentados abordan estas limitaciones. Se han diseñado dispositivos con latencias de lazo cerrado inferiores a 200us que permiten combinar cientos de canales electrofisiológicos con otras fuentes de datos, como vídeo o seguimiento. El software de control para estos dispositivos se ha diseñado manteniendo la flexibilidad como objetivo. Se han desarrollado interfaces y estándares de naturaleza abierta para incentivar el desarrollo de herramientas compatibles entre ellas. Para resolver los problemas de cableado se siguieron dos métodos distintos. Uno fue el desarrollo de headstages ligeros combinados con cables coaxiales ultra finos y conmutadores activos, gracias al seguimiento de animales. Este desarrollo permite reducir el esfuerzo impuesto a los animales, permitiendo espacios amplios y experimentos de larga duración, al tiempo que permite el uso de headstages con características avanzadas. Paralelamente se desarrolló un tipo diferente de headstage, con tecnología inalámbrica. Se creó un algoritmo de compresión digital especializado capaz de reducir el ancho de banda a menos del 65% de su tamaño original, ahorrando energía. Esta reducción permite baterías más ligeras y mayores tiempos de operación. El algoritmo fue diseñado para ser capaz de ser implementado en una gran variedad de dispositivos. Los desarrollos presentados abren la puerta a nuevas posibilidades experimentales para la neurociencia, combinando adquisición elextrofisiológica con estudios conductuales en condiciones naturales y estímulos complejos en tiempo real. / [CA] L'electrofisiologia extracel·lular és una tècnica àmpliament utilitzada en la investigació neurocientífica, la qual permet estudiar el funcionament del cervell mitjançant el mesurament de camps elèctrics generats per l'activitat neuronal. Això es realitza a través d'elèctrodes implantats al cervell, connectats a dispositius electrònics per a l'amplificació i digitalització dels senyals. Dels molts models animals utilitzats en experimentació electrofisiològica, les rates i els ratolins es troben entre les espècies més utilitzades. Actualment, l'experimentació electrofisiològica busca condicions cada vegada més complexes, limitades per la tecnologia dels dispositius d'adquisició. Dos aspectes són d'especial interès: La realimentació de sistemes de llaç tancat i el comportament en condicions naturals. En aquesta tesi es presenten desenvolupaments amb l'objectiu de millorar diferents aspectes d'aquestos dos problemes. La realimentació de sistemes de llaç tancat es refereix a totes aquestes tècniques on els estímuls es produeixen en resposta a un esdeveniment generat per l'animal. La latència ha d'ajustar-se a les escales temporals sota estudi. Els sistemes moderns d'adquisició presenten latències en l'ordre dels 10ms. No obstant això, per a respondre a esdeveniments ràpids, com poden ser els potencials d'acció, es requereixen latències per davall de 1ms. A més a més, els algoritmes per a detectar els esdeveniments o generar els estímuls poden ser complexos, integrant varies entrades de dades a temps real. Integrar el desenvolupament d'aquests algoritmes en les eines d'adquisició forma part del disseny dels experiments. Per a estudiar comportaments naturals, els animals han de ser capaços de moure's lliurement en ambients emulant condicions naturals. Aquestos experiments es veuen limitats per la natura cablejada dels sistemes d'adquisició. Altres restriccions físiques, com el pes dels implants o el consum d'energia, poden també limitar la duració dels experiments. L'experimentació es pot enriquir quan les dades electrofisiològiques es complementen amb dades de múltiples fonts. Per exemple, el seguiment d'animals o microscòpia. Eines capaces d'integrar dades independentment del seu origen obrin la porta a noves possibilitats. Els avanços tecnològics presentats tracten aquestes limitacions. S'han dissenyat dispositius amb latències de llaç tancat inferiors a 200us que permeten combinar centenars de canals electrofisiològics amb altres fonts de dades, com vídeo o seguiment. El software de control per a aquests dispositius s'ha dissenyat mantenint la flexibilitat com a objectiu. S'han desenvolupat interfícies i estàndards de naturalesa oberta per a incentivar el desenvolupament d'eines compatibles entre elles. Per a resoldre els problemes de cablejat es van seguir dos mètodes diferents. Un va ser el desenvolupament de headstages lleugers combinats amb cables coaxials ultra fins i commutadors actius, gràcies al seguiment d'animals. Aquest desenvolupament permet reduir al mínim l'esforç imposat als animals, permetent espais amplis i experiments de llarga durada, al mateix temps que permet l'ús de headstages amb característiques avançades. Paral·lelament es va desenvolupar un tipus diferent de headstage, amb tecnologia sense fil. Es va crear un algorisme de compressió digital especialitzat capaç de reduir l'amplada de banda a menys del 65% de la seua grandària original, estalviant energia. Aquesta reducció permet bateries més lleugeres i majors temps d'operació. L'algorisme va ser dissenyat per a ser capaç de ser implementat a una gran varietat de dispositius. Els desenvolupaments presentats obrin la porta a noves possibilitats experimentals per a la neurociència, combinant l'adquisició electrofisiològica amb estudis conductuals en condicions naturals i estímuls complexos en temps real. / [EN] Extracellular electrophysiology is a technique widely used in neuroscience research. It can offer insights on how the brain works by measuring the electrical fields generated by neural activity. This is done through electrodes implanted in the brain and connected to amplification and digitization electronic circuitry. Of the many animal models used in electrophysiology experimentation, rodents such as rats and mice are among the most popular species. Modern electrophysiology experiments seek increasingly complex conditions that are limited by acquisition hardware technology. Two particular aspects are of special interest: Closed-loop feedback and naturalistic behavior. In this thesis, we present developments aiming to improve on different facets of these two problems. Closed-loop feedback encompasses all techniques in which stimuli is produced in response of an event generated by the animal. Latency, the time between trigger event and stimuli generation, must adjust to the biological timescale being studied. While modern acquisition systems feature latencies in the order of 10ms, response to fast events such as high-frequency electrical transients created by neuronal activity require latencies under $1ms$. In addition, algorithms for triggering or generating closed-loop stimuli can be complex, integrating multiple inputs in real-time. Integration of algorithm development into acquisition tools becomes an important part of experiment design. For electrophysiology experiments featuring naturalistic behavior, animals must be able to move freely in ecologically meaningful environments, mimicking natural conditions. Experiments featuring elements such as large arenaa, environmental objects or the presence of another animals are, however, hindered by the wired nature of acquisition systems. Other physical constraints, such as implant weight or power restrictions can also affect experiment time, limiting their duration. Beyond the technical limits, complex experiments are enriched when electrophysiology data is integrated with multiple sources, for example animal tracking or brain microscopy. Tools allowing mixing data independently of the source open new experimental possibilities. The technological advances presented on this thesis addresses these topics. We have designed devices with closed-loop latencies under 200us while featuring high-bandwidth interfaces. These allow the simultaneous acquisition of hundreds of electrophysiological channels combined with other heterogeneous data sources, such as video or tracking. The control software for these devices was designed with flexibility in mind, allowing easy implementation of closed-loop algorithms. Open interface standards were created to encourage the development of interoperable tools for experimental data integration. To solve wiring issues in behavioral experiments, we followed two different approaches. One was the design of light headstages, coupled with ultra-thin coaxial cables and active commutator technology, making use of animal tracking. This allowed to reduce animal strain to a minimum allowing large arenas and prolonged experiments with advanced headstages. A different, wireless headstage was also developed. We created a digital compression algorithm specialized for neural electrophysiological signals able to reduce data bandwidth to less than 65.5% its original size without introducing distortions. Bandwidth has a large effect on power requirements. Thus, this reduction allows for lighter batteries and extended operational time. The algorithm is designed to be able to be implemented in a wide variety of devices, requiring low hardware resources and adding negligible power requirements to a system. Combined, the developments we present open new possibilities for neuroscience experiments combining electrophysiology acquisition with natural behaviors and complex, real-time, stimuli. / The research described in this thesis was carried out at the Polytechnic University of Valencia (Universitat Politècnica de València), Valencia, Spain in an extremely close collaboration with the Neuroscience Institute - Spanish National Research Council - Miguel Hernández University (Instituto de Neurociencias - Consejo Superior de Investigaciones Cientí cas - Universidad Miguel Hernández), San Juan de Alicante, Spain. The projects described in chapters 3 and 4 were developed in collabo- ration with, and funded by, Open Ephys, Cambridge, MA, USA and OEPS - Eléctronica e produção, unipessoal lda, Algés, Portugal. / Cuevas López, A. (2021). Development of Advanced Closed-Loop Brain Electrophysiology Systems for Freely Behaving Rodents [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/179718