With the explosive growth in wireless applications, the last decade witnessed an ever-increasing test challenge for radio frequency (RF) circuits. While the design community has pushed the envelope far into the future, by expanding CMOS process to be used with high-frequency wireless devices, test methodology has not advanced at the same pace. Consequently, testing such devices has become a major bottleneck in high-volume production, further driven by the growing need for tighter quality control. RF devices undergo testing during the prototype phase and during high-volume manufacturing (HVM). The benchtop test equipment used throughout prototyping is very precise yet specialized for a subset of functionalities. HVM calls for a different kind of test paradigm that emphasizes throughput and sufficiency, during which the projected performance parameters are measured one by one for each device by automated test equipment (ATE) and compared against defined limits called specifications. The set of tests required for each product differs greatly in terms of the equipment required and the time taken to test individual devices. Together with signal integrity, precision, and repeatability concerns, the initial cost of RF ATE is prohibitively high. As more functionality and protocols are integrated into a single RF device, the required number of specifications to be tested also increases, adding to the overall cost of testing, both in terms of the initial and recurring operating costs. In addition to the cost problem, RF testing proposes another challenge when these components are integrated into package-level system solutions. In systems-on-packages (SOP), the test problems resulting from signal integrity, input/output bandwidth (IO), and limited controllability and observability have initiated a paradigm shift in high-speed analog testing, favoring alternative approaches such as built-in tests (BIT) where the test functionality is brought into the package. This scheme can make use of a low-cost external tester connected through a low-bandwidth link in order to perform demanding response evaluations, as well as make use of the analog-to-digital converters and the digital signal processors available in the package to facilitate testing. Although research on analog built-in test has demonstrated hardware solutions for single specifications, the paradigm shift calls for a rather general approach in which a single methodology can be applied across different devices, and multiple specifications can be verified through a single test hardware unit, minimizing the area overhead. Specification-based alternate test methodology provides a suitable and flexible platform for handling the challenges addressed above. In this thesis, a framework that integrates ATE and system constraints into test stimulus generation and test response extraction is presented for the efficient production testing of high-performance RF devices using specification-based alternate tests. The main components of the presented framework are as follows: Constraint-driven RF alternate test stimulus generation: An automated test stimulus generation algorithm for RF devices that are evaluated by a specification-based alternate test solution is developed. The high-level models of the test signal path define constraints in the search space of the optimized test stimulus. These models are generated in enough detail such that they inherently define limitations of the low-cost ATE and the I/O restrictions of the device under test (DUT), yet they are simple enough that the non-linear optimization problem can be solved empirically in a reasonable amount of time. Feature extractors for BIT: A methodology for the built-in testing of RF devices integrated into SOPs is developed using additional hardware components. These hardware components correlate the high-bandwidth test response to low bandwidth signatures while extracting the test-critical features of the DUT. Supervised learning is used to map these extracted features, which otherwise are too complicated to decipher by plain mathematical analysis, into the specifications under test. Defect-based alternate testing of RF circuits: A methodology for the efficient testing of RF devices with low-cost defect-based alternate tests is developed. The signature of the DUT is probabilistically compared with a class of defect-free device signatures to explore possible corners under acceptable levels of process parameter variations. Such a defect filter applies discrimination rules generated by a supervised classifier and eliminates the need for a library of possible catastrophic defects.

Les systèmes Ultra Larges Bandes (ULB) représentent une nouvelle technologie de communication sans fil pour la transmission d'informations à très hauts débits. Avec une Densité Spectrale de Puissance (DSP) inférieure à -41.3 dBm/MHz, les distances de propagation sont considérablement restreintes (< 20 m). C'est en février 2002, que la commission fédérale de communication des Etats-Unis (FCC) a alloué pour les systèmes ULB une bande de fréquence comprise entre 3,1 et 10,6 GHz. Depuis, de nombreuses divergences sont apparues suivant les grandes zones géographiques de normalisation.<br />Au travers des nombreuses techniques de transmission ULB, nous étudions la modulation multi-bandes à répétition en fréquence orthogonale (MB-OFDM) qui a été proposée par l'European Computing Machineries Association (ECMA) en temps que standard en 2005.L'efficacité des signaux Impulsionnel Radio (IR), précurseur à l'ULB, a cependant été montrée. Nous présentons à ce titre une topologie originale nommée Multi-Bandes On-Off Keying (MB-OOK).<br />Dans le cadre de cette thèse, nous développons un outil de simulation système (SST) afin d'analyser les performances d'un système ULB-OFDM. Nous étudions l'influence sur le taux d'erreur binaire (BER) des non-linéarités du front-end de réception Radio-Fréquence (RF) et de l'amplification de puissance faible bruit (LNA) dans une chaîne de transmission globale.<br />Dans le cadre des systèmes MB-OOK, différents types de filtres sont envisagés. L'impact sur le BER est étudié afin de connaître les propriétés prédominantes des filtres sur les performances de transmission.<br />Le domaine de l'optique peut également être avantageusement employé dans des processus de transmission ULB. Les réseaux larges bandes radio sur fibres (RoF) présentent de nombreux intérêts pour fournir à moindre coût la distribution du signal mais également la possibilité de pouvoir le traiter directement. Dans ce sens, Nous évaluons les possibilités de convertir fréquentiellement des signaux de type ULB-OFDM, ainsi que d'évaluer l'impact de transmetteurs Electrique/Optique (E/O) sur les performances d'un système ULB sur fibre à partir des valeurs du vecteur d'erreur d'amplitude (EVM).

No description available.

La qPCR est actuellement la technique de référence en matière de quantification d'ADN. Elle peut être définie comme une amplification exponentielle, cyclique et ciblée de la séquence d'ADN cible. Le caractère exponentiel de la qPCR est à la fois à l'origine de la sensibilité de la méthode mais aussi d'une potentielle variabilité inter-échantillons. Cette variabilité est compensée par le caractère cyclique de la méthode qui entraine une synchronisation de la réaction pour tous les échantillons à chaque cycle. L'amplification ciblée de la séquence choisie traduit quand à elle la spécificité de la méthode. Néanmoins, cette dernière propriété de la PCR reste la moins vérifiée. La spécificité de la qPCR est indiscutable lorsque la cible à détecter se trouve en quantité suffisante (>100 copies environ). En revanche, pour des quantités plus faibles ou en présence d'inhibiteurs, la spécificité diminue ou disparaît (limites de détection et de quantification). Cette perte de spécificité retentit de façon significative sur la précision et la reproductibilité du dosage à réaliser. Cependant, si l'hybridation non spécifique est inéluctable dans ces conditions, l'amplification non spécifique consécutive peut être limitée, voire supprimée au moyen d'amorces de PCR soit plus spécifiques, soit moins susceptibles de générer des différences inter-échantillons. La RT-qPCR, grâce à une étape initiale de transcription inverse (conversion d'un ARN en un ADN complémentaire) permet la quantification des ARN. Cependant, la transcription inverse reste moins reproductible que la PCR, générant des différences inter-échantillons délétères en diagnostic comme en recherche. Au cours de ces travaux, je propose des méthodes originales améliorant de façon significative différentes étapes de ces techniques. Premièrement, je propose une amélioration de la standardisation de l'étape de transcription inverse capable de diminuer de façon significative la variabilité inter-échantillons ; l'utilisation d'un volume constant d'extrait d'ARN pour chaque échantillon améliore considérablement la précision de la quantification d'ARN messagers. Deuxièmement, je propose une modification des amorces par incorporation de résidus d'acides nucléiques bloqués (LNA) ; cette modification permet d'augmenter la spécificité des amorces aux limites de la détection. Enfin, je propose une méthode simple et économique permettant la mesure directe de la température de fusion (Tm) dans les conditions réelles de la PCR. Ce paramètre, considéré comme capital pour la réalisation des dosages par qPCR, est généralement obtenu par des méthodes prédictives peu précises. De plus, cette méthode doit permettre de déterminer précisément les paramètres thermodynamiques des amorces (∆G, ∆H et ∆S) et ainsi avoir accès d'une part au pourcentage réel d'amorces hybridées en fonction de la température et d'autre part d'identifier la susceptibilité des amorces aux inhibiteurs.

Um novo circuito amplificador de 50 MHz - 1 GHz com alta linearidade para o padrão IEEE 802.22 “wireless regional area” (WRAN) é apresentado. Ele foi implementado sem nenhum indutor e oferece uma saída diferencial para ser utilizada como balun. Técnicas de cancelamento de ruído e aumento de linearidade foram usadas para melhorar a performace do amplificador de modo que eles pudessem ser otimizados separadamente. A linearidade foi melhorada utilizando transistores conectados como diodo. O amplificador foi implementado em um processo CMOS 130 nm, em uma área compacta de 136 m x 71 m. As simulações são apresentadas para esquemáticos pós-leiaute para duas classes diferentes de projeto: um visando a melhor linearidade e o outro a melhor Figura de Ruído (FR). Quando otimizado para melhor linearidade, os resultados de simulação atingem um ganho de tensão > 23.7 dB (ganho de potência > 19.1 dB), uma figura de ruído < 3.6 dB na banda inteira (com 2.4 dB min), um ponto de intersecção de terceira ordem (IIP3) > 3.3 dBm (7.6 dBm max) e um coeficiente de reflexão de entrada S11 < -16 dB. Quando otimizado para melhor figura de ruído, ele atinge um ganho de tensão > 24.7 dB (ganho de potência > 19.8 dB), uma FR < 2 dB na banda inteira, um IIP3 > -0.3 dBm e um S11 < -11 dB. Resultados de simulação Monte Carlo confirmam baixa sensibilidade à variabilidade de processo. Além disso, uma baixa sensibilidade com a temperatura na faixa de -55 até 125 C foi observada para Ganho, FR e S11. Consumo de potência é 17.6 mA sob fonte de alimentação de 1.2 V. / A new 50 MHz - 1 GHz low-noise amplifier circuit with high linearity for IEEE 802.22 wireless regional area network (WRAN) is presented. It was implemented without any inductor and offers a differential output for balun use. Noise cancelling and linearity boosting techniques were used to improve the amplifier performance in a way they can be separately optimized. Linearity was improved using diode-connected transistors. The amplifier was implemented in a 130 nm CMOS process in a compact 136 m x 71 m area. Simulations are presented for post-layout schematics for two classes of design: one for best linearity, another for best noise figure (NF). When optimized for best linearity, simulation results achieve a voltage gain > 23.7 dB (power gain > 19.1 dB), a NF < 3.6 dB over the entire band (with 2.4 dB min figure), an input third-order intercept point (IIP3) > 3.3 dBm (7.6 dBm max.) and an input power reflection coefficient S11 < -16 dB. When optimized for best NF, it achieves a voltage gain > 24.7 dB (power gain > 19.8 dB), a NF < 2 dB over the entire band, an IIP3 > -0.3 dBm and an S11 < -11 dB. Monte Carlo simulation results confirm low sensitivity to process variations. Also a low sensitivity to temperature within the range -55 to 125 C was observed for Gain, NF and S11. Power consumption is 17.6 mA under a 1.2 V supply.

La qPCR est actuellement la technique de référence en matière de quantification d'ADN. Elle peut être définie comme une amplification exponentielle, cyclique et ciblée de la séquence d'ADN cible. Le caractère exponentiel de la qPCR est à la fois à l'origine de la sensibilité de la méthode mais aussi d'une potentielle variabilité inter-échantillons. Cette variabilité est compensée par le caractère cyclique de la méthode qui entraine une synchronisation de la réaction pour tous les échantillons à chaque cycle. L'amplification ciblée de la séquence choisie traduit quand à elle la spécificité de la méthode. Néanmoins, cette dernière propriété de la PCR reste la moins vérifiée. La spécificité de la qPCR est indiscutable lorsque la cible à détecter se trouve en quantité suffisante (>100 copies environ). En revanche, pour des quantités plus faibles ou en présence d'inhibiteurs, la spécificité diminue ou disparaît (limites de détection et de quantification). Cette perte de spécificité retentit de façon significative sur la précision et la reproductibilité du dosage à réaliser. Cependant, si l'hybridation non spécifique est inéluctable dans ces conditions, l'amplification non spécifique consécutive peut être limitée, voire supprimée au moyen d'amorces de PCR soit plus spécifiques, soit moins susceptibles de générer des différences inter-échantillons. La RT-qPCR, grâce à une étape initiale de transcription inverse (conversion d'un ARN en un ADN complémentaire) permet la quantification des ARN. Cependant, la transcription inverse reste moins reproductible que la PCR, générant des différences inter-échantillons délétères en diagnostic comme en recherche. Au cours de ces travaux, je propose des méthodes originales améliorant de façon significative différentes étapes de ces techniques. Premièrement, je propose une amélioration de la standardisation de l'étape de transcription inverse capable de diminuer de façon significative la variabilité inter-échantillons ; l'utilisation d'un volume constant d'extrait d'ARN pour chaque échantillon améliore considérablement la précision de la quantification d'ARN messagers. Deuxièmement, je propose une modification des amorces par incorporation de résidus d'acides nucléiques bloqués (LNA) ; cette modification permet d'augmenter la spécificité des amorces aux limites de la détection. Enfin, je propose une méthode simple et économique permettant la mesure directe de la température de fusion (Tm) dans les conditions réelles de la PCR. Ce paramètre, considéré comme capital pour la réalisation des dosages par qPCR, est généralement obtenu par des méthodes prédictives peu précises. De plus, cette méthode doit permettre de déterminer précisément les paramètres thermodynamiques des amorces (∆G, ∆H et ∆S) et ainsi avoir accès d'une part au pourcentage réel d'amorces hybridées en fonction de la température et d'autre part d'identifier la susceptibilité des amorces aux inhibiteurs. / QPCR is currently the gold standard for DNA quantification of DNA. It can be defined as an exponential, cyclic and targeted amplification of a known DNA sequence. The exponential character of qPCR is both the cause of the sensitivity of the method but also a potential variability between samples. This variability is offset by the cyclical nature of the method that causes a synchronization of the reaction for all samples in each cycle. The targeted amplification of the selected sequence translates the specificity of the method. Nevertheless, this last property of PCR remains the least tested. The specificity of qPCR is unquestionable when the target to detect is sufficient (> 100 copies). However, for smaller quantities or in the presence of inhibitors, the specificity decreases or disappears (limits of detection and quantification). This loss of specificity will have a significant effect upon the accuracy and reproducibility of the assay to be performed. However, if non-specific hybridizations are unavoidable in these conditions, consecutive nonspecific amplification may be limited or eliminated using more specific PCR primers or less likely to produce inter-sample differences. RT-qPCR allows RNA quantification because of an initial step of reverse transcription (conversion of an RNA into a complementary DNA). However, reverse transcription is less reproducible than PCR, generating harmful inter-sample differences both in diagnosis and in the research field. In this work, I propose innovative methods improving significantly different steps of these techniques. First, I propose an improved standardization of the reverse transcription step that can significantly reduce the variability between samples; using a constant volume of RNA extract for each sample substantially improves mRNA quantification accuracy. Second, I propose a primers modification by incorporating Locked Nucleic Acid residues (LNA); this modification increases the specificity of the primers to the limits of detection. Finally, I propose a simple and economical method for direct measurement of the melting temperature (Tm) in real PCR conditions. This essential parameter in the achievement of qPCR assays is generally obtained by predictive methods with low accuracy. Furthermore, this method should determine the thermodynamic parameters of the oligonucleotide sequence (∆G, ∆H and ∆S), on the one hand allowing access to the actual percentage of annealed primers according to the temperature and on the other hand, to identify the primers susceptibility in the presence of inhibitors.

Um novo circuito amplificador de 50 MHz - 1 GHz com alta linearidade para o padrão IEEE 802.22 “wireless regional area” (WRAN) é apresentado. Ele foi implementado sem nenhum indutor e oferece uma saída diferencial para ser utilizada como balun. Técnicas de cancelamento de ruído e aumento de linearidade foram usadas para melhorar a performace do amplificador de modo que eles pudessem ser otimizados separadamente. A linearidade foi melhorada utilizando transistores conectados como diodo. O amplificador foi implementado em um processo CMOS 130 nm, em uma área compacta de 136 m x 71 m. As simulações são apresentadas para esquemáticos pós-leiaute para duas classes diferentes de projeto: um visando a melhor linearidade e o outro a melhor Figura de Ruído (FR). Quando otimizado para melhor linearidade, os resultados de simulação atingem um ganho de tensão > 23.7 dB (ganho de potência > 19.1 dB), uma figura de ruído < 3.6 dB na banda inteira (com 2.4 dB min), um ponto de intersecção de terceira ordem (IIP3) > 3.3 dBm (7.6 dBm max) e um coeficiente de reflexão de entrada S11 < -16 dB. Quando otimizado para melhor figura de ruído, ele atinge um ganho de tensão > 24.7 dB (ganho de potência > 19.8 dB), uma FR < 2 dB na banda inteira, um IIP3 > -0.3 dBm e um S11 < -11 dB. Resultados de simulação Monte Carlo confirmam baixa sensibilidade à variabilidade de processo. Além disso, uma baixa sensibilidade com a temperatura na faixa de -55 até 125 C foi observada para Ganho, FR e S11. Consumo de potência é 17.6 mA sob fonte de alimentação de 1.2 V. / A new 50 MHz - 1 GHz low-noise amplifier circuit with high linearity for IEEE 802.22 wireless regional area network (WRAN) is presented. It was implemented without any inductor and offers a differential output for balun use. Noise cancelling and linearity boosting techniques were used to improve the amplifier performance in a way they can be separately optimized. Linearity was improved using diode-connected transistors. The amplifier was implemented in a 130 nm CMOS process in a compact 136 m x 71 m area. Simulations are presented for post-layout schematics for two classes of design: one for best linearity, another for best noise figure (NF). When optimized for best linearity, simulation results achieve a voltage gain > 23.7 dB (power gain > 19.1 dB), a NF < 3.6 dB over the entire band (with 2.4 dB min figure), an input third-order intercept point (IIP3) > 3.3 dBm (7.6 dBm max.) and an input power reflection coefficient S11 < -16 dB. When optimized for best NF, it achieves a voltage gain > 24.7 dB (power gain > 19.8 dB), a NF < 2 dB over the entire band, an IIP3 > -0.3 dBm and an S11 < -11 dB. Monte Carlo simulation results confirm low sensitivity to process variations. Also a low sensitivity to temperature within the range -55 to 125 C was observed for Gain, NF and S11. Power consumption is 17.6 mA under a 1.2 V supply.

Um novo circuito amplificador de 50 MHz - 1 GHz com alta linearidade para o padrão IEEE 802.22 “wireless regional area” (WRAN) é apresentado. Ele foi implementado sem nenhum indutor e oferece uma saída diferencial para ser utilizada como balun. Técnicas de cancelamento de ruído e aumento de linearidade foram usadas para melhorar a performace do amplificador de modo que eles pudessem ser otimizados separadamente. A linearidade foi melhorada utilizando transistores conectados como diodo. O amplificador foi implementado em um processo CMOS 130 nm, em uma área compacta de 136 m x 71 m. As simulações são apresentadas para esquemáticos pós-leiaute para duas classes diferentes de projeto: um visando a melhor linearidade e o outro a melhor Figura de Ruído (FR). Quando otimizado para melhor linearidade, os resultados de simulação atingem um ganho de tensão > 23.7 dB (ganho de potência > 19.1 dB), uma figura de ruído < 3.6 dB na banda inteira (com 2.4 dB min), um ponto de intersecção de terceira ordem (IIP3) > 3.3 dBm (7.6 dBm max) e um coeficiente de reflexão de entrada S11 < -16 dB. Quando otimizado para melhor figura de ruído, ele atinge um ganho de tensão > 24.7 dB (ganho de potência > 19.8 dB), uma FR < 2 dB na banda inteira, um IIP3 > -0.3 dBm e um S11 < -11 dB. Resultados de simulação Monte Carlo confirmam baixa sensibilidade à variabilidade de processo. Além disso, uma baixa sensibilidade com a temperatura na faixa de -55 até 125 C foi observada para Ganho, FR e S11. Consumo de potência é 17.6 mA sob fonte de alimentação de 1.2 V. / A new 50 MHz - 1 GHz low-noise amplifier circuit with high linearity for IEEE 802.22 wireless regional area network (WRAN) is presented. It was implemented without any inductor and offers a differential output for balun use. Noise cancelling and linearity boosting techniques were used to improve the amplifier performance in a way they can be separately optimized. Linearity was improved using diode-connected transistors. The amplifier was implemented in a 130 nm CMOS process in a compact 136 m x 71 m area. Simulations are presented for post-layout schematics for two classes of design: one for best linearity, another for best noise figure (NF). When optimized for best linearity, simulation results achieve a voltage gain > 23.7 dB (power gain > 19.1 dB), a NF < 3.6 dB over the entire band (with 2.4 dB min figure), an input third-order intercept point (IIP3) > 3.3 dBm (7.6 dBm max.) and an input power reflection coefficient S11 < -16 dB. When optimized for best NF, it achieves a voltage gain > 24.7 dB (power gain > 19.8 dB), a NF < 2 dB over the entire band, an IIP3 > -0.3 dBm and an S11 < -11 dB. Monte Carlo simulation results confirm low sensitivity to process variations. Also a low sensitivity to temperature within the range -55 to 125 C was observed for Gain, NF and S11. Power consumption is 17.6 mA under a 1.2 V supply.

Grace aux récents développements des technologies d’intégration, il est aujourd’hui possible d’envisager la réalisation de circuits et systèmes intégrés sur Silicium fonctionnant à des fréquences auparavant inatteignables. Par conséquence, depuis quelques années, on assiste à la naissance de nouvelles applications en bande millimétrique, comme la communication sans fil à haut-débit à 60GHz, les radars automobiles à 76-77 et 79-82GHz, et l’imagerie millimétrique à 94GHz.Cette thèse vise, en premier lieu, à la définition d’une méthodologie de conception des circuits intégrés en bande millimétrique. Elle est par la suite validée au travers de son application à la conception des amplificateurs faible-bruit en technologie BiCMOS SiGe. Dans ce contexte, une attention particulière a été portée au développement d’une stratégie de conception et de modélisation des inductances localisées. Plusieurs exemples d’amplificateurs faible-bruit ont été réalisés, à un ou deux étages, employant des composants inductifs localisés ou distribués, à 60, 80 et 94 GHz. Tous ces circuits présentent des caractéristiques au niveau de l’état de l’art dans le domaine, ainsi en confirmant l’exactitude de la méthodologie de conception et son efficacité sur toute la planche de fréquence considérée. En outre, la réalisation d’un récepteur intégré pour applications automobiles à 80GHz est aussi décrite comme exemple d’une possible application système, ainsi que la co-intégration d’un amplificateur faible-bruit avec une antenne patch millimétrique intégrée sur Silicium. / The interest towards millimeter waves has rapidly grown up during the last few years, leading to the development of a large number of potential applications in the millimeter wave band, such as WPANs and high data rate wireless communications at 60GHz, short and long range radar at 77-79GHz, and imaging systems at 94GHz.Furthermore, the high frequency performances of silicon active devices (bipolar and CMOS) have dramatically increased featuring both fT and fmax close or even higher than 200GHz. As a consequence, modern silicon technologies can now address the demand of low-cost and high-volume production of systems and circuits operating within the millimeter wave range. Nevertheless, millimeter wave design still requires special techniques and methodologies to overcome a large number of constraints which appear along with the augmentation of the operative frequency.The aim of this thesis is to define a design methodology for integrated circuits operating at millimeter wave and to provide an experimental validation of the methodology, as exhaustive as possible, focusing on the design of low noise amplifiers (LNAs) as a case of study.Several examples of LNAs, operating at 60, 80, and 94 GHz, have been realized. All the tested circuits exhibit performances in the state of art. In particular, a good agreement between measured data and post-layout simulations has been repeatedly observed, demonstrating the exactitude of the proposed design methodology and its reliability over the entire millimeter wave spectrum. A particular attention has been addressed to the implementation of inductors as lumped devices and – in order to evaluate the benefits of the lumped design – two versions of a single-stage 80GHz LNA have been realized using, respectively, distributed transmission lines and lumped inductors. The direct comparison of these circuits has proved that the two design approaches have the same potentialities. As a matter of fact, design based on lumped inductors instead of distributed elements is to be preferred, since it has the valuable advantage of a significant reduction of the circuit dimensions.Finally, the design of an 80GHz front-end and the co-integration of a LNA with an integrated antenna are also considered, opening the way to the implementation a fully integrated receiver.

Les travaux de thèse présentés se situent dans le contexte de la conception de circuits intégrés en technologie CMOS avancée pour des applications radiofréquence à très faible consommation de puissance. Les circuits sont conçus à travers deux concepts. Le premier est l'utilisation du coefficient d'inversion qui permet de normaliser le transistor en fonction de sa taille et de sa technologie, ceci permet une analyse rapide pour différentes performances visées ou différentes technologies. La deuxième approche est d'utiliser un facteur de mérite pour trouver la polarisation la plus adéquate d'un circuit en fonction de ses performances. Ces deux principes ont été utilisés pour définir des méthodes de conception efficaces pour deux blocs radiofréquence : l'amplificateur faible bruit et l'oscillateur. / Thesis work are presented in the context of the integrated circuits design in advanced CMOS technology for ultra low power RF applications. The circuits are designed around two concepts. The first is the use of the inversion coefficient to normalize the transistor as a function of its size and its technology, this allows a quick analysis for different performances or different technologies. The second approach is to use a figure of merit to find the most appropriate polarization of a circuit based on its performance. These two principles were used to define effective design methods for two RF blocks: low noise amplifier and oscillator.