[pt] A perda de percurso (PL) é um parâmetro essencial em modelos de propagação e crucial na determinação da área de cobertura de sistemas móveis. Os métodos de aprendizado de máquina (ML) tornaram-se ferramentas promissoras para a previsão de propagação de rádio. No entanto, ainda existem alguns desafios para sua implantação completa, relacionados à seleção das entradas mais significativas do modelo, à compreensão de suas contribuições para as previsões do modelo e à avaliação adicional da capacidade de generalização para amostras desconhecidas. Esta tese tem como objetivo projetar modelos de PL baseados em ML otimizados para diferentes aplicações das tecnologias 5G e além. Essas aplicações abrangem links de ondas milimétricas (mmWave) para ambientes indoor e outdoor na faixa de frequência de 26,5 a 40 GHz, cobertura de macrocélulas no espectro sub-6 GHz e comunicações veiculares usando campanhas de medições desenvolvidas em CETUC, Rio de Janeiro, Brazil. Vários algoritmos de ML são explorados, como redes neurais artificiais (ANN), regressão de vetor de suporte (SVR), floresta aleatória (RF) e aumento de árvore de gradiente (GTB). Além disso, estendemos dois modelos empíricos para mmWave com previsão de PL melhorada. Propomos uma metodologia para seleção robusta de modelos de ML e uma metodologia para selecionar os preditores mais adequados para as máquinas consideradas com base na melhoria de desempenho e na interpretabilidade do modelo. Além disso, para o canal veículo-veículo (V2V), uma técnica de rede neural convolucional (CNN) também é proposta usando uma abordagem de aprendizado por transferência para lidar com conjuntos de dados pequenos. Os testes de generalização propostos mostram a capacidade dos modelos de ML de aprender o padrão entre as entradas do modelo e a PL, mesmo em ambientes e cenários mais desafiadores de amostras desconhecidas. / [en] Path loss (PL) is an essential parameter in propagation models and critical in determining mobile systems’ coverage area. Machine learning (ML) methods have become promising tools for radio propagation prediction. However, there are still some challenges for its full deployment, concerning to selection of the most significant model s inputs, understanding their contributions to the model s predictions, and a further evaluation of the generalization capacity for unknown samples. This thesis aims to design optimized ML-based PL models for different applications for the 5G and beyond technologies. These applications encompass millimeter wave (mmWave) links for indoor and outdoor environments in the frequency band from 26.5 to 40 GHz, macrocell coverage in the sub-6 GHz spectrum, and vehicular communications using measurements campaign carried out by the Laboratory of Radio-propagation, CETUC, in Rio de Janeiro, Brazil. Several ML algorithms are exploited, such as artificial neural network (ANN), support vector regression (SVR), random forest (RF), and gradient tree boosting (GTB). Furthermore, we have extended two empirical models for mmWave with improved PL prediction. We proposes a methodology for robust ML model selection and a methodology to select the most suitable predictors for the machines considered based on performance improvement and the model’s interpretability. In adittion, for the vehicle-to-vehicle (V2V) channel, a convolutional neural network (CNN) technique is also proposed using a transfer learning approach to deal with small datasets. The generalization tests proposed shows the ability of the ML models to learn the pattern between the model’s inputs and PL, even in more challenging environments and scenarios of unknown samples.

[pt] O uso de dispositivos móveis cresceu exponencialmente nos últimos anos, tornando-se um desafio a melhoria da qualidade de sinal e o pleno atendimento a serviços que exigem altas taxas de dados e baixa latência. A fim de atender a crescente demanda de tráfego, as faixas de freqüências de ondas milimétricas começam a ser exploradas para aplicação à Quinta Geração de Comunicações Móveis. A partir de medições realizadas em ambientes fechados, nas faixas de freqüências de 28 GHz e 38 GHz, foi possível analisar o canal de comunicação com base nos modelos propagação para as faixas de freqüências de ondas milimétricas. / [en] Mobile usage has grown exponentially in recent years, making it a challenge to improve signal quality and fully support to services that require high data rates and low latency. In order to meet the growing traffic demand, millimeter-wave frequency bands are being explored for application to the 5th Generation of Mobile Communications. From measurements performed in indoor environments, in the 28 GHz and 38 GHz frequency range, it was possible to analyze the communication channel based on the propagation models for the millimeter wave frequency bands.

[pt] A perda de percurso (PL) é um parâmetro essencial em modelos de propagação e crucial na determinação da área de cobertura de sistemas móveis. Osmétodos de aprendizado de máquina (ML) tornaram-se ferramentas promissoras para a previsão de propagação de rádio. No entanto, ainda existem algunsdesafios para sua implantação completa, relacionados à seleção das entradasmais significativas do modelo, à compreensão de suas contribuições para asprevisões do modelo e à avaliação adicional da capacidade de generalizaçãopara amostras desconhecidas. Esta tese tem como objetivo projetar modelosde PL baseados em ML otimizados para diferentes aplicações das tecnologias5G e além. Essas aplicações abrangem links de ondas milimétricas (mmWave)para ambientes indoor e outdoor na faixa de frequência de 26,5 a 40 GHz,cobertura de macrocélulas no espectro sub-6 GHz e comunicações veicularesusando campanhas de medições desenvolvidas em CETUC, Rio de Janeiro,Brazil. Vários algoritmos de ML são explorados, como redes neurais artificiais(ANN), regressão de vetor de suporte (SVR), floresta aleatória (RF) e aumentode árvore de gradiente (GTB). Além disso, estendemos dois modelos empíricospara mmWave com previsão de PL melhorada. Propomos uma metodologiapara seleção robusta de modelos de ML e uma metodologia para selecionar ospreditores mais adequados para as máquinas consideradas com base na melhoria de desempenho e na interpretabilidade do modelo. Além disso, para o canalveículo-veículo (V2V), uma técnica de rede neural convolucional (CNN) também é proposta usando uma abordagem de aprendizado por transferência paralidar com conjuntos de dados pequenos. Os testes de generalização propostosmostram a capacidade dos modelos de ML de aprender o padrão entre as entradas do modelo e a PL, mesmo em ambientes e cenários mais desafiadoresde amostras desconhecidas. / [en] Path loss (PL) is an essential parameter in propagation models and critical in determining mobile systems coverage area. Machine learning (ML) methods have become promising tools for radio propagation prediction. However, there are still some challenges for its full deployment, concerning to selection of the most significant model s inputs, understanding their contributions to the model s predictions, and a further evaluation of the generalization capacity for unknown samples. This thesis aims to design optimized ML-based PL models for different applications for the 5G and beyond technologies. These applications encompass millimeter wave (mmWave) links for indoor and outdoor environments in the frequency band from 26.5 to 40 GHz, macrocell coverage in the sub-6 GHz spectrum, and vehicular communications using measurements campaign carried out by the Laboratory of Radio-propagation, CETUC, in Rio de Janeiro, Brazil. Several ML algorithms are exploited, such as artificial neural network (ANN), support vector regression (SVR), random forest (RF), and gradient tree boosting (GTB). Furthermore, we have extended two empirical models for mmWave with improved PL prediction. We proposes a methodology for robust ML model selection and a methodology to select the most suitable predictors for the machines considered based on performance improvement and the model s interpretability. In adittion, for the vehicle-to-vehicle (V2V) channel, a convolutional neural network (CNN) technique is also proposed using a transfer learning approach to deal with small datasets. The generalization tests proposed shows the ability of the ML models to learn the pattern between the model’s inputs and PL, even in more challenging environments and scenarios of unknown samples.

L'évolution des Réseaux Locaux Domestiques (RLD) est influencée par l'augmentation irréfrénée du nombre de terminaux connectés dans nos domiciles et par le déploiement de réseaux d'accès optiques haut débit qui délivrent désormais des services dépassant le Gigabit/s. Pour continuer à échanger efficacement les données, les RLD doivent évoluer vers le multi-Gigabit/s, et plus particulièrement la très appréciée connectivité sans fil. Les systèmes radio actuels ont des capacités limitées, néanmoins de nouveaux standards dont le débit excède 7Gbit/s voient le jour. Mais comme ces derniers exploitent la bande millimétrique, entre 57 et 66 GHz, leur couverture radio est limitée à la taille d'une pièce. En effet, à de telles fréquences, les pertes en espace libre sont très élevées et les ondes ne traversent pas les murs. Cette thèse propose de résoudre cette limitation au moyen de solutions Radio sur Fibre (RoF pour Radio over Fiber). Cette technique consiste à capturer les signaux radio émis dans une première pièce, puis à les convertir en signaux optiques pour les transmettre par fibre optique jusqu'à une seconde pièce où ils seront réémis. Plusieurs transducteurs RoF seront donc installés dans le domicile et interconnectés entre eux au moyen d'une infrastructure optique adaptée, créant ainsi des systèmes se comportant à la fois comme répéteurs and comme système de distribution. Pour le marché du RLD, de telles solutions sont compétitives que si elles sont bas coût. Par conséquent, ce travail se focalise sur la modulation directe avec détection directe (IM-DD) en Fréquence Intermédiaire (FI). En d'autres termes, le signal radio à 60GHz est translaté à une plus faible fréquence, autour de 5GHz, pour moduler directement un laser, puis il est restitué à 60GHz après la détection directe. Concernant la fibre optique, la multimode silice (MMF) est privilégiée puisqu'elle permet l'utilisation de composants optoélectroniques bas coût et largement disponible. Cette thèse propose différentes architectures RoF, de la point-à-point reliant deux pièces à la multipoint-à-multipoint agissant comme un bus logique. Après une caractérisation analogique des composants optoélectroniques, du lien RoF et du câble domestique, ces architectures sont construites et caractérisées étape par étape sur la base d'une modulation OFDM conforme aux standards radio 60GHz. Des transmissions temps réel entre terminaux commerciaux sont également réalisées afin de valider ces architectures. Pour finir, des infrastructures RoF avancées sont proposées. Premièrement, les systèmes RoF peuvent être améliorés si l'accès à leur couche optique est contrôlé par la couche MAC radio. Cette approche est donc étudiée, démontrant ainsi sa faisabilité. Deuxièmement, les systèmes optiques, perçus par les utilisateurs comme des produits premium, doivent supporter aussi bien les services de base du RLD que ceux qui vont émerger dans les années à venir. Ainsi, des infrastructures multiservice and multiformat innovantes transportant sur un unique câble optique des données IP, la télévision hertzienne et satellite, les signaux radio 60GHz et d'autres formats spécifiques tels que l'HDMI sont proposées puis testées / The evolution of the Home Area Network (HAN) is lead by the proliferation of connected devices inside the home and the deployment of high broadband access network which now allows the delivery of services that can exceed 1Gbit/s. To ensure efficient in-house exchanges, the HAN has to move rapidly toward multi-Gigabit/s connections, in particular the wireless connectivity generally preferred by the customers. Current wireless systems have limited capacities, but new radio standards delivering data-rates up to 7Gbit/s are emerging. Nevertheless, as they address the unlicensed millimeter-wave band, from 57 to 66 GHz, their radio coverage is limited to a single room. Indeed, at such frequencies, the free-space losses are high and the waves do not cross the walls. This thesis proposes to solve this problem by means of the Radio over Fiber (RoF) technology. This consists in capturing the 60GHz radio signals emitted in one room, converting them into optical signals for transmission through optical fibers, and reemitting them in another room. Thus, several RoF transducers will be installed in the home and interconnected by a suitable optical infrastructure to create systems acting at the same time as repeaters and as distribution systems. From the viewpoint of the HAN market, such systems will be competitive only if they are low cost. As a consequence, this work focuses on direct modulation with direct detection (IM-DD) at Intermediate Frequency (IF). In other words, the 60GHz signal is down-converted at a lower frequency around 5GHz before the laser modulation and up-converted to 60GHz after the photodetection. Concerning the optical fiber, silica multimode fiber (MMF) is privileged as it allows the use of low cost and largely available optoelectronic devices working at 850nm.The thesis proposes different RoF architectures, from point-to-point interconnecting two rooms to multipoint-to-multipoint acting as logical buses. After an analog characterization of the optoelectronic components, the RoF link and the domestic cable, these architectures are designed, built and characterized step by step using OFDM modulation according to existing wireless 60GHz standards. Real-time transmissions between commercial devices have also been performed to validate these architectures. Moreover, advanced RoF infrastructures are proposed. First, the RoF systems can be easily improved if the access to their optical media is managed by the radio MAC layer. This approach is therefore studied showing its feasibility. Secondly, an optical system, seen by users as a premium product, has to support the legacy home services commonly used as well as the new ones that could emerge in the future. Thus, innovative multiservice and multiformat infrastructures conveying on a unique optical cable wired IP data, broadcast terrestrial or satellite television, the 60GHz wireless connectivity, and specific formats as HDMI signals are proposed and tested

Le travail de cette thèse a consisté à concevoir, réaliser et optimiser un montage expérimental de Polarisation Dynamique Nucléaire multi-échantillons pour des applications biomédicales en Imagerie Spectroscopique par Résonance Magnétique. Ce montage est constitué d'un aimant à fort champ magnétique (3,35T), dans lequel se place un système cryogénique à bain d'hélium (He4) liquide pompé pouvant atteindre des températures inférieures à 1,2K. Un ensemble d'inserts permet d'effectuer les différentes étapes du processus PDN dont l'irradiation des échantillons par un champ micro-onde (f=94GHz et P=50mW) et le suivi de leur polarisation par Résonance Magnétique Nucléaire. Ce système permet de polariser jusqu'à trois échantillons, de volume proche de 1mL, à des taux de polarisation de quelques pourcents. Il présente une forte autonomie supérieure à quatre heures, autorisant ainsi la polarisation de molécules à longues constantes de temps de polarisation. La possibilité de disposer quasi-simultanément, après dissolution, de plusieurs échantillons fortement polarisés ouvre la voie à de nouvelles applications dans le domaine de l'imagerie biomédicale / The aim of this thesis work was to design, build and optimize a large volume multisamples DNP (Dynamic Nuclear Polarization) polarizer dedicated to Magnetic Resonance Spectroscopic Imaging applications. The experimental system is made up of a high magnetic field magnet (3,35T) in which takes place a cryogenic system with a pumped bath of liquid helium (4He) allowing temperatures lower than 1,2K. A set of inserts is used for the different steps of DNP : irradiation of the sample by a microwave field (f=94GHz and P=50mW), polarization measurement by Nuclear Magnetic Resonance. . . With this system, up to three samples of 1mL volume can be polarized to a rate of few percents. The system has a long autonomy of four hours, so it can be used for polarizing molecules with a long time constant of polarization. Finally, the possibility to get quasisimultaneously, after dissolution, several samples with a high rate of polarization opens the way of new applications in biomedical imaging

La focalisation du champ électromagnétique dans les zones de champ proche et lointain est un sujet de forte actualité pour l'imagerie médicale et la radiométrie des microondes jusqu'aux ondes millimétriques. Dans ce cadre, la théorie des ondes de fuite est un formalisme élégant qui permet de décrire d'une même façon les problèmes radiatifs en champ proche et en champ lointain des microondes aux fréquences optiques. Dans cette thèse, on utilise la polyvalence de la théorie des ondes de fuite pour développer des systèmes rayonnants complexes afin de contrôler les caractéristiques radiatives en champ lointain aux fréquences submillimétriques et pour focaliser la radiation électromagnétique en champ proche aux fréquences millimétriques. Ainsi, l'utilisation de matériaux uniques comme le graphène et les cristaux liquides ont été considérés pour la conception des antennes à ondes de fuite, en obtenant des résultats très intéressants en termes de reconfigurabilité, d'efficience et de directivité. Dans ce contexte, une analyse théorique originale a fourni de nouvelles formules pour l'évaluation des caractéristiques radiatives (c.à.d. la largeur de faisceau, le niveau des lobes secondaires, etc.) des antennes à ondes de fuite. En effet, la largeur du faisceau de ces antennes est, jusqu'à présent, estimée au moyen des formules proposées pour la première fois dans les années '60 par Prof. Arthur A. OLINER. Ces formules ne tiennent en compte ni de la longueur de l'antenne (sauf pour des cas très particuliers), ni du rayonnement longitudinal, elles ne permettent donc pas une évaluation rigoureuse.En complément à la reconfigurabilité en champ lointain, les ondes de fuite offrent aussi la possibilité de focaliser la radiation en champ proche. Dans ce cas, on voit que les ondes de fuite peuvent être utilisées d'une façon efficace pour générer des faisceaux non diffractifs de Bessel à travers des systèmes rayonnants à bande étroite aux ondes millimétriques. De plus, le caractère non diffractif des faisceaux de Bessel peut aussi être utilisé pour générer des impulsions très localisées (comme les solitons en optiques) à travers la superposition continue des faisceaux de Bessel sur une large bande de fréquence. Dans ce cadre, une nouvelle formulation a été développée afin de comprendre les limitations physiques et technologiques concernant la génération des impulsions non diffractives et non dispersives, c.à.d. les X-waves. Les résultats ont montré qu'un type de systèmes rayonnants à large bande, notamment les antennes RLSA (en anglais « Radial Line Slot Array »), semblent très favorables pour la génération des X-waves. / In recent years, microwave, millimeter-wave, and THz applications such as medical and security imaging, wireless power transfer, and near-field focusing, just to mention but a few, have gained much attention in the area of ICT due to their potentially high social impact. On one hand, the need of highly-directive THz sensors with tunable radiating features in the far-field region has recently boosted the research activity in the design of flexible, low-cost and low-profile devices. On the other hand, it is of paramount importance to focus energy in the near-field region, and thus the generation of limited-diffraction waves in the microwave and millimeter-wave regime is a topic of recent increasing interest. In this context, leaky-wave theory is an elegant and extremely useful formalism which allows for describing in a common fashion guiding and radiating phenomena in both the near field and the far field, spanning frequencies from microwaves to optics passing through THz. In this PhD thesis we aim to exploit the intrinsic versatility of the leakywave approach to design advanced radiating systems for controlling the far-field radiating features at THz frequencies and for focusing electromagnetic radiation in the near field at millimeter waves. Specifically, the use of relatively new materials such as graphene and liquid crystals has been considered for the design of leaky-wave based radiators, achieving very promising results in terms of reconfigurability, efficiency, and radiating capabilities. In this context, an original theoretical analysis has provided new general formulas for the evaluation of the radiating features (e.g., half-power beamwidth, sidelobe level, etc.) of leaky-wave antennas. Indeed, the current formulations are based on several simplifying hypotheses which do not allow for an accurate evaluation of the beamwidth in different situations. In addition to the intriguing reconfigurable capabilities offered by leaky waves in far-field applications, interesting focusing capabilities can be obtained in the near field. In particular, it is shown that leaky waves can profitably be used to generate limited-diffraction Bessel beams by means of narrow-band radiators in the microwave range. Also, the use of higherorder leaky-wave modes allows for achieving almost the same performance in the millimeter-wave range, where previous designs were subjected to severe fabrication issues. Even more interestingly, the limited-diffractive character of Bessel beams can also be used to generate limited-diffraction pulses as superpositions of monochromatic Bessel beams over a considerable fractional bandwidth. In this context, a novel theoretical framework has been developed to understand the practical limitations to efficiently generate limited-diffraction, limited-dispersion pulses, such as X-waves, in the microwave/millimeter-wave range. As a result of this investigation, a class of wideband radiators has been thoroughly analyzed, showing promising capabilities for the generation of both zeroth-order and higher-order Xwaves. The latter may pave the way for the first localized transmission of orbital angular momentum in the microwave range.

[ES] Las redes inalámbricas interiores son una parte integral de las comunicaciones actuales. Debido a la alta demanda de confiabilidad y la enorme densidad de redes interiores, algunas de ellas están migrando hacia las frecuencias de onda milimétrica (mmW), donde están disponibles mayor ancho de banda y rendimiento. Diseñar la red para la banda mmW ofrece soluciones a algunos problemas presentes en frecuencias m 'as bajas pero también plantea nuevos desafíos. Esta Tesis se centra en particular en casos de uso de conferencias en entornos grandes. La principal funcionalidad de la red de conferencias es realizar conferencias. Principalmente, esto significa transmitir señales de audio desde los participantes y fuentes de audio a otros participantes. Por lo tanto, se presenta el conjunto de requisitos para el diseño y despliegue de la red resultantes de los estándares de la conferencia, tipos de datos, estructura de los datos transmitidos, propagación en ambientes interiores y fenómenos físicos relacionados con la longitud de onda. Las soluciones de conferencias inalámbricas existentes se proporcionan en frecuencias Wi-Fi, es decir, 2,4 GHz y 5 GHz. Sin embargo, su objetivo no es conectar a los participantes de la conferencia a Internet, sino crear una red que cumpla con los requisitos de la conferencia basada en la funcionalidad proporcionada en la familia de estándares 802.11. La enmienda 802.11ay se proporciona para la comunicación en la banda de frecuencia de 60 GHz. Para permitir una fácil transición de las redes de conferencias inalámbricas hacia mmW, se estudia la enmienda 802.11ay y el canal de radio de 60 GHz. Se considera que el estándar es inadecuado para el tipo y la cantidad de datos enviados en las redes de conferencias inalámbricas. Está diseñado para trabajar con grandes cantidades de datos, mientras que los datos principales en las redes de conferencias son audio en tiempo real. Enviar tramas de datos con esa poca información introduce una sobrecarga significativa debido a los datos del pre 'ambulo en PPDU. Sin embargo, se considera que algunas de sus características merecen una mayor exploración. Al estudiar el caso de uso de la conferencia, es importante centrarse en las configuraciones dedicadas a las conferencias, es decir, la influencia de las propiedades de los materiales interiores, la forma y el tamaño del entorno, y el tipo y ubicación de la antena en la propagación. Para el canal de radio de 60 GHz se estudia el efecto multitrayecto, lo que lleva a la creación del modelo de predicción RMS DS. Además, se muestra que la directividad en esta banda proporciona una gran ventaja si se utiliza la transmisión LoS o si el reflector conocido ayuda a dirigir el haz en la dirección de receptor (RX). Además, esta Tesis también se centra en frecuencias distintas a 60 GHz, precisamente 27 GHz y 28 GHz, las frecuencias introducidas en el Rango de Frecuencia 2 (FR2) 5G NR. Para esta Tesis se realizaron varias campañas de medición en diferentes ambientes, como el auditorio, el aula grande, la sala de reuniones y la sala de seminarios. Los resultados resaltan la importancia de conocer el entorno para optimizar el despliegue de la red. El ray tracing (RT) se utiliza para profundizar la comprensión del comportamiento de los canales de radio medidos. Se identifica el origen de los componentes de trayectorias múltiples y los objetos que son reflectores de mayor impacto. Además, las simulaciones de entornos disponibles se realizan para frecuencias, configuraciones de antenas y configuraciones de mobiliario distintas a las de las mediciones realizadas. Se midieron e introdujeron en el simulador RT los valores de las pérdidas de transmisión y reflexión introducidas por objetos presentes en entornos grandes en frecuencias de 26-40 GHz y 55-65 GHz para permitir resultados m 'as precisos de las simulaciones. La comparación entre mediciones y simulaciones muestra la utilidad de las simulaciones RT en el análisis de canales de radio. / [CA] Les xarxes sense fil d'interior són una part integral de les comunicacions actuals. A causa de l'alta demanda de fiabilitat i la gran densitat de xarxes interiors, algunes d'elles migren cap a les freqüències de les ones mil·límetres (mmW), on hi ha una amplada de banda i un rendiment més grans disponibles. Dissenyar la xarxa per a la banda mmW ofereix solucions a alguns problemes presents en freqüències més baixes però també planteja nous reptes. Aquesta tesi se centra especialment en casos d'ús de conferències en entorns grans. La principal funcionalitat de la xarxa de conferències és la realització de conferències. Principalment, això significa transmetre senyals d'àudio dels participants i fonts d'àudio a altres participants. Per tant, es presenta el conjunt de requisits per al disseny i desplegament de la xarxa resultant dels estàndards de conferència, els tipus de dades, l'estructura de les dades transmeses, la propagació en entorns interiors i els fenòmens físics relacionats amb la longitud d'ona. Les solucions de conferències sense fil existents es proporcionen en freqüències Wi-Fi, és a dir, 2,4 GHz i 5 GHz. Tanmateix, no pretenen connectar els participants de la conferència a Internet, sinó crear una xarxa que compleixi els requisits de la conferència basada en la funcionalitat proporcionada a la família d'estàndards 802.11. L'esmena 802.11ay es proporciona per a la comunicació a la banda de freqüència de 60 GHz. Per permetre una fàcil transició de les xarxes de conferències sense fil cap al mmW, s'estudia l'esmena 802.11ay i el canal de ràdio de 60 GHz. Es considera que l'estàndard és inadequat per al tipus i la quantitat de dades enviades a les xarxes de conferències sense fil. Està dissenyat per treballar amb grans quantitats de dades, mentre que les dades principals de les xarxes de conferències són àudio en temps real. L'enviament de trames de dades amb aquesta poca informació introdueix una sobrecàrrega important a causa de les dades del preàmbul a PPDU. No obstant això, es considera que algunes de les seves característiques mereixen una exploració més detallada. Mentre s'estudia el cas d'ús de la conferència, és important centrar-se en les configuracions dedicades a les conferències, és a dir, la influència de les propietats dels materials interiors, la forma i la mida de l'entorn, i el tipus i col·locació d'antena en la propagació. Per al canal de ràdio de 60 GHz, s'estudia l'efecte multicamí, que condueix a la creació del model de predicció RMS DS. A més, es demostra que la directivitat en aquesta banda dóna un gran avantatge si s'utilitza la transmissió LoS o si el reflector conegut ajuda a dirigir el feix en la direcció de receptor (RX). A més, aquesta Tesi també se centra en freqüències diferents de 60 GHz, precisament 27 GHz i 28 GHz, les freqüències introduïdes en el rang de freqüències 2 (FR2) 5G NR. Per a aquesta Tesi s'han realitzat diverses campanyes de mesura en diferents entorns, com ara l'auditori, l'aula gran, la sala de reunions i la sala de seminaris. Els resultats posen de manifest la importància de conèixer l'entorn per tal d'optimitzar el desplegament de la xarxa. El ray tracing (RT) s'utilitza per aprofundir en la comprensió del comportament dels canals de ràdio mesurats. S'identifica l'origen dels components multipath i els objectes que són reflectors més impactants. A més, es realitzen simulacions dels entorns disponibles per a freqüències, configuracions d'antenes i configuracions de mobles diferents de les de les mesures realitzades. Es van mesurar els valors de les pèrdues de transmissió i reflexió introduïdes per objectes presents en entorns grans en freqüències de 26-40 GHz i 55-65 GHz i es van introduir al simulador RT per permetre els resultats més precisos de les simulacions. La comparació entre mesures i simulacions mostra la utilitat de les simulacions RT en l'anàlisi de canals de ràdio. / [EN] Indoor wireless networks are an integral part of today's communications. Due to the high demand for reliability and the huge density of indoor networks, some of them are migrating towards the Millimeter Wave (mmW) frequencies, where larger bandwidth and throughput are available. Designing the network for the mmW band offers solutions to some problems present in lower frequencies but also poses new challenges. This Thesis focuses in particular on conference use cases in large environments. The main functionality of the conference network is to conduct conferences. Primarily, this means transmitting audio signals from participants and audio sources to other participants. Therefore, the set of requirements for the network design and deployment resulting from conference standards, data types, structure of transmitted data, propagation in indoor environments, and physical phenomena related to the wavelength are presented. Existing wireless conference solutions are provided in Wi-Fi frequencies, that is 2.4 GHz and 5 GHz. However, they do not aim to connect conference participants to the Internet but to create a network fulfilling the conference requirements based on the functionality provided in 802.11 standards family. The 802.11ay amendment is provided for communication in the 60 GHz frequency band. To allow an easy transition of wireless conference networks towards the mmW, the 802.11ay amendment and the 60 GHz radio channel are studied. The standard is found to be inadequate for the type and amount of data sent in the wireless conference networks. It is designed to work with large amounts of data, whereas the primary data in conference networks is real-time audio. Sending data frames with that little information introduces significant overhead due to the preamble data in Physical Protocol Data Unit (PPDU). Nevertheless, some of its features are considered worth further exploration. While studying the conference use case, it is important to focus on setups dedicated to the conferences, that is, the influence of indoor materials properties, the environment's shape and size, and antenna type and placement on the propagation. For the 60 GHz radio channel, the multipath effect is studied, which leads to the creation of the Root Mean Square Delay Spread (RMS DS) prediction model. Moreover, it is shown that directivity in this band gives a great advantage if the Line of Sight (LoS) transmission is used or the known reflector helps to steer the beam into the direction of the receiver. Furthermore, this Thesis also focuses on frequencies other than 60 GHz, precisely 27 GHz and 28 GHz, the frequencies introduced in Frequency Range 2 (FR2) Fifth Generation (5G) New Radio (NR). Several measurement campaigns were performed for this Thesis in different environments, such as the auditory, the large classroom, the meeting room, and the seminar room. The results highlight the importance of knowing the environment in order to optimize network deployment. The Ray Tracing (RT) is used to deepen the understanding of the measured radio channels' behaviour. The origin of multipath components and the objects that are most impactful reflectors are identified. Moreover, the simulations of available environments are performed for frequencies, antenna setups, and furniture configurations other than those in performed measurements. Values of transmission and reflection losses introduced by objects present in large environments in frequencies 26-40 GHz and 55-65 GHz were measured and introduced into the RT simulator to allow the most precise results of simulations. Comparison between measurements and simulations shows the usefulness of RT simulations in radio channel analysis. / This thesis was partly founded by the H2020 Marie Sklodowska-Curie Actions Program within Project Grant No. 766231 WAVECOMBE - ITN - 2017. Measurement campaigns in Gdansk University of Technology were partially funded by the COST Action CA20120, ”Intelligence-Enabling Radio Commu- nications for Seamless Inclusive Interactions” (INTERACT) under the Short Term Scientific Mission No. CA20120-2146922e. / Drozdowska, MK. (2024). Intelligent Design of mmWave Conference Networks in Built Environments [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/211182