In this thesis, we have established an optical measurement system to measure the device characteristics. We focus on the investigation of semiconductor optical amplifier, Fabry-Perot laser, and ring cavity filter. We used InP-based multiple quantum wells epitaxial wafer with modulation doping in the active layer. A 1.41 £gm symmetric InGaAlAs/InP quantum well structure is used to fabricate the optical waveguide ring resonator devices for the optical communication region at 1.55£gm wavelength. For the semiconductor optical amplifier and lasers, we designed two different types: Fabry-Perot Amplifier (FPA), and Traveling Wave Amplifier (TWA). The InGaAlAs-FPA structure has three lasing peaks at 1514 nm, 1528 nm, and 1544 nm. The InGaAlAs-TWA-a structure has only one peak at 1510 nm. The InGaAsP-TWA-b structure has a gross gain = 8.5 dB (wavelength = 1575 nm) at pumping current = 22 mA. We used Hakki-Paoli method and transparency current to calculate gain spectrum. For ring cavity filter, the optical spectrum has a FSR = 41.25 GHz.

Bandpass filters are an essential component of wireless communication systems that only transmits frequencies corresponding to the communication band and rejects all other frequencies. As the deployment of 5G draws nearer, first deployments are expected in 2020 [1], the need for viable filters at the new frequency bands becomes more imminent. Size and performance are two critical considerations for a filter that will be used in emerging mobile communication applications. The high frequency of 5G communication, 28 GHz as opposed to sub 6 GHz for nearly all previous communication protocols, means that previously utilized lumped component based solutions cannot be implemented since they are ill-suited for mm-wave applications. The focus of this work is the miniaturization of a high-performance filter. The Substrate Integrated Waveguide (SIW) is a high performance and promising structure and Low Temperature Co-Fired Ceramic (LTCC) is a high-performance material that both can operate at higher frequencies than the technologies used for previous telecommunication generations. To miniaturize the structure, a compact folded four-cavity SIW filter is designed, implemented and tested. The feeding structure is integrated into the filter to exploit the System on Package (SoP) attributes of LTCC and further reduce the total area of the filter individually and holistically when looking at the final integrated system. Two unique three dimensional (3D) integrated SoP LTCC two-stage SIW single cavity filters and one unique four-cavity filter all with embedded planar resonators are designed, fabricated and tested. The embedded resonators create a two-stage effect in a single cavity filter. The better single cavity design provides a 15% fractional bandwidth at a center frequency of 28.12 GHz, and with an insertion loss of -0.53 dB. The fabricated four-cavity filter has a 3-dB bandwidth of .98GHz centered at 27.465 GHz, and with an insertion loss of -2.66 dB. The designs presented highlight some of the previously leveraged advantages of SoP designs while also including additions of embedded planar resonators to feed the SIW cavity. The integration of both elements realizes a compact and high-performance filter that is well suited for future mm-wave applications including 5G.

Les travaux réalisés lors de cette thèse ont pour objectif la fabrication de filtres accordables fort Q0. Pour cela nous nous sommes naturellement orienté vers des structures volumiques puisque elles permettent l’obtention de Q0 important. Nous avons également utilisé des éléments d’accord permettant de maximiser ce Q0, tels que des MEMS-RF ou bien encore des patchs de Matériaux à Transition de Phase (PTM) comme le dioxyde de vanadium (VO2). Nous avons démontré lors de notre étude que ces dispositifs garantissaient des Q0 supérieurs à 1000 sur une large plage d’accord. / This work is based on the fabrication of High-Q tunable filters. We looked at cavity filters since they allow us to achieve a high Q0. We also used RF-MEMS or vandium Dioxyde as tuning element. Since they provide a high-Q tuning. We have demonstrated Q0 better than one thousand over a wide tuning range.

In this Master's thesis the option of using deep reinforcement learning for cavity filter tuning has been explored. Several reinforcement learning algorithms have been explained and discussed, and then the deep deterministic policy gradient algorithm has been used to solve a simulated filter tuning problem. Both the filter environment and the reinforcement learning agent were implemented, with the filter environment making use of existing circuit models. The reinforcement learning agent learned how to tune filters with four poles and one transmission zero, or eight tune-able screws in total. A comparison was also made between constant exploration noise and exploration noise decaying over time, together with different maximum lengths of the episodes. For the particular noise used here, decaying exploration noise was shown to be better than constant, and a maximum length of 100 steps was shown to be better than 200 for the 8 screw filter.

This thesis deals with the analysis of coaxial cavity filters and synthesis of generic topology of these filters. Transfer function comes out the synthesis, which is based on low-pass prototype and similarity with real structure. The next aim of this thesis is to build the basic software for coaxial cavity filters synthesis from engaged characteristic parameters of the filter. This work also includes the analysis of coupling matrix synthesis based on transfer function, where we can use direct coupling or cross-coupling between resonators. The last point is focused on the diplexer filter based on specified parameters and obtained coupling matrix. Entire diplexer model is optimized using Tuning-Space Mapping method.

Dispositivos fotônicos têm estado continuamente no foco das pesquisas científicas, particularmente em aplicações para comunicações ópticas e sensoriamento. Por outro lado, as dimensões desses dispositivos são restringidas pelo limite de difração de Abbe. Esse limite tem se mostrado como o grande gargalo no desenvolvimento de novas tecnologias em microscopia óptica, litografia de projeção óptica, óptica integrada, e armazenamento óptico de dados, por limitar as dimensões e a capacidade de integração destes dispositivos. Felizmente, a \"plasmônica\" surgiu como um novo campo de estudo, possibilitando a superação dessa limitação por meio da propagação da luz em modos de plasmon-poláritons de superfície - SPP (Surface Plasmon Polariton). De maneira simplificada, SPPs são campos eletromagnéticos confinados em regiões menores que o comprimento de onda da luz. A geração de SPP ocorre por meio da excitação coletiva de elétrons na interface entre dois meios, metal-dielétrico, que se acoplam com a onda eletromagnética incidente. Pesquisadores logo perceberam que guias de onda baseados em SPP poderiam transportar a mesma banda de informação que um dispositivo fotônico convencional e serem tão localizados quanto dispositivos eletrônicos (elétrons têm maior capacidade de confinamento que fótons). Dessa maneira, alterando a estrutura da superfície de um metal, as propriedades dos SPPs - em particular sua interação com a luz - podem ser manipuladas, oferecendo potencial para o desenvolvimento de novos tipos de dispositivos fotônicos. Com isso, nanoestruturas capazes de guiar, dividir ou mesmo sintonizar a luz tornaram-se realidade. No presente trabalho, o fenômeno de geração de SPPs é estudado teoricamente e aplicado na modelagem de diversas estruturas de interesse científico e tecnológico, tais como filtros de cavidade ressonante e ressoadores em anel. O objetivo principal é a obtenção de estruturas capazes de filtrar ou sintonizar comprimentos de onda, minimizando as perdas ao máximo. Com isso, espera-se estender e explorar ainda mais o leque de possíveis aplicações. / Photonic devices have continuously been in the focus of scientific research, particularly for optical communications and sensing applications. On the other hand, the dimensions of these devices are well known to be limited by the Abbe\'s diffraction limit. This limit has been the major bottleneck in developing new technologies in optical microscopy, lithography projection optics, integrated optics, and optical data storage, as it limits the size and ability to integrate these devices. Fortunately, the field of \"Plasmonics\" has emerged and devices whose dimensions overcome the difraction limit have now become reality. This is possible with the propagation of light in the form of Surface Plasmon Polariton - SPP that, in a simplified way, is an electromagnetic field confined in regions smaller than the wavelength of light. SPP occurs via collective excitation of electrons at the interface between two media, metal-dielectric, as a result of the coupling with an incident electromagnetic wave. Researchers soon realized that waveguides based on SPP could carry the same band of information as that of a conventional photonic device and yet be as localized as electronic devices (electrons have a greater capacity for confinement than photons). Thus, changing the structure of the surface of a metal, the properties of SPPs - in particular its interaction with light - can be manipulated, offering potential for the development of new types of photonic devices. Thus, nanostructures capable of transferring, guiding, splitting, or even tuning the light have now become reality. In this work, the phenomenon of generation of SPPs is theoretically investigated and applied to various structures of scientific and technological interest, such as filters and cavity resonators. The main objective is to obtain structures that are able to filter or tune wavelengths, minimizing losses as much as possible. As a result, we expect to extend and explore even further the range of possible applications.

Dispositivos fotônicos têm estado continuamente no foco das pesquisas científicas, particularmente em aplicações para comunicações ópticas e sensoriamento. Por outro lado, as dimensões desses dispositivos são restringidas pelo limite de difração de Abbe. Esse limite tem se mostrado como o grande gargalo no desenvolvimento de novas tecnologias em microscopia óptica, litografia de projeção óptica, óptica integrada, e armazenamento óptico de dados, por limitar as dimensões e a capacidade de integração destes dispositivos. Felizmente, a \"plasmônica\" surgiu como um novo campo de estudo, possibilitando a superação dessa limitação por meio da propagação da luz em modos de plasmon-poláritons de superfície - SPP (Surface Plasmon Polariton). De maneira simplificada, SPPs são campos eletromagnéticos confinados em regiões menores que o comprimento de onda da luz. A geração de SPP ocorre por meio da excitação coletiva de elétrons na interface entre dois meios, metal-dielétrico, que se acoplam com a onda eletromagnética incidente. Pesquisadores logo perceberam que guias de onda baseados em SPP poderiam transportar a mesma banda de informação que um dispositivo fotônico convencional e serem tão localizados quanto dispositivos eletrônicos (elétrons têm maior capacidade de confinamento que fótons). Dessa maneira, alterando a estrutura da superfície de um metal, as propriedades dos SPPs - em particular sua interação com a luz - podem ser manipuladas, oferecendo potencial para o desenvolvimento de novos tipos de dispositivos fotônicos. Com isso, nanoestruturas capazes de guiar, dividir ou mesmo sintonizar a luz tornaram-se realidade. No presente trabalho, o fenômeno de geração de SPPs é estudado teoricamente e aplicado na modelagem de diversas estruturas de interesse científico e tecnológico, tais como filtros de cavidade ressonante e ressoadores em anel. O objetivo principal é a obtenção de estruturas capazes de filtrar ou sintonizar comprimentos de onda, minimizando as perdas ao máximo. Com isso, espera-se estender e explorar ainda mais o leque de possíveis aplicações. / Photonic devices have continuously been in the focus of scientific research, particularly for optical communications and sensing applications. On the other hand, the dimensions of these devices are well known to be limited by the Abbe\'s diffraction limit. This limit has been the major bottleneck in developing new technologies in optical microscopy, lithography projection optics, integrated optics, and optical data storage, as it limits the size and ability to integrate these devices. Fortunately, the field of \"Plasmonics\" has emerged and devices whose dimensions overcome the difraction limit have now become reality. This is possible with the propagation of light in the form of Surface Plasmon Polariton - SPP that, in a simplified way, is an electromagnetic field confined in regions smaller than the wavelength of light. SPP occurs via collective excitation of electrons at the interface between two media, metal-dielectric, as a result of the coupling with an incident electromagnetic wave. Researchers soon realized that waveguides based on SPP could carry the same band of information as that of a conventional photonic device and yet be as localized as electronic devices (electrons have a greater capacity for confinement than photons). Thus, changing the structure of the surface of a metal, the properties of SPPs - in particular its interaction with light - can be manipulated, offering potential for the development of new types of photonic devices. Thus, nanostructures capable of transferring, guiding, splitting, or even tuning the light have now become reality. In this work, the phenomenon of generation of SPPs is theoretically investigated and applied to various structures of scientific and technological interest, such as filters and cavity resonators. The main objective is to obtain structures that are able to filter or tune wavelengths, minimizing losses as much as possible. As a result, we expect to extend and explore even further the range of possible applications.

This thesis deals with optimization of coaxial cavity filters. It thoroughly describes optimization method Tuning – Space Mapping (TSM) in CST Studio. It focuses on the explanation of the fine and the coarse model and their mutual link. The whole algorithm is illustrated and tested on the third-order coaxial cavity filter working in the band of 880 – 960MHz. In order to check the function of TSM, the algorithm is applied on fourth-order filter, triplet filter and quadruplet filter. Another part of the thesis focuses on the automatization of the whole optimization procedure. Macros for automatic determination of calibration constants and automatic calibration process between the coarse and the fine model are proposed and programmed in CST. The complete optimization procedure is applied on optimization of seventh-order filter with two cross couplings.

As telecommunication continues developing, the demand for mass production of well calibrated Base Transceiver Stations (BTS) components increases. Cavity Filters are an essential piece of every BTS; however, manufacturing tolerances often lead to detuned filters which require costly post-production fine tuning. Model Free Reinforcement Learning has been proposed to automate this process; however agents are not sample efficient. This is especially problematic, as agent training with newer, more precise environment simulators is time demanding. This work aims to leverage Model Based Reinforcement Learning to improve sample efficiency, while maintaining the same degree of accuracy. To this end, we evaluate and improve upon the performance of three state-of-the-art methods, present in the literature. The proposed modifications on these methods can serve as a template for their application on other, high dimensional non image data problems. In particular, the proposed modification on the Dreamer is modular, improves training stability and greatly decreases sample complexity. More specifically, sample complexity was reduced by a factor of 4 for the 6p2z filter and by a factor of 10 for 8p4z. Furthermore, hyperparameter sensitivity analysis is provided to add extra insight behind each approach. Overall, results facilitate further research in this field. The reduced sample complexity opens the possibility of training on more accurate simulators of more complicated filters, which would previously be intractable due to the high amount of samples required. / Moderna mobilnät är uppbyggda av massproducerade basstationer (Base Tranciever Stations), som var och en innehåller ett antal kavitetsfilter. Dessa filter är mycket känsliga, vilket gör att de efter produktion behöver finjusteras manuellt för att fungera som avsett. För att automatisera denna process har man tidigare använt Model Free Reinforcement Learning (MFRL). Denna teknik kräver dock mycket beräkningar, vilket är problematiskt, eftersom man skulle vilja genomföra träningen med mer komplexa simuleringsmodeller, vilket inte går i dagsläget. I detta arbete skall vi undersöka möjligheten att använda Model Based Reinforcement Learning (MBRL) för att lösa samma problem med färre beräkningssteg. Vi utvärderar, och anpassar, därför tre befintliga MBRL-algoritmer till problemet. Dessa anpassningar kan även överföras till andra problem. Den anpassning som görs på Dreamer-algoritmen är modulär, förbättrar stabiliteten i träningen, och minskar antalet beräkningar som behövs. I detalj så minskade antalet beräkningar med en faktor 4 för ett så-kallat 6p2z-filter och en faktor 10 för ett 8p4z-filter. En känslighetsanalys vad gäller hyperparametrar har också gjorts för varje metod. Rapportens resultat kan användas i vidare forskning på så sätt att det minskade antalet beräkningar gör att man kan använda mer realistiska modeller, av mer komplicerade filter, på ett sätt som tidigare inte var möjligt.

Les travaux présentés dans cette thèse concernent l’étude et la fabrication de composants passifs hyperfréquence possédant un facteur de qualité important sur les bandes Q et V. Pour cela, nous nous sommes orientés vers la fabrication de composants volumiques à air. Des lignes coaxiales à section rectangulaire à air ainsi que des filtres à cavité métalliques à air ont été réalisés. Aussi, ces travaux consistent à définir une méthode de fabrication permettant d’obtenir les composants cités précédemment. Plusieurs techniques de micro-fabrication additives ont ainsi été établies afin de répondre à la problématique. / This work concerns the study and the fabrication of passive RF components with a high quality factor on Q/V bands. The components were fabricated with an air-filled 3D architecture. Air-filled rectangular micro-coaxial lines and air filled cavity filters were demonstrated. Also, micro-additive fabrication processes were demonstrated for the realization of the 3D air-filled components.