[pt] Esta tese apresenta novas estruturas e algoritmos para detecção multiusuário e supressão de interferência em sistemas DS-CDMA. São investigadas estruturas baseadas em redes neurais recorrentes para projeto de receptores com decisão realimentada e desenvolvidos algoritmos adaptativos para combater a interferência de múltiplo acesso e a interferência entre símbolos. Novos algoritmos baseados na minimização da taxa de erro de bits são examinados e generalizados para esquemas de detecção com cancelamento de interferência. Para situações onde uma seqüência de treinamento não é disponibilizada, é considerado um novo critério de projeto às cegas de receptores com restrições lineares baseado na função custo módulo constante. Algoritmos adaptativos às cegas baseados neste novo critério são usados para estimar os parâmetros de um receptor linear e do canal de comunicações. São também desenvolvidos novos mecanismos às cegas de ajuste do passo para algoritmos do tipo gradiente estocástico em receptores lineares com base no critério de mínima variância com restrições. Com base nos critérios de mínima variância e módulo constante com restrições, são desenvolvidos critérios de projeto às cegas para receptores com decisão realimentada e propostos algoritmos adaptativos para essas estruturas. Um novo esquema de cancelamento sucessivo de interferência baseado no conceito de arbitragem é proposto e incorporado a uma estrutura de recepção com decisão realimentada para o enlace reverso. Em seguida, o novo esquema de cancelamento de interferência é combinado com uma estrutura iterativa que emprega múltiplos estágios, resultando em melhores estimativas do receptor e um desempenho uniforme para os usuários. Finalmente, são apresentadas novas estruturas de recepção com posto reduzido, baseadas em filtros FIR interpolados e interpoladores variantes no tempo, e desenvolvidos algoritmos adaptativos às cegas e supervisionados para o novo esquema. / [en] This thesis presents new structures and algorithms for multiuser detection and interference suppression in DS-CDMA systems. Structures based on recurrent neural networks are investigated for decision feedback receivers and adaptive algorithms are developed for combatting multiple access interference and intersymbol interference. New algorithms based on the minimization of the bit error rate are examined and generalized for detection schemes with interference cancellation. For situations where a training sequence is not available, a new blind criterion, based on the constant modulus cost function with linear constraints is considered. Based on this novel criterion, blind adaptive algorithms are used for estimating the parameters of linear receivers and the channel. New blind adaptive mechanisms for adjusting the step size of stochastic gradient algorithms, using the constrained minimum variance criterion, are also presented for estimating the parameters of linear receivers and the channel. Based on constrained minimum variance and constrained constant modulus criteria, the blind design of decision feedback structures is considered and blind adaptive algorithms are derived. A new successive interference cancellation scheme using the concept of arbitration is proposed and incorporated within a decision feedback structure for uplink scenarios. Then, the new interference cancellation scheme is combined with an iterative structure that employs multiple stages, resulting in improved receiver estimates and providing uniform performance over the users. Finally, novel reduced-rank receiver structures, based on interpolated FIR filters with time-varying interpolators, are presented and blind and supervised adaptive algorithms are developed for this new scheme.

The doctoral thesis focuses on signal representation in the time-frequency domain with constant resolution. In theoretical introduction the possibilities of displaying a signal in time and frequency are summarized. Attention is concentrated on comparison of short-time Fourier Transform (STFT) and Wigner-Ville Distribution (WVD). The latter achieves a significantly better resolution, especially for a linearly modulated signal. The disadvantage of WVD, which is the presence of interferences resulting from the calculation of the instantaneous autocorrelation function, is described in detail. These interferences are due to the presence of multiple components in the signal or its non-linear modulation. Subsequently, several methods are discussed, which can suppress these interferences, but at the cost of resolution loss. One of the interference suppression methods is smoothed pseudo Wigner-Ville distribution. It is further used in this thesis for the analysis of interference suppression when various filtrations in the time-frequency plane are applied. Several signals with multiple components or various non-linear modulations are used. Based on the analysis, a method using a set of variously smoothed pseudo Wigner-Ville distributions is designed to estimate the time-frequency representation with high resolution and minimal interferences. To compare the results to other methods, the quantitative metrics used in the literature are compared. To select the appropriate one a new metric is suggested. It is applicable to simulated signals and uses mean square error. Based on the comparison, the Stankovi\' measure is selected as the most appropriate for comparing results. The selected metric is used to determine the appropriate minimal number of differently smoothed pseudo Wigner-Ville distributions. Using the selected metric, the proposed method is compared with other methods. These are STFT with optimized window length, S-method with optimized parameter and optimization method using radial Gaussian kernel (RGK). These methods are compared based on the set of signals previously used for interference suppression analysis. In addition, noises are added to the signals. Finally, the methods are also compared based on the real bat echo signal. In conclusion, the proposed method outperforms the compared methods in suppressing interference and resolution.

The design of a 24 GHz frequency divider-by-10 for accurate indoor localisation systems is presented. It is proposed to use frequency dividers as active reflector tags in a frequency-modulated continuous wave indoor localisation system in order to reduce interferences caused by direct reflections of the interrogating signal. Since frequency dividers are subharmonic generators, this allows achieving conversion gain in the reflected signal. The frequency divider is fabricated using GLOBAL FOUNDRIES 45 nm-silicon-on-insulator technology. It consumes only 5.7 mW from a 1 V supply. It has a wide locking range of 33% and an efficiency of 3.58 GHz/mW. To the best of authors' knowledge, the use of frequency dividers as active reflectors was not studied before.

[pt] Este trabalho tem como finalidade o estudo das possibilidade do sistema GFDM (Generalized Frequency Division Multiplexing). Para o estudo feito foi apresentado um modelo matricial para representar os sinais gerados no sistema GFDM, a semelhança do modelo de sinal do sistema OFDM (Orthogonal Frequency Division Multiplexing). Tal modelo permitiu a obtenção de expressões analíticas para a Densidade Espectral de Potência (DEP, Spectral Power Density) dos sinais e sua comparação com a DEP dos sinais transmitidos em sistemas OFDM. A partir do modelo matricial apresentado são estudados o desempenho de diferentes tipos de equalizadores/detectores lineares clássicos passíveis de utilização neste sistema de comunicações digitais, tais como Zero Forcing, Minimum Mean Square Error e Matched Filter. Além disso o trabalho propõe e analisa o desempenho resultante da aplicação de técnicas de supressão de interferência PIC (Parallel Interference Cancellation) em conjunto com os detectores lineares mencionados e dos detectores LAS (Likelihood Ascent Search) precedidos por equalizadores Matched Filter (MF-LAS). O número de estágios PIC realizados em cada detecção é controlado por uma estratégia de parada baseada na métrica de distância. Diferentes esquemas de detecção MF-LAS em conjunto com PIC são também propostos e examinadas. Finalmente, partindo do modelo matricial desenvolvido neste trabalho é realizada a estimação de canal empregando a estratégia de símbolos pilotos ortogonais. As diferentes estratégias de detecção examinadas para o sistemas GFDM são comparadas em termos de desempenho BER (Bit Error Rate) e da complexidade computacional associada aos respectivos detectores. Comparações entre os sistemas GFDM e OFDM com destaque na complexidade na geração de sinais, eficiência espectral e desempenho estão também incluídos nesta dissertação. / [en] The main goal of the presented work is to study the possibilities of the GFDM system (Generalized Frequency Division Multiplexing). For achieving this purpose, a matrix model is presented which represents the signals generated in the GFDM system, similar to the signal model of the OFDM (Orthogonal Frequency Division Multiplexing) system. This model allows the obtainment analytical expressions for the Spectral Power Density (DEP) of the signals and their comparison with the DEP of the signals transmitted in OFDM systems. Furthermore, we study the performance of different types of classical linear equalizers/detectors that can be used in the digital communications systems, such as Zero Forcing, Minimum Mean Square Error and Matched Filter. In addition, we propose and analyze the performance resulting from the application of PIC (Parallel Interference Cancellation) interference suppression techniques together with the linear detectors mentioned and LAS (Likelihood Ascent Search) detectors preceded by Matched Filter (MF-LAS) equalizers. The number of PIC stages performed at each detection is controlled by a stop strategy based on the distance metric. Different MF-LAS detection schemes together with PIC are also proposed and examined. Finally, the channel estimation is performing based on the matrix model developed in this work and using orthogonal pilots symbols. The differents strategies of detection examined for GFDM systems are compared in terms of BER performance (Bit Error Rate) and the computational complexity associated with the respective detectors. Comparisons between GFDM and OFDM systems based on criterions as the complexity of the signal generation, spectral efficiency and performance are also included in this dissertation.