Nesta tese analisamos e propomos métodos para a equalização não-treinada (cega) de canais de comunicação lineares FIR baseados em filtros de partículas, que são técnicas recursivas para a solução Bayesiana de problemas de filtragem estocástica. Iniciamos propondo novos métodos para equalização sob ruído gaussiano que prescindem do uso de codificação diferencial, ao contrário dos métodos existentes. Empregando técnicas de evolução artificial de parâmetros, estendemos estes resultados para o caso de ruído aditivo com distribuição não-gaussiana. Em seguida, desenvolvemos novos métodos baseados nos mesmos princípios para equalizar e decodificar conjuntamente sistemas de comunicação que empregam códigos convolucionais ou em bloco. Através de simulações numéricas, observamos que os algoritmos propostos apresentam desempenhos, medidos em termos de taxa média de erro de bit e velocidade de convergência, marcadamente superiores aos de métodos tradicionais, freqüentemente aproximando o desempenho dos algoritmos ótimos (MAP) treinados. Além disso, observamos que os métodos baseados em filtros de partículas determinísticos exibem desempenhos consistentemente superiores aos dos demais métodos, sendo portanto a melhor escolha caso o modelo de sinal empregado permita a marginalização analítica dos parâmetros desconhecidos do canal. / In this thesis, we propose and analyze blind equalization methods suitable for linear FIR communications channels, focusing on the development of algorithms based on particle filters - recursive methods for approximating Bayesian solutions to stochastic filtering problems. Initially, we propose new equalization methods for signal models with gaussian additive noise that dispense with the need for differentially encoding the transmitted signals, as opposed to the previously existing methods. Next, we extend these algorithms to deal with non-gaussian additive noise by deploying artificial parameter evolution techniques. We next develop new joint blind equalization and decoding algorithms, suitable for convolutionally or block-coded communications systems. Via numerical simulations we show that the proposed algorithms outperform traditional approaches both in terms of mean bit error rate and convergence speed, and closely approach the performance of the optimal (MAP) trained equalizer. Furthermore, we observed that the methods based on deterministic particle filters consistently outperform those based on stochastic approaches, making them preferable when the adopted signal model allows for the analytic marginalization of the unknown channel parameters.

Nesta tese analisamos e propomos métodos para a equalização não-treinada (cega) de canais de comunicação lineares FIR baseados em filtros de partículas, que são técnicas recursivas para a solução Bayesiana de problemas de filtragem estocástica. Iniciamos propondo novos métodos para equalização sob ruído gaussiano que prescindem do uso de codificação diferencial, ao contrário dos métodos existentes. Empregando técnicas de evolução artificial de parâmetros, estendemos estes resultados para o caso de ruído aditivo com distribuição não-gaussiana. Em seguida, desenvolvemos novos métodos baseados nos mesmos princípios para equalizar e decodificar conjuntamente sistemas de comunicação que empregam códigos convolucionais ou em bloco. Através de simulações numéricas, observamos que os algoritmos propostos apresentam desempenhos, medidos em termos de taxa média de erro de bit e velocidade de convergência, marcadamente superiores aos de métodos tradicionais, freqüentemente aproximando o desempenho dos algoritmos ótimos (MAP) treinados. Além disso, observamos que os métodos baseados em filtros de partículas determinísticos exibem desempenhos consistentemente superiores aos dos demais métodos, sendo portanto a melhor escolha caso o modelo de sinal empregado permita a marginalização analítica dos parâmetros desconhecidos do canal. / In this thesis, we propose and analyze blind equalization methods suitable for linear FIR communications channels, focusing on the development of algorithms based on particle filters - recursive methods for approximating Bayesian solutions to stochastic filtering problems. Initially, we propose new equalization methods for signal models with gaussian additive noise that dispense with the need for differentially encoding the transmitted signals, as opposed to the previously existing methods. Next, we extend these algorithms to deal with non-gaussian additive noise by deploying artificial parameter evolution techniques. We next develop new joint blind equalization and decoding algorithms, suitable for convolutionally or block-coded communications systems. Via numerical simulations we show that the proposed algorithms outperform traditional approaches both in terms of mean bit error rate and convergence speed, and closely approach the performance of the optimal (MAP) trained equalizer. Furthermore, we observed that the methods based on deterministic particle filters consistently outperform those based on stochastic approaches, making them preferable when the adopted signal model allows for the analytic marginalization of the unknown channel parameters.

This thesis studies novel and efficient computational sampling methods for appli- cations in three types of stochastic inversion problems: seismic waveform inversion, filtering problems, and static parameter estimation. A primary goal of a large class of seismic inverse problems is to detect parameters that characterize an earthquake. We are interested to solve this task by analyzing the full displacement time series at a given set of seismographs, but approaching the full waveform inversion with the standard Monte Carlo (MC) method is prohibitively expensive. So we study tools that can make this computation feasible. As part of the inversion problem, we must evaluate the misfit between recorded and synthetic seismograms efficiently. We employ as misfit function the Wasserstein metric origi- nally suggested to measure the distance between probability distributions, which is becoming increasingly popular in seismic inversion. To compute the expected values of the misfits, we use a sampling algorithm called Multi-Level Monte Carlo (MLMC). MLMC performs most of the sampling at a coarse space-time resolution, with only a few corrections at finer scales, without compromising the overall accuracy. We further investigate the Wasserstein metric and MLMC method in the context of filtering problems for partially observed diffusions with observations at periodic time intervals. Particle filters can be enhanced by considering hierarchies of discretizations to reduce the computational effort to achieve a given tolerance. This methodology is called Multi-Level Particle Filter (MLPF). However, particle filters, and consequently MLPFs, suffer from particle ensemble collapse, which requires the implementation of a resampling step. We suggest for one-dimensional processes a resampling procedure based on optimal Wasserstein coupling. We show that it is beneficial in terms of computational costs compared to standard resampling procedures. Finally, we consider static parameter estimation for a class of continuous-time state-space models. Unbiasedness of the gradient of the log-likelihood is an important property for gradient ascent (descent) methods to ensure their convergence. We propose a novel unbiased estimator of the gradient of the log-likelihood based on a double-randomization scheme. We use this estimator in the stochastic gradient ascent method to recover unknown parameters of the dynamics.