Sobre equalizadores autodidatas de decisão realimentada aplicados a sistemas multiusuário

• Ano de defesa: 2007
• Autor(a) principal: Mendes Filho, João
• Orientador(a): Silva, Magno Teófilo Madeira da
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Presbiteriana Mackenzie
• Palavras-chave em Português: comunicação móvel; equalizador de decisão realimentada; transmissão de dados; sistema multiusuário
• Palavras-chave em Inglês: mobile communication; decision feedback equalizer; data transmission; multiuser systems
• Área do conhecimento CNPq: CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA
• Link de acesso: http://dspace.mackenzie.br/handle/10899/24364

Due to the growing demand for mobile communications, adaptive equalizers play an important role for enhancing the efficiency of data transmission. In this scenario, the Decision Feedback Equalizer (DFE) stands out. It presents a favorable tradeoff between computational cost and efficient behavior, mainly when compared to Linear Transversal Equalizer. In this work, the blind adaptation of DFE is investigated for the single and multiuser cases. In the single user case, the perfect equalization conditions for the DFE are revisited, considering the absence of noise and feedback of correct decisions. Assuming the joint blind adaptation of the DFE's feedforward and feedback filters, two stochastic gradient algorithms are also revisited. The first is based on the Constant Modulus cost function, subjected to a constraint to avoid degenerate solutions. The second considers the minimization of a cost function that takes into account the probability density function of the equalizers's output. This latter, known in the literature as the Soft Decision-Directed (SDD) algorithm, was proposed for the recovery of signals based on the Quadrature Amplitude Modulation (QAM). From the division of the complex plane into regions containing 4-QAM type constellations, we propose a modification in the SDD algorithm based on the centers of these regions. The resulting algorithm presents a more favorable tradeoff between convergence rate and computational cost. Moreover, in order to mitigate the steady-state mean-square error, we consider concurrent algorithms based on the previous mentioned. As a core of this dissertation, the perfect equalization conditions and the remarked algorithms are extended to the multiuser case. Simulation results point out that the Modified SDD algorithm and its concurrent adaptation with the constrained Constant Modulus Algorithm present advantages in terms of convergence rate for the blind adaptation of DFE in the recovering of QAM signals.