Digital signal processing techniques for the compensation of analog circuit impairments in impedance mismatched radio-frequency transmitters.

Detalhes bibliográficos
Ano de defesa: 2021
Autor(a) principal: Hemsi, Cyro Scarano
Orientador(a): Não Informado pela instituição
Banca de defesa: Não Informado pela instituição
Tipo de documento: Tese
Tipo de acesso: Acesso aberto
Idioma: eng
Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
Programa de Pós-Graduação: Não Informado pela instituição
Departamento: Não Informado pela instituição
País: Não Informado pela instituição
Palavras-chave em Português:
Link de acesso: https://www.teses.usp.br/teses/disponiveis/3/3142/tde-23092021-112819/
Resumo: This thesis investigates and proposes new models for the digital compensation of radio frequency (RF) imperfections in broadband wireless transmitters, more specifically for non-linear power amplifiers (PA) subject to load impedance mismatch (LMM). Such imperfections in RF transmitters, together with the in-phase and quadrature (IQ) imbalance in digital modulators, are responsible for degrading the transmitter\'s performance, in terms of spectral purity, modulation quality and bit error rate (BER). Several practical scenarios in which PAs are subject to LMM motivate the research for more advanced non-linear behavioral models with memory for digital predistortion (DPD), capable of overcoming the limitations reported in the literature of traditional polynomial models, while being less complex than existing approaches to PA LMM. This thesis proposes the application of the Wiener-Hammerstein with feedback (WHFB) polynomial model as a simplified behavioral model for DPD in the context of PAs subject to LMM. The high dimensionality of the proposed WHFB structure can be reduced through sparse estimation techniques, such as the least absolute shrinkage and selection operator (LASSO) and group LASSO extensions, which are able to significantly decrease the number of coefficients needed, thus reducing the length of the DPD filter and, proportionally, the cost of filtering. In addition, block-oriented LASSO extensions, such as Group-LASSO and Sparse-group LASSO, are applied in the context of model sizing, that is, in the task of determining appropriate values for the model parameters, which traditionally requires an exhaustive search. Dense and sparse WHFB models are experimentally validated through measurements from an experimental test set-up and also compared to others, including models based on parallel factors (PARAFAC) decomposition and on Laguerre expansion, thus demonstrating their ability to adequately compensate for subject PAs the LMM. Finally, a new strategy for reducing the Volterra model is proposed, which results in a more flexible and modular memory polynomial, in which the parameters are chosen independently for each order of nonlinearity. This flexible approach is able to accurately describe a wide range of operating/environmental conditions of PAs.