Date of Award
Master of Engineering (Research)
Dr. Oliver Gough
Dr. Patrick Pratt (Motorola)
In order to make use of spectrally efficient linear modulation methods for mobile communications, it is necessary to linearise the underlying RF power amplifier (PA). A number of techniques for linearizing a PA have been studied, focusing on the Adaptive Baseband Predistortion. The work is centred on the research concentrates on the analysis of different algorithms, their simulation using MatLab, and implementation of a predistorter as a subsystem of a prototype Cartesian based transmitter.
The thesis describes the typical architecture of a transceiver with an emphasis on the nonlinear analysis. Cavers’ algorithm has been examined in terms of its computational complexity and compared with a novel LMS based predistorter algorithm. The Polar implementation of Cavers’ algorithm was found to be less sensitive to noise and can reach deeper compression points than the Cartesian, while the Cartesian is more stable. Although the LMS demands an a priori calculation for the complex adaptation constant of each sample, it is nearly immune to the noise and, as opposed to Cavers’ algorithm, its ultimate power error is independent of the compression point. Nevertheless, Caver’s algorithm obtains faster convergence under low noise conditions and higher compression points than the LMS algorithm.
A prototype transceiver was built based on an FPGA (Virtex-E) together with appropriate converters, filters and modulators. The results obtained from simulation compared well with the results from the prototype.
Faus-Ferrer, Laura, "Adaptive Baseband Predistortion for RF Power Amplifier Linearization." (2003). Theses [online].
Available at: https://sword.cit.ie/allthe/168