Performance comparison of training sequences for power amplifier linearization systems

Abstract

The suitable design of a calibration signal for power amplifier (PA) behavioral modeling and digital predistortion design is addressed in this work. The proposed alternatives are a manageable triangular chirp-like waveform or a multisine-type sequence that resembles as much as possible to the transmitted signal. In this research, multisines are generated accordingly to the statistical characteristics of the transmitted signal, whereas chirp sequence is defined by the system transmission parameters. Laboratory tests show that, in spite of the remarkable existing similarities between the transmitted signal and the multisines, the chirp sequence accomplishes the same enhancement in nonlinear distortion reduction than the multisines (up to 16 dB) while reducing the training time needed for computing the predistorter coefficients.