Characterization Using Synthetic Modulated Signals

Synthetic test signals have also been proposed for the characterization of RF power amplifiers and transmitters in behavioral modeling and predistortion applications. These signals tend to emulate the behavior that the DUT would exhibit with the modulated test signal generated according to communication standards, while reducing the complexity associated with the identification process [15-17] or improving its robustness [18]. In [16], triangular chirp signal was applied for the training step of the behavioral modeling and DPD of a power amplifier driven by an OFDM (orthogonal frequency division multiplexing) test signal. Compared to a predistorter trained with the same OFDM signal, the adoption of the chirp test signal during the DPD learning process led to comparable linearization performance in terms of spectral regrowth with a significantly lower calibration time and computational load [16]. Furthermore, it was reported that the use of the chirp signal for the predistorter training in OFDM driven power amplifiers can lead to better linearization performance than the case where the predistorter training is done using multi-tone test signals [17].

In [18], the problem of the ill-conditioning of the data matrix in memory polynomial model was circumvented by the proper design of the test signal. This signal is built by controlling the probability distribution function (pdf) of the real and imaginary components of the baseband waveform. A generalized Gaussian distribution was proposed to generate a CDMA (code division multiple access)-like test signal that has a parameterized probability density function optimized for a minimum condition number of the autocorrelation matrix. It was demonstrated that such signal tackles the ill-conditioning problem present in memory polynomial models and improves the robustness of the model identification. Similar to the chirp signal, the parameterized-pdf synthetic test signal was found to lead to DPD performance similar to that obtained when the actual modulated signal is used for the predistorter training.

 
Source
< Prev   CONTENTS   Source   Next >