In a base station, the power amplifier (PA) experiences rapid fluctuations in behavior because of the varying power levels and bandwidth of its excitation signals. High peak-to-average power ratios and nonconstant envelopes of wideband signals impose stringent linear and efficiency requirements on the PA required for amplification.

It is necessary to design fast and noncomplex digital pre-distorters that are able to achieve and maintain acceptable linearization performance and compensate for the dynamic distortions generated by the PA. This thesis explores an adaptive DPD scheme using multiple models that combines a switching and a simple adaptation algorithm.

Experimental data are used in evaluating the proposed scheme. The multiple model scheme reduces the transient error response and requires a relatively small number of data samples for identification, switching and adaptation. The scheme is demonstrated to be fast, less complex, and capable of maintaining and achieving the linearity requirement of the PA.