Simplified parameter-extraction process for digital predistortion based on the indirect learning architecture

In this paper, simplified parameter-extraction process is presented for digital predistortion (DPD) design. The predistorter identification is based on the indirect learning architecture. In the proposed parameter-extraction procedure, the feedback data from PA output is normalized by its own peak value instead of expected gain of PA, and then this data is used for parameters extraction. The estimated coefficients are directly used in the predistortion function. The normalized data to predistorter training block has peak value of unity. This leads to that the original input data can be normalized by its own peak value when predistorter is inserted in the signal path. The predistorter using simplified parameter-extraction process is equivalent to the conventional one using maximum gain as normalization gain. However, gain measurement or calculation is eliminated in the predistorter training step, which significantly reduces the DPD implementation complexity. A comparison with conventional parameter-extraction process based predistorter is also presented.