Weighted PLUME digital predistorter model with a reduced-complexity identification algorithm

Abstract

This paper investigates the implementation of a robust linearization technique for high-efficiency power amplifier using digital predistortion. First, a modified version of the conventional PLUME predistorter, termed as weighted PLUME (W-PLUME), is developed by augmenting the conventional PLUME model with an input-power-dependent weighting function. The W-PLUME model accounts for the different trends of the power amplifier behavior at low and high input power levels. Second, a reduced-complexity digital predistorter identification algorithm is proposed in order to select the significant coefficients among all of the DPD coefficients extracted using standard Least Squares algorithm. The model and the identification technique are assessed using LTE-based experimental data. A numerical comparison between the proposed size-reduced DPD and the full size digital predistorter model shows that up to 75% reduction in the number of coefficients (from 174 to 43) can be achieved while maintaining the DPD performance within 0.1dB in terms of ACPR and 0.4dB in terms of NMSE.