A Direct Learning Digital Predistortion Method to Compensate for the Power Amplifier Nonlinear Distortion

Abstract

In this paper, a novel direct learning digital predistortion (DPD) relied on gradient-based algorithm to identify the model of power amplifier (PA) is proposed. Unlike the conventional DPD introducing an inevitable calculation error in model identification, the proposed method accurately calculates the predistortion function by constructing a univariate polynomial and finding its roots to obtain the accurate value of the DPD function, which linearizes the PA more precisely. Simulations show that the proposed linearization scheme outperforms the conventional DPD in the normalized mean square error (NMSE) performance, and the adjacent channel leakage ratio (ACLR) performance as well.