A Low Complexity NARX Structure using IndirectLearning Architecture for Digital Pre-Distortion

In this paper, we demonstrate a nonlinear autoregressive with exogenous input (NARX) DPD technique which is more compact, less computationally intensive and less susceptible to errors caused by noise in the PA output compared to an equivalent memory polynomial based DPD. Experimental validation is performed with a 20 MHz LTE signal for a GaN Doherty power amplifier.