Wiener G-functionals for nonlinear power amplifier digital predistortion

This paper expounds on the pruning of Volterra series used to linearize power amplifiers (PAs) exhibiting memory effects. This pruning approach starts with the identification of the minimum set of dominant kernels needed in the Volterra series modeling for a given PA. The pruned Volterra series is then applied to synthesize a digital predistortion (DPD) function. The proposed pruned Volterra series DPD achieved more than 50 dBc ACPR and −38 dB EVM when a 45 Watts GaN PA at 2.14 GHz was driven by a 20 MHz WCDMA signal. In addition, the proposed model was found to lead to reduced span of the kernels values and better numerical conditioning.