Enhanced Training Technique for Nested Look-Up Table Based Behavioral Modeling of Nonlinear Power Amplifiers

摘要

This paper addresses the high complexity of standard behavioral models for dynamic nonlinear power amplifiers. Specifically, it tackles key limitations related to the identification of the nested look-up table model. Enhanced training technique suitable for this type of models is introduced. The proposed training algorithm is experimentally assessed using measured data of a Gallium Nitride based Doherty power amplifier driven by a multi-carrier test signal. The experimental results demonstrate that the use of a new interpolation technique enhances the nested look-up table model accuracy by approximately 2 to 4 dB for short training data. Moreover, the proposed training algorithm achieves the same accuracy as its conventional counterpart while requiring 40% to 60% shorter training data lengths.