Comparison of 2-D behavioral models for modeling and digital predistortion of envelope tracking power amplifiers

In this paper we present a comparison of dual- input single-output static polynomial models for behavioral modeling and digital predistortion of envelope tracking RF power amplifiers. The 2-D Taylor polynomial model is here introduced and compared to state-of-the-art 2-D models for different values of their nonlinearity orders. The comparison is performed using measurement results on an envelope tracking system assembled using commercially available modules. The Normalized Mean Square Error (NMSE) and the Adjacent Channel Error Ratio (ACEPR) are used to evaluate the modeling performance. Memoryless dual-input DPD is also applied using 2-D Taylor polynomial model and compared to state-of-the-art 2-D static polynomial DPD. The result comparison allows to identify the model nonlinear kernels that can achieve the best modeling and linearization performance when applied to ET PAs.