Extraction and Improvements of a Behavioral Model Based on the Wiener-Bose Structure Used for Baseband Volterra Kernels Estimation

Abstract

In this paper a strategy to extract a first-zone filtered behavioral model based on Wiener-Bose structure, which has a similar structure to the Volterra series, is introduced, and improvements applied to the model identification are highlighted. Considerations regarding the Volterra kernels symmetry for complex signals are described. A closed form for determining the number of independent parameters for a baseband Volterra series was developed and applied in the Wiener-Bose model parametrization process. A method to improve the Hessian condition number in the model calculation procedure is presented. The modeling and validation results based on measurements for an industry standard signal (WCDMA) are displayed. It is shown that this model is capable to represent efficiently RF-power amplifier (PA) memory effects with good accuracy.