A dual branch Hammerstein-Wiener architecture for behavior modeling of wideband RF transmitters

Abstract

This paper proposes a dual branch Hammerstein-Wiener system suitable for behavioral modeling of dynamic nonlinear RF power amplifiers and transmitters. The model consists of a Hammerstein system in parallel with a Wiener system. The model performances in time and frequency domains are experimentally evaluated for a 3G high power Doherty amplifier driven by multi-carrier WCDMA signals. For various orders, the proposed dual branch Hammerstein-Wiener model performances are benchmarked against those of the Hammerstein model, Wiener model, and the well established memory polynomial model. The proposed dual branch model achieves better performance than single branch Wiener and Hammerstein models with higher number of coefficients. Furthermore, it leads to normalized mean square error performance comparable to that of the memory polynomial model while requiring 30% to 40% less coefficients.