Skip to Main Content
A novel enhanced Hammerstein behavior model consisting of a weighted memoryless polynomial followed by a Volterra filter is proposed. The weighted polynomial is used for predicting the strong static nonlinear behaviors of the power amplifiers (PAs). Since the Volterra filter is employed only for the mild dynamic nonlinearities, the filter can be implemented with low nonlinear order. Thus, this proposed model is capable of predicting both the static and dynamic nonlinearities of RF PAs with the acceptable complexity. The modeling performance of the proposed model is assessed in terms of in-band and out-of-band errors, such as normalized mean square error and adjacent channel error power ratio, and it is compared with a conventional Hammerstein, an augmented Hammerstein, and a Volterra series with respect to computation complexities such as the number of floating point operations and coefficients. The excellent estimation capability of the enhanced Hammerstein model is validated by two kinds of PAs: Si lateral diffusion metal-oxide-semiconductor and GaN high electron-mobility transistor amplifiers. Furthermore, the proposed scheme is applied to the digital predistortion (DPD) to cancel the nonlinearities of the PAs. The modeling performances and DPD experimental results clearly demonstrate the superiority of the enhanced Hammerstein scheme: the computational complexity is comparable with the augmented Hammerstein behavioral model, but the modeling performance is similar to the Volterra filter, which is the most accurate model.