Behaviour modelling of wideband RF transmitters using Hammerstein-Wiener models

The authors concern the application and use of a class of block-oriented systems, known as Hammerstein-Wiener, to model the dynamic non-linear behaviour of radio frequency (RF) transmitters. This class of system models is found to better mimic strong dynamic non-linear systems, as compared to simpler individual Hammerstein and Wiener models. The identification, which may look complicated at first glance, is skillfully tackled in an iterative two-stage identification approach by applying a class of global pattern search techniques along with the Narendra-Gallman method. The number of parameters processed at each iteration is therefore reduced, which considerably facilitates the progress and convergence of the identification algorithm. The model is identified and validated for two classes of power amplifiers. The validation results are obtained for various orders and are compared to the results of memory polynomial models. The Hammerstein-Wiener model achieves comparable results with a much lower level of complexity, in terms of the number of parameters associated with the model.