Performance assessment of RF power amplifier memory polynomial models under different signal statistics

Memory polynomial models are widely used in the behavioural modelling and linearization of RF power amplifiers (PAs). In this paper we assess the performances of theses models under different signal bandwidths, statistics and power distributions. For this purpose, Normalized mean square error (NMSE) has been used to compare the model output to measured data when the PA is driven by different input signals. Several simulations and measurements have been carried out and the results demonstrated the effects of the signal characteristics on the performances of the model. Real multi carrier WCDMA and multi tone signals were used with an experimental Doherty amplifier. The obtained results revealed a degradation of the polynomial model performances when the statistics of the input signals change.