A complexity-reduced band-limited memory polynomial behavioral model for wideband power amplifier

Abstract

Band-limited Volterra series model is an effective way for power amplifier (PA) modeling and digital predistortion (DPD), in the presence of limited DAC/ADC sampling rate. However, it suffers from the issue of requiring very high order baseband band-limiting filter to guarantee the model accuracy, which increases computational complexity drastically. This paper proposes a more efficient band-limited model to reduce the required order of the band-limiting filter significantly, by adding an extra error compensation module into the original band-limited memory polynomial (BLMP) model. A class-AB GaN PA excited by a 60MHz bandwidth orthogonal frequency division multiplexing (OFDM) signal was tested to validate this approach. Experimental results prove that the new model has much lower complexity and higher accuracy than BLMP model.