Comparison of RF Power Amplifier Behavioral Models with Respect to Their Modeling Capabilities in Adjacent and Alternate Bands

In this work a comparison of three different behavioral models for radio frequency (RF) power amplifier (PA) is done. The used models were a Volterra series model, the memory polynomial model (MPM) and the generalized memory polynomial model (GMPM). A special focus is put on their modeling capabilities in the adjacent and alternate band. It was found that the Volterra model gives the best modeling performance but also has a high amount of parameter. The GMPM has a significantly lower amount of parameter yet a comparable modeling performance. The MPM uses only very few parameter but still achieves a good modeling performance. The modeling performance for each model was highest in the inband, is less in the adjacent band and decrease further in the alternate band. It also is highly dependent on the used order of nonlinearity.