Behavioral modeling of nonlinear RF power amplifiers for digital wireless communication systems with implications for predistortion linearization systems

The objective of the research in this dissertation is to develop accurate and efficient measurement based behavioral models of nonlinear RF power amplifiers in digital wireless communication systems to predict nonlinear performance metrics such as adjacent channel power ratio, carrier-to-interference ratio, and error vector magnitude. A problem in the current behavioral modeling of nonlinear RF power amplifiers is the discrepancy between the predicted performance metrics and the measured values when the signal used in the modeling and the signal used in the real application are different. We developed an efficient method to predict the performance metrics accurately using an intrinsic CIR kernel function that is extracted from two-tone measurement and input signal statistics for digitally modulated signal. Another problem in the modeling is how to treat memory effects in PAs accurately and efficiently. We developed two models that have a parallel Wiener structure. One uses infinite impulse response filters and memoryless functions, and the other uses sparse delay taps and memoryless functions. These models can predict nonlinear phenomena that occur from memory effects accurately. A figure-of-merit derived from the proposed models was developed that provide a means to quantify the magnitude of PA memory effects in a given device. The developed figure-of-merit is applied to investigate the degradation of performance in a memoryless predistortion that occurs from memory effects. This work will contribute to the area of nonlinear behavioral modeling of RF power amplifier for computer aided design tools and to the development of more accurate predistortion functions for power amplifier linearization.