Broadband Communications, Microwave Power Amplifier Modeling And Polynomial Predistortion Techniques

The application of those multi-carrier / multi-channel digital modulation methods based on the non-constant envelope results in modulation signals with large peak-to-average ratio (PAR), i.e., high fluctuations in their signal envelopes. For the sensitive of these modulation signals to the power amplifier (PA) nonlinearities, the PA needs to have a higher linearity to satisfy communication specifications. Unfortunately, there is an inherent tradeoff between linearity and efficiency in power amplifier design. Thus, PA linearization is necessary for improving linearity without compromising efficiency.Based on the complete and deep study of the phenomena and the causes of PA nonlinear characteristics and memory effects, this paper focuses on the PA polynomial model and its methods of parameters identification. In this paper, a realistic, accurate and simple complex behavior test bed for PA characterization is proposed. Using this method, a power amplifier (MHL21336) based on Freescale-LDMOS transistors was measured under a single carrier WCDMA signal. The PA output power at 1-dB gain compression is approximately 35dBm. The results obtained demonstrate that the proposed method is efficient in characterizing the nonlinearities and memory effects of the power amplifier. A memory polynomial model for the amplifier was constructed based on the measured data from the test bed, and it was used to design a baseband predistorter. It employs the polynomial model and the indirect learning architecture. Experiment results show that, when inputting a single carrier WCDMA signal with 5 MHz bandwidth, the performance of the adjacent channel leakage ratio (ACLR) was improved. It implies that the presented predistortion approach effectively restrains the outband spectrum regrowth.