Signal-Based Digital Predistortion for Linearization of Power Amplifiers

In this paper, a signal-based digital predistortion (DPD) scheme is proposed for the linearization of power amplifiers (PAs) in broadcasting transmitters. The proposed DPD scheme is called the signal-based DPD to distinguish it from the conventional model-based DPD, where the motivation of the former is to synthesize the predistorted signal and that of the latter is to construct the preinverse function of the PA. Our work on this signal-based DPD scheme mainly consists of two parts: a signal-based predistorter and its identification. The signal-based predistorter is derived according to the typical fixed point approach associated with the contraction mapping theorem (CMT), and it has an online cascade structure that is constructed by reproducing offline iterations of the CMT. Identification methods are proposed by tackling the divergent output problem of the signal-based predistorter with a cascade structure. In particular, three coefficient extraction strategies, which have different complexities and different linearization results, are proposed to identify the signal-based predistorter. The experimental results verify that the proposed signal-based DPD scheme can achieve the best linearization performance compared with all the typical model-based DPD methods, including the indirect learning architecture-based DPD, direct learning architecture-based DPD, and iterative learning control-based DPD.