A Digital Predistortion for Concurrent Dual-Band Power Amplifier Linearization Based on Periodically Nonuniform Sampling Theory

Abstract

In this paper, we propose a novel technique of digital predistortion (DPD) for dual-band power amplifiers (PA) based on periodically nonuniform sampling (PNS) theory. In contrast to conventional 2D-DPD models, the proposed PNS-DPD has only a single input which can largely reduce the model complexity. We fold the two stimuli with aliasing and feed it to a simple single-band DPD model. The desired predistorted signals are reconstructed from aliased DPD output through the PNS theory. Compared with conventional multi-input models which include numerous intermodulation products of the input signals, the complexity of the proposed single-input PNS-DPD model is hugely decreased. The model coefficients of the proposed PNS-DPD can be easily extracted with conventional direct or indirect learning architecture. We experimentally evaluate the proposed DPD on a test bench and compare it with other DPD techniques in the literature. The implementation complexity can be reduced by over 30% and the identification complexity is also largely reduced.

Physics Informed Spiking Neural Networks: Application to Digital Predistortion for Power Amplifier Linearization

“…where the input signal is u(n), k is the index for nonlinearity, m and l are the indices for memory, and a kl , b kml and c kml are the complex coefficients. The extraction of DPD model coefficients using direct or indirect learning architecture (DLA/ILA) has been well described in [41] and [42]. If we denote U (ω) and X (ω) as the Fourier transform of u(n) and an FD-GMP model can be deduced from ( 6) for the FD-DPD:…”

Dual-band Transmitter Linearization using Spiking Neural Networks

“…Neural networks has been applied to address the dual-band linearization problem [12], which helps to guide the implementation of an SNN-based solution. In [13], a technique based on periodical nonuniform sampling theory is believed to reach the lowest computational complexity for dual-band linearization methods. However, this technique is still based on classical continuevalue computing which always consumes a large amount of energy compared with event-driven computing [14].…”