Iterative Learning Control for Digital Predistortion with Undersampled Feedback Signal

A novel undersampled feedback signal iterative learning control (US-ILC-DPD) technique is proposed to linearize RF power amplifiers (PAs). The proposed method combines the existing iterative learning control (ILC) with the undersampled processing. It can keep the good linearization performance and avoid using high speed ADCs, which are normally expensive and power costly. In addition, the advantages of ILC-DPD, such as low computational complexity and fast convergence rate, are all preserved in this approach. In linearization, an RFPA is excited by a 20-MHz LTE signal, which is sampled at 100, 25 and 12.5MSPS, respectively. Both the proposed approach and the conventional undersampled-DPD based on direct learning architecture (US-DLA) are validated. The experimental results show that, compared with the conventional US-DLA, the US-ILC-DPD has better robustness and linearization performance when the sampling rate is reduced. Meanwhile, the proposed US-ILC-DPD has much lower computational complexity.