Low-Complexity Adaptive Digital Pre-Distortion with Meta-Learning based Neural Networks

In this paper, we study a meta-learning based neural network (NN) model to enhance the energy-efficiency related to power amplification in wireless communication systems. Specifically, we introduce a low-complexity adaptive solution to perform digital pre-distortion (DPD) for power amplifiers (PAs) using neural networks. Thus, we design a dedicated NN architecture to derive pre-distortion functions using few neurons. Moreover, we develop a meta-learning training approach that allows better generalization and faster adaptation, over time-varying PAs, compared to classical DPD schemes. Thereby, we propose a new approach to realize an adaptive digital predisorter based on meta-learning. A dedicated architecture allows to achieve lowcomplexity while meta-learning permits adapting to most parameters change in the system using few data. Through the simulation results, we have shown that this approach can offer a metatrained DPD function that can provide satisfying performance for different PA models using only 3000 IQ symbols during training phase. Contrary to fixed DPD architectures, the performance of our meta-trained DPD can be improved through only few gradient steps and over few samples during online calibration, achieving excellent performance with moderate complexity.