Deep Learning-Aided Receiver Against Nonlinear Distortion of HPA in OFDM Systems
Shuntian Zheng, Sheng Yi Wu, Haihan Li, Chunxiao Jiang, Xiaojun Jing · 2021 13th International Conference on Wireless Communications and Signal Processing (WCSP) · 2021
In this paper, we focus on the design of a novel receiver for wireless systems against the channel and the nonlinear distortion caused by high-power amplifiers. With the aid of deep learning, a channel estimator and a signal detector with robustness and efficiency are designed under the framework of the traditional system. In particular, the channel matrix derived from the pilot is converted into a two-dimensional image, and the residual feature aggregation network is applied to reconstruct the channel state information (CSI). In addition, based on the outcome of the estimator and the flexible denoising convolutional neural network, a detection network is derived to denoise and detect signals. The proposed scheme is evaluated in orthogonal frequency division multiplexing (OFDM) systems with fading channels and nonlinear amplifiers, and the experimental results confirmed its superiority in terms of estimation and detection performance.