Performance of OFDM system receiver Based on Deep Learning
Yupeng An, Zhongdong Wu, Chunyang Tang · 2021
In order to explore the performance of deep learning in wireless communication receivers, a neural network-based method is proposed to realize the functions of OFDM receivers. This method is to directly reduce the noise and restore the transmitted symbols, and does not add the existing channel estimation function network module. Specifically, the mathematical calculation function of Discrete Fourier Transform (DFT) and the training of OFDM signal samples are used to reduce the ser error rate (SER) through the Convolutional Neural Network(CNN). In addition, The robustness of this method is proved by data sets under different modulation modes and different SNR conditions. This article introduces the preliminary research results of OFDM receivers. The experimental results of the simulation in this article show that the method we propose is better than the SER of OFDM receiver based on traditional linear least squares (LS) channel estimation, which drop by 1.6% ~ 6.7% in the conditions of different SNRs and modulation modes.