Channel Estimation of 5G New Radio using Convolutional Neural Network

Due to high throughput, low latency, reliability, and massive communication, 5G and beyond networks have become popular in the era of high-speed wireless communication. This high-frequency multi-path transmission suffers from attenuation, delay, and distortion which affects the signal transmission. In this practical scenario, it is necessary to know or estimate the channel to reconstruct the original signal. The effective channel estimate is a crucial component of the receiver. In this study, deep learning aids in channel estimation for a 5G-and-beyond network under the level of mobility given by the Doppler impacts. Utilizing the deep learning feature to estimate the channel, lowers complexity and improves the accuracy of wireless systems. The mean square error (MSE) results of channel estimation show that the convolutional neural network (CNN) outperforms the linear interpolation for all the tapped delay line (TDL) modes. Similarly, CNN MSE results for TDL-A are 72%, TDL-B 7%, TDL-C 10%, TDL-D and TDL-E 44% less than the practical estimator for 10 dB SNR.