Learning Aided Intelligent Mechanism for Channel Estimation in 5G Wireless Networks

The attenuation of signals due to propagation environment is the major challenge of wireless communication which results in frequent call drops, reduced signal strength and low transmission rates. The propagation channel often degrades the signal quality at the receiver due to channel effects including fading, shadowing, path loss and other overhead. The propagation channel and its successful estimation is very important for ensuring communication reliability in next generation wireless systems. This paper presents an intelligent mechanism based on deep learning to estimate the wireless channel such that the system spectral efficiency is enhanced. The impact of signal distortion due to hardware effects is also considered. Further, the proposed scheme is compared with the conventional LMMSE channel estimation scheme. Also, to extract the data using the estimated channel, the performance of proposed scheme is evaluated using three receivers namely, RZF, MMSE and modified MMSE. It is observed that the proposed scheme outperforms the LMMSE scheme in terms of normalised mean square error (NMSE) by 14.43% and by 27.16% in the absence of distortion. In the end, the performance comparison with the system with known perfect channel state information (CSI) is also performed.