WCSE 2022 Spring
ISBN: 978-981-18-5852-9 DOI: 10.18178/wcse.2022.04.073

Laplacian Densely Connected Channel Estimation Network in NOMA System

Zhong Yu, Pan Wang, Yongbin Xie

Abstract— Non-orthogonal multiple access (NOMA) is an essential technology in wireless communications because it can increase the number of users. However, it needs the accurate channel state information (CSI) to achieve successive interference cancellation (SIC) and the signal detection. To that end, we propose a laplacian densely connected channel estimation network in NOMA system. The network integrates the laplacian pyramid structure between the dense connected blocks to reconstruct the complete channel matrix based on channel matrix at pilot positions. The Laplacian pyramid structure can gradually increase the size of channel matrix which make full use of channel matrix information of different sizes, and the dense connected network reuse the network information to enhance the network performance. The 3rd Generation Partnership Project (3GPP) channel models and mean square error as estimation error are adopted to evaluate the network performance. The estimation error shows that the proposed network is better than least squares (LS) estimation with linear interpolation, and competitive to the minimum mean square error (MMSE) estimation.

Index Terms— Channel estimation, densely connected network, image super-resolution, non-orthogonal multiple access (NOMA).

Zhong Yu
School of Information and Communication Engineering, Xi‘an University of Posts & Telecommunications
Pan Wang
School of Information and Communication Engineering, Xi‘an University of Posts & Telecommunications
Yongbin Xie
School of Information and Communication Engineering, Xi‘an University of Posts & Telecommunications

[Download]


Cite: Zhong Yu, Pan Wang, Yongbin Xie, " Laplacian Densely Connected Channel Estimation Network in NOMA System, " WCSE 2022 Spring Event: 2022 9th International Conference on Industrial Engineering and Applications, pp. 615-621, Sanya, China, April 15-18, 2022.