Channel Estimation and Pilot Design for Uplink Sparse Code Multiple Access System Based on Complex-Valued Sparse Autoencoder

Channel estimation is one of the most important aspects of wireless communication. Especially, in sparse code multiple access (SCMA) system, the accuracy of channel estimation has a significant impact on decoding performance. Various methods, so far, have been developed for channel estimation. Most of these methods regard channel estimation as a parameter estimation problem of linear models. However, these methods require lots of time–frequency resources to ensure high estimation accuracy. In massive connection scenarios, high pilot overhead makes the spectrum resource more scarce. Therefore, the drawback of conventional channel estimation methods limits the further improvement of system capacity in the Internet of Things (IoT) when time–frequency resources are restricted. To address this problem, in this paper, we propose an efficient channel estimation scheme and sparse pilot structure design method in an SCMA system based on complex-valued sparse autoencoder which is effective to learn features of the wireless channel. Complex-valued sparse autoencoder is a kind of neural network with complex-valued weights. It contains two parts: encoder and decoder. In this paper, the encoder part is used to realize the pilot design. Channel estimation is implemented by the decoder. Complex-valued weights obtained from training are used as baseband pilots. Compared with maximum likelihood channel estimation (MLE) of the linear model, the proposed method can achieve higher channel estimation accuracy with more sparse pilot structure. The bit-error rates’ performance of the SCMA receiver in this paper is very close to that of the perfect channel state information (CSI).

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.