Intelligent FCDAE Denoiser for Reliable Decoding Over Correlated Noise Channel

In practical systems, colored noise commonly imposes on the data delivered, since the ideal white noise can hardly be achieved. In order to suppress the noise, we propose a denoiser based on a fully convolutional denoising autoencoder (FCDAE) for more reliable information recovery. The smart FCDAE c...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on vehicular technology Vol. 73; no. 10; pp. 15659 - 15663
Main Authors: Li, Yan, Lu, Huaiyin, Zhang, Lin, Chen, Jianyuan, Wu, Zhiqiang
Format: Journal Article
Language:English
Published: New York IEEE 01.10.2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Bit error rate
Convolutional neural networks
Correlated noise
Decoders
Decoding
deep learning
Error correcting codes
fully convolutional denoising autoencoder (FCDAE)
intelligent denosier
Noise measurement
Noise reduction
Reliability
Signal to noise ratio
System reliability
White noise
ISSN:0018-9545, 1939-9359
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In practical systems, colored noise commonly imposes on the data delivered, since the ideal white noise can hardly be achieved. In order to suppress the noise, we propose a denoiser based on a fully convolutional denoising autoencoder (FCDAE) for more reliable information recovery. The smart FCDAE consists of a fully convolution-based encoder and decoder to learn the noise structure distribution in the received signals. Besides, the loss function is selected as a combination of mean square error (MSE) and normality test loss. The proposed denoiser can be further combined with decoders such as the low-density parity-check (LDPC) codes. The simulation results over channels with correlated noise demonstrate that the proposed FCDAE denoiser can effectively improve the signal-to-noise ratio (SNR) and bit error rate (BER) performance, thereby achieving superior reliability performance to benchmark systems.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-9545
1939-9359
DOI:10.1109/TVT.2024.3399838