Deep-Learning-based Channel Estimation for Chaotic Wireless Communication

A deep-learning-based channel estimation method for chaotic wireless communication is proposed in this letter, which is based on a deep neural network (DNN) pre-trained by the stacked denoising autoencoder (SDAE) structure. The DNN learns the channel parameters by using the autocorrelation function...

Full description

Saved in:
Bibliographic Details
Published in:IEEE wireless communications letters Vol. 13; no. 1; p. 1
Main Authors: Yin, Hui-Ping, Zhao, Xiao-Hui, Yao, Jun-Ling, Ren, Hai-Peng
Format: Journal Article
Language:English
Published: Piscataway IEEE 01.01.2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Artificial neural networks
Autocorrelation functions
Channel estimation
chaotic baseband wireless communication system (CBWCS)
Chaotic communication
Communications systems
Decoding
Deep learning
deep neural network (DNN)
Estimation
Identification methods
Machine learning
Noise reduction
Robustness (mathematics)
stacked denoising autoencoder (SDAE)
Wireless communication
Wireless communications
ISSN:2162-2337, 2162-2345
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A deep-learning-based channel estimation method for chaotic wireless communication is proposed in this letter, which is based on a deep neural network (DNN) pre-trained by the stacked denoising autoencoder (SDAE) structure. The DNN learns the channel parameters by using the autocorrelation function (ACF) of the received signal in the sense of minimizing the mean squared error (MSE). Numerical results demonstrate that the proposed scheme learns the channel very well and significantly outperforms the conventional schemes in terms of the channel estimation MSE, as well as the BER performance of the communication system. The proposed channel estimation method based on the ACF of chaotic signal is robust to the noise because of the effect of the double noise resistance operation including the autocorrelation operation and the denoising autoencoder. The proposed scheme is a blind identification method, which uses the received signal directly, by this way, saves the valuable bandwidth resource without any probe signal.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2162-2337
2162-2345
DOI:10.1109/LWC.2023.3323683