Research on ultra wide band NLoS/LoS recognition method based on the fusion of 1DCNN and LSTM

A deep learning model based on one-dimensional-convolutional neural network-convolutional long short-term memory (LSTM) attention network (1DCNN-CLANet) was proposed to improve the ranging accuracy and positioning performance of ultra wide band (UWB) localization systems under non-line-of-sight (NLo...

Full description

Saved in:
Bibliographic Details
Main Authors: ZHENG Enrang, MENG Xin, JIANG Suying, XUE Jing, ZHANG Yi, LI Qiang
Format: Article
Language:zho
Published: Editorial Department of Journal on Communications 2025-06-01
Series:Tongxin xuebao
Subjects:
Online Access:http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2025102/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850089927546503168
author ZHENG Enrang
MENG Xin
JIANG Suying
XUE Jing
ZHANG Yi
LI Qiang
author_facet ZHENG Enrang
MENG Xin
JIANG Suying
XUE Jing
ZHANG Yi
LI Qiang
author_sort ZHENG Enrang
collection DOAJ
description A deep learning model based on one-dimensional-convolutional neural network-convolutional long short-term memory (LSTM) attention network (1DCNN-CLANet) was proposed to improve the ranging accuracy and positioning performance of ultra wide band (UWB) localization systems under non-line-of-sight (NLoS) conditions. Convolutional neural network (CNN) was first employed to extract spatial features from channel impulse response (CIR) data, and LSTM network was used to capture their temporal characteristics. Then, CNN was further applied to extract additional features such as distance data, signal amplitude, and maximum noise strength. Finally, An attention mechanism was then incorporated to fuse the CIR and additional feature branches for accurate NLoS/LoS classification. Experimental results show that 1DCNN-CLANet achieves classification accuracies of 99.51% for binary classification and 98.47% for four-class classification in complex environments, outperforming other approaches. The model demonstrates strong potential for robust NLoS identification in practical UWB localization systems.
format Article
id doaj-art-3d3f82aea53643dfaadcd52577fda50a
institution DOAJ
issn 1000-436X
language zho
publishDate 2025-06-01
publisher Editorial Department of Journal on Communications
record_format Article
series Tongxin xuebao
spelling doaj-art-3d3f82aea53643dfaadcd52577fda50a2025-08-20T02:42:39ZzhoEditorial Department of Journal on CommunicationsTongxin xuebao1000-436X2025-06-0146285302114256710Research on ultra wide band NLoS/LoS recognition method based on the fusion of 1DCNN and LSTMZHENG EnrangMENG XinJIANG SuyingXUE JingZHANG YiLI QiangA deep learning model based on one-dimensional-convolutional neural network-convolutional long short-term memory (LSTM) attention network (1DCNN-CLANet) was proposed to improve the ranging accuracy and positioning performance of ultra wide band (UWB) localization systems under non-line-of-sight (NLoS) conditions. Convolutional neural network (CNN) was first employed to extract spatial features from channel impulse response (CIR) data, and LSTM network was used to capture their temporal characteristics. Then, CNN was further applied to extract additional features such as distance data, signal amplitude, and maximum noise strength. Finally, An attention mechanism was then incorporated to fuse the CIR and additional feature branches for accurate NLoS/LoS classification. Experimental results show that 1DCNN-CLANet achieves classification accuracies of 99.51% for binary classification and 98.47% for four-class classification in complex environments, outperforming other approaches. The model demonstrates strong potential for robust NLoS identification in practical UWB localization systems.http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2025102/UWBNLOSdeep learning modelCNNLSTM network
spellingShingle ZHENG Enrang
MENG Xin
JIANG Suying
XUE Jing
ZHANG Yi
LI Qiang
Research on ultra wide band NLoS/LoS recognition method based on the fusion of 1DCNN and LSTM
Tongxin xuebao
UWB
NLOS
deep learning model
CNN
LSTM network
title Research on ultra wide band NLoS/LoS recognition method based on the fusion of 1DCNN and LSTM
title_full Research on ultra wide band NLoS/LoS recognition method based on the fusion of 1DCNN and LSTM
title_fullStr Research on ultra wide band NLoS/LoS recognition method based on the fusion of 1DCNN and LSTM
title_full_unstemmed Research on ultra wide band NLoS/LoS recognition method based on the fusion of 1DCNN and LSTM
title_short Research on ultra wide band NLoS/LoS recognition method based on the fusion of 1DCNN and LSTM
title_sort research on ultra wide band nlos los recognition method based on the fusion of 1dcnn and lstm
topic UWB
NLOS
deep learning model
CNN
LSTM network
url http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2025102/
work_keys_str_mv AT zhengenrang researchonultrawidebandnloslosrecognitionmethodbasedonthefusionof1dcnnandlstm
AT mengxin researchonultrawidebandnloslosrecognitionmethodbasedonthefusionof1dcnnandlstm
AT jiangsuying researchonultrawidebandnloslosrecognitionmethodbasedonthefusionof1dcnnandlstm
AT xuejing researchonultrawidebandnloslosrecognitionmethodbasedonthefusionof1dcnnandlstm
AT zhangyi researchonultrawidebandnloslosrecognitionmethodbasedonthefusionof1dcnnandlstm
AT liqiang researchonultrawidebandnloslosrecognitionmethodbasedonthefusionof1dcnnandlstm