Accurate localization method combining optimized hybrid neural networks for geomagnetic localization with multi-feature dead reckoning

Location-based services provide significant economic and social benefits. The ubiquity, low cost, and accessibility of geomagnetism are highly advantageous for localization. However, the existing geomagnetic localization methods suffer from location ambiguity. To address these issues, we propose a f...

Full description

Bibliographic Details
Main Authors: Yan, Suqing, Luo, Baihui, Sun, Xiyan, Xiao, Jianming, Ji, Yuanfa, Kamarul Hawari, Ghazali
Format: Article
Language:English
Published: MDPI 2025
Subjects:
Online Access:http://umpir.ump.edu.my/id/eprint/44411/
http://umpir.ump.edu.my/id/eprint/44411/1/Accurate%20localization%20method%20combining%20optimized%20hybrid%20neural%20networks.pdf
_version_ 1848827097570607104
author Yan, Suqing
Luo, Baihui
Sun, Xiyan
Xiao, Jianming
Ji, Yuanfa
Kamarul Hawari, Ghazali
author_facet Yan, Suqing
Luo, Baihui
Sun, Xiyan
Xiao, Jianming
Ji, Yuanfa
Kamarul Hawari, Ghazali
author_sort Yan, Suqing
building UMP Institutional Repository
collection Online Access
description Location-based services provide significant economic and social benefits. The ubiquity, low cost, and accessibility of geomagnetism are highly advantageous for localization. However, the existing geomagnetic localization methods suffer from location ambiguity. To address these issues, we propose a fusion localization algorithm based on particle swarm optimization. First, we construct a five-dimensional hybrid LSTM (5DHLSTM) neural network model, and the 5DHLSTM network structure parameters are optimized via particle swarm optimization (PSO) to achieve geomagnetic localization. The eight-dimensional BiLSTM (8DBiLSTM) algorithm is subsequently proposed for heading estimation in dead reckoning, which effectively improves the heading accuracy. Finally, fusion localization is achieved by combining geomagnetic localization with an improved pedestrian dead reckoning (IPDR) based on an extended Kalman filter (EKF). To validate the localization performance of the proposed PSO-5DHLSTM-IPDR method, several extended experiments using Xiaomi 10 and Hi Nova 9 are conducted in two different scenarios. The experimental results demonstrate that the proposed method improves localization accuracy and has good robustness and flexibility
first_indexed 2025-11-15T03:55:18Z
format Article
id ump-44411
institution Universiti Malaysia Pahang
institution_category Local University
language English
last_indexed 2025-11-15T03:55:18Z
publishDate 2025
publisher MDPI
recordtype eprints
repository_type Digital Repository
spelling ump-444112025-04-29T06:37:41Z http://umpir.ump.edu.my/id/eprint/44411/ Accurate localization method combining optimized hybrid neural networks for geomagnetic localization with multi-feature dead reckoning Yan, Suqing Luo, Baihui Sun, Xiyan Xiao, Jianming Ji, Yuanfa Kamarul Hawari, Ghazali QA75 Electronic computers. Computer science TK Electrical engineering. Electronics Nuclear engineering Location-based services provide significant economic and social benefits. The ubiquity, low cost, and accessibility of geomagnetism are highly advantageous for localization. However, the existing geomagnetic localization methods suffer from location ambiguity. To address these issues, we propose a fusion localization algorithm based on particle swarm optimization. First, we construct a five-dimensional hybrid LSTM (5DHLSTM) neural network model, and the 5DHLSTM network structure parameters are optimized via particle swarm optimization (PSO) to achieve geomagnetic localization. The eight-dimensional BiLSTM (8DBiLSTM) algorithm is subsequently proposed for heading estimation in dead reckoning, which effectively improves the heading accuracy. Finally, fusion localization is achieved by combining geomagnetic localization with an improved pedestrian dead reckoning (IPDR) based on an extended Kalman filter (EKF). To validate the localization performance of the proposed PSO-5DHLSTM-IPDR method, several extended experiments using Xiaomi 10 and Hi Nova 9 are conducted in two different scenarios. The experimental results demonstrate that the proposed method improves localization accuracy and has good robustness and flexibility MDPI 2025 Article PeerReviewed pdf en cc_by_4 http://umpir.ump.edu.my/id/eprint/44411/1/Accurate%20localization%20method%20combining%20optimized%20hybrid%20neural%20networks.pdf Yan, Suqing and Luo, Baihui and Sun, Xiyan and Xiao, Jianming and Ji, Yuanfa and Kamarul Hawari, Ghazali (2025) Accurate localization method combining optimized hybrid neural networks for geomagnetic localization with multi-feature dead reckoning. Sensors, 25 (5). pp. 1-24. ISSN 1424-8220. (Published) https://doi.org/10.3390/s25051304 https://doi.org/10.3390/s25051304
spellingShingle QA75 Electronic computers. Computer science
TK Electrical engineering. Electronics Nuclear engineering
Yan, Suqing
Luo, Baihui
Sun, Xiyan
Xiao, Jianming
Ji, Yuanfa
Kamarul Hawari, Ghazali
Accurate localization method combining optimized hybrid neural networks for geomagnetic localization with multi-feature dead reckoning
title Accurate localization method combining optimized hybrid neural networks for geomagnetic localization with multi-feature dead reckoning
title_full Accurate localization method combining optimized hybrid neural networks for geomagnetic localization with multi-feature dead reckoning
title_fullStr Accurate localization method combining optimized hybrid neural networks for geomagnetic localization with multi-feature dead reckoning
title_full_unstemmed Accurate localization method combining optimized hybrid neural networks for geomagnetic localization with multi-feature dead reckoning
title_short Accurate localization method combining optimized hybrid neural networks for geomagnetic localization with multi-feature dead reckoning
title_sort accurate localization method combining optimized hybrid neural networks for geomagnetic localization with multi-feature dead reckoning
topic QA75 Electronic computers. Computer science
TK Electrical engineering. Electronics Nuclear engineering
url http://umpir.ump.edu.my/id/eprint/44411/
http://umpir.ump.edu.my/id/eprint/44411/
http://umpir.ump.edu.my/id/eprint/44411/
http://umpir.ump.edu.my/id/eprint/44411/1/Accurate%20localization%20method%20combining%20optimized%20hybrid%20neural%20networks.pdf