A method for mitigating NLOS error for UWB based on CNN-stacked-LSTM model
Authors:
- 30 January 2025
- journal article
- research article
- Published by IOP Publishing in Engineering Research Express
- Vol. 7 (1) , 015239
Abstract
To address the shortcomings of traditional Long Short-Term Memory (LSTM) network in Non-Line-of-Sight (NLOS) mitigation, such as the large amount of training data required and the lengthy training times, in order to enhance the model’s ability to process spatial features and multi-level features, this paper proposes an NLOS mitigation method based on Stacked Long Short-Term Memory (Stacked-LSTM) network and Convolutional Neural Network (CNN). This method combines CNN and Stacked-LSTM models to efficiently extract spatial and higher-level temporal features from the Channel Impulse Response (CIR) signal, reducing the input dimension and improving the performance of the model. The constructed CNN-Stacked-LSTM model is used to mitigate NLOS errors and reduce the impact of NLOS in the original ranging data. In the model performance validation experiment, the accuracy of the CNN-Stacked-LSTM model was improved by 4%–14% compared with the CNN-LSTM, Transformer, Attention-LSTM and LSTM models, and the training time was reduced by 0.07h compared with the traditional LSTM model. The experimental results in the two actual Ultra-WideBand (UWB) environments show that compared with the other four models, the RMSE value of the CNN-Stacked-LSTM model proposed in this paper is reduced by 19.55%–58.96% and 8.64%–45.52%, respectively. It has the best mitigation effect on NLOS and the highest positioning accuracy.Keywords
Funding Information
- National Natural Science Foundation of China (62063006)
- Leading Local Science and Technology Development Fund Project of Wuzhou (202201001)
This publication has 22 references indexed in Scilit:
- Accurate Multiclass NLOS Channels Identification in UWB Indoor Positioning System-Based Deep Neural NetworkIEEE Access, 2024
- An improved positioning method based on compensation and optimization of ultra-wideband ranging resultsMeasurement Science and Technology, 2024
- A CNN-based hybrid model to detect glaucoma diseaseMultimedia Tools and Applications, 2023
- Performance Improvement of LSTM-based Deep Learning Model for Streamflow Forecasting Using Kalman FilteringWater Resources Management, 2023
- Sustainable groundwater management using stacked LSTM with deep neural networkUrban Climate, 2023
- UWB Sensor-Based Indoor LOS/NLOS Localization With Support Vector Machine LearningIEEE Sensors Journal, 2023
- UWB Positioning System Based on LSTM Classification With Mitigated NLOS EffectsIEEE Internet of Things Journal, 2022
- Filtering Data Bins of UWB Radars for Activity Recognition with Random ForestProcedia Computer Science, 2022
- NLOS identification using parallel deep learning model and time-frequency information in UWB-based positioning systemMeasurement, 2022
- Robust ultra-wideband range error mitigation with deep learning at the edgeEngineering Applications of Artificial Intelligence, 2021