Yitong Song
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yitong Song has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Afeng Yang
School of Communication Engineering, Hangzhou Dianzi University, Hangzhou, China
2 shared publications - Xin Xu
College of Communication Engineering, Hangzhou Dianzi University, Hangzhou, Zhejiang, China
2 shared publications - Yingbiao Yao
College of Communication Engineering, Hangzhou Dianzi University, Hangzhou, Zhejiang, China
2 shared publications - Zhuang Wang
College of Communication Engineering, Hangzhou Dianzi University, Hangzhou, Zhejiang, China
2 shared publications - Shuo Shi
POWERCHINA Beijing Engineering Corporation Ltd., Beijing, China
1 shared publication - Zhengjing Zhou
Bank of Hangzhou Company Ltd., Hangzhou, Zhejiang, China
1 shared publication
Latest Publications
- 4 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-13
A CNN and DRL-Based Heading Estimation Method for Pedestrian Dead Reckoning Localization
Pedestrian dead reckoning (PDR) has been widely applied in indoor localization due to its independence from additional infrastructure. However, its heading estimation is highly susceptible to environmental magnetic disturbances, leading to accuracy degradation. To address this issue, this article proposes an adaptive heading estimation method based on convolutional neural networks (CNNs) and deep reinforcement learning (DRL). First, the magnetometer signals are transformed into 2-D images using continuous wavelet transform (CWT), and magnetic disturbances are identified through CNN-based classification. Then, a DRL framework based on deep deterministic policy gradient (DDPG) is used to dynamically adjust the fusion weights of the magnetometer and gyroscope headings, with a reward function that incorporates a heading smoothness constraint. Experimental results demonstrate that the proposed method achieves a magnetic disturbance classification accuracy of 99% and maintains a mean heading estimation error within 3°, outperforming benchmark algorithms. Furthermore, in real-world PDR localization experiments, the proposed adaptive heading estimation method achieves an average positioning error (PE) of 0.89 m, validating its effectiveness and robustness.
- 4 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-13
A CNN and DRL-Based Heading Estimation Method for Pedestrian Dead Reckoning Localization
Pedestrian dead reckoning (PDR) has been widely applied in indoor localization due to its independence from additional infrastructure. However, its heading estimation is highly susceptible to environmental magnetic disturbances, leading to accuracy degradation. To address this issue, this article proposes an adaptive heading estimation method based on convolutional neural networks (CNNs) and deep reinforcement learning (DRL). First, the magnetometer signals are transformed into 2-D images using continuous wavelet transform (CWT), and magnetic disturbances are identified through CNN-based classification. Then, a DRL framework based on deep deterministic policy gradient (DDPG) is used to dynamically adjust the fusion weights of the magnetometer and gyroscope headings, with a reward function that incorporates a heading smoothness constraint. Experimental results demonstrate that the proposed method achieves a magnetic disturbance classification accuracy of 99% and maintains a mean heading estimation error within 3°, outperforming benchmark algorithms. Furthermore, in real-world PDR localization experiments, the proposed adaptive heading estimation method achieves an average positioning error (PE) of 0.89 m, validating its effectiveness and robustness.
- 19 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-15
Pedestrian Dead Reckoning and Ultrawideband Fusion Positioning Algorithm Based on Adaptive Gradient Descent
In order to address the challenges of unknown initial positions and accumulated long-distance positioning errors in Pedestrian Dead Reckoning (PDR), as well as significant ranging errors caused by non-line-of-sight (NLOS) conditions in Ultra-Wideband (UWB) localization, this paper proposed a PDR and UWB fusion positioning algorithm based on adaptive gradient descent, briefly as AGD-PDR-UWB, for the indoor positioning scenarios where embedded terminals have limited resources. Firstly, the initial pedestrian position is estimated using the least-squares method based on UWB-ranging information from three base stations to solve the problem of unknown initial positions in PDR. Secondly, a novel method, which evaluates the degree of NLOS based on PDR positioning results and UWB-ranging information from two consecutive steps, is proposed to identify NLOS conditions in UWB signals. Finally, the proposed NLOS recognition method is used to adjust the weight of the loss function in the fusion positioning process, and the optimal pedestrian position that minimizes the loss function is obtained by the adaptive gradient descent method. Experimental results demonstrate that the proposed AGD-PDR-UWB algorithm outperforms four other comparative algorithms in two different positioning scenarios. Under the condition of three UWB base stations, the positioning error is within 0.5 m, effectively mitigating the cumulative errors of PDR and the positioning errors caused by NLOS conditions in UWB.
- 19 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-15
Pedestrian Dead Reckoning and Ultrawideband Fusion Positioning Algorithm Based on Adaptive Gradient Descent
In order to address the challenges of unknown initial positions and accumulated long-distance positioning errors in Pedestrian Dead Reckoning (PDR), as well as significant ranging errors caused by non-line-of-sight (NLOS) conditions in Ultra-Wideband (UWB) localization, this paper proposed a PDR and UWB fusion positioning algorithm based on adaptive gradient descent, briefly as AGD-PDR-UWB, for the indoor positioning scenarios where embedded terminals have limited resources. Firstly, the initial pedestrian position is estimated using the least-squares method based on UWB-ranging information from three base stations to solve the problem of unknown initial positions in PDR. Secondly, a novel method, which evaluates the degree of NLOS based on PDR positioning results and UWB-ranging information from two consecutive steps, is proposed to identify NLOS conditions in UWB signals. Finally, the proposed NLOS recognition method is used to adjust the weight of the loss function in the fusion positioning process, and the optimal pedestrian position that minimizes the loss function is obtained by the adaptive gradient descent method. Experimental results demonstrate that the proposed AGD-PDR-UWB algorithm outperforms four other comparative algorithms in two different positioning scenarios. Under the condition of three UWB base stations, the positioning error is within 0.5 m, effectively mitigating the cumulative errors of PDR and the positioning errors caused by NLOS conditions in UWB.