UKF-FNN-RIC: A Highly Accurate UWB Localization Algorithm for TOA Scenario
Shuning Zhang,Ershen Wang,Zhanchen Zhu,Jingyi Yi,Yongkang Wang,Enze Kuai
DOI: https://doi.org/10.1109/tim.2024.3476531
IF: 5.6
2024-10-25
IEEE Transactions on Instrumentation and Measurement
Abstract:UWB positioning has become an important branch of indoor localization technology, which has high accuracy in the ideal state but is easily affected by the environment and other factors. To address the accuracy enhancement challenges of indoor UWB positioning across various conditions, a versatile UKF-FNN-RIC algorithm is proposed, leveraging the unscented Kalman filter (UKF), feedforward neural network (FNN), and redundant information correction (RIC). First, various existing UWB positioning algorithms are analyzed, the redundant information in the UWB solution model is studied in depth, and the relationship between the redundant information and the measurement vectors is clarified, thus the error correction algorithm of gradient descent of redundant information is proposed, and the data meets the preconditions of its application through UKF and FNN transforms. The UKF-FNN-RIC algorithm is proposed in time of arrival (TOA) scenarios. Finally, simulation experiments and practical experiments are carried out for two-way ranging (TWR) TOA scenario, which show that the algorithm in the article can greatly improve the accuracy and robustness. The accuracy of fixed point in the practical experiment reaches the millimeter level, the accuracy of track determination reaches the centimeter level, and the algorithms can be easily migrated to other positioning systems.
engineering, electrical & electronic,instruments & instrumentation