Get the latest tech news

Bias and Deviation Weighted Graph Search for NLOS Indoor RTLS Calibration

Recently, UWB-based technology providing centimeter-level accuracy has been developed and widely utilized in indoor real-time location tracking systems. However, location accuracy varies due to factors such as frequency interference, collisions, reflected signals, and whether line-of-sight (LOS) conditions are met, and it can be challenging to ensure high accuracy in specific environments. Fortunately, when anchor positions are fixed, the locations of large obstacles such as columns or furniture remain relatively stable, leading to similar patterns of positioning bias at specific points. This study proposes an algorithm that corrects inaccurate positioning to more closely reflect the actual location based on bias and deviation maps generated using natural neighbor interpolation. Initially, positioning bias and deviations at specific points are sampled, and bias and deviation maps are created using natural neighbor interpolation. During location tracking, the algorithm detects candidate clusters and determines the centroid to estimate the actual location by applying the bias and deviation maps to the measured positions derived through trilateration. To validate the proposed algorithm, experiments were conducted in a non-LOS (NLOS) indoor environment. The results demonstrate that the proposed algorithm can reduce the positioning bias of a UWB-based RTLS by approximately 71.34% compared to an uncalibrated system.

Additionally, the experiment aims to demonstrate the superiority of the proposed method by comparing it with other time series data calibration techniques, such as moving average, a Kalman filter, and robust local regression [ 34, 35, 36]. This work was supported by Institute of Information and communications Technology Planning and Evaluation (IITP) under the Artificial Intelligence Convergence Innovation Human Resources Development (IITP-2024-RS-2023-00254177) grant funded by the Korean government (MSIT). [ Google Scholar] [ CrossRef] Pei, Y.; Chen, R.; Li, D.; Xiao, X.; Zheng, X. FCN-Attention: A deep learning UWB NLOS/LOS classification algorithm using fully convolution neural network with self-attention mechanism.

Get the Android app