On Mitigation of Ranging Errors for Through-the-Body NLOS Conditions using Convolutional Neural Networks

Nam, Sung-Chan; Choi, Hong-Beom; Ko, Young-Bae

doi:10.23919/icact53585.2022.9728942

Cited by 3 publications

(1 citation statement)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In paper [19], they use neural network-assisted variational Bayesian volume Kalman filter in the IMM framework to reduce LOS and NLOS errors. Papers [20] and [21] propose NLOS identification techniques using convolutional neural networks (CNNs). In [24], the LOS and NLOS data collected in channel measurements were prelabeled to train the CNN network.…”

Section: Related Workmentioning

confidence: 99%

A Constrained Filter for Indoor Localization Based on Multi-filter Fusion and NLOS Identification

Wang

Chen

2023

Preprint

View full text Add to dashboard Cite

Wireless sensor networks (WSN) are increasingly important in various fields. One of the main technologies of WSN is mobile node localization. And how to achieve precise node positioning in non-line-of-sight (NLOS) environments is a main challenge. In this paper, we propose a constrained filter NI-HCF based on NLOS identification and multi-filter fusion. We first process the measurements using Extended H-infinity filter (EHF) to obtain the predicted state vector. Subsequently, NLOS identification is conducted on measurements via residual analysis between predicted state vectors and measurements, categorizing them into LOS and NLOS groups. We send the classified measured values into Constrained Square-Root Unscented Kalman Filter (CSRUKF). This means that the measured value of LOS is processed with Square Root Unscented Kalman Filter (SRUKF). And NLOS measurements are utilized to determine a feasible domain. The sigma points outside the feasible domain can be projected into the feasible domain by solving the Semi-Definite Programming (SDP) problem. The final position estimate is obtained by the projected sigma points. Simulations and experimental tests indicate that the proposed algorithm achieves improved positioning precision in NLOS conditions and has better results compared with MPDA, IMM-EKF, RIMM, EKF and REKF.

show abstract

Section: Related Workmentioning

confidence: 99%