22 December 2020 in IEEE Wireless Communications Letters
Non-line-of-sight (NLOS) propagation conditions can severely degrade wireless localization accuracy due to the biases in range measurements. Machine learning methods such as support vector machine (SVM) can mitigate the effect of NLOS biases when sufficient labeled ranging measurements are available. This paper proposes a semi-supervised learning approach for NLOS identification and mitigation, which leverages low-cost unlabeled measurements by self-training to complement only a small portion of labeled ones. Experimental results show that the proposed semi-supervised approach can increase the NLOS identification probability from 90% to 94% and reduce the ranging error by 10% by exploiting the unlabeled measurements.
22 March 2022 in IEEE Wireless Communications Letters
Wi-Fi indoor localization has witnessed an increasing interest, benefiting from its low cost, easy accessibility and wide deployment. However, its applications are often restricted due to harsh multipath propagation in indoor environments. In this letter, we propose a tensor-based localization algorithm that fully exploits the multipath components (MPCs) within the channel state information. We first formulate the location parameter estimation problem as a tensor decomposition problem and ensure the uniqueness by a preprocessing step. Then, a multipath-aided localization method is proposed to achieve high-accuracy localization via fusing time-difference-of-arrival and angle-of-arrival information from all MPCs. Experimental results show that the proposed method can provide a stable sub-meter-level localization performance in multipath environments.
23 August 2021 in IEEE Wireless Communications Letters
This letter presents a tensor-based joint AoA and ToF estimation method using the channel state information for Wi-Fi systems. Our method can mitigate the estimation error caused by the signal phase offset and noise interference using linear regression and discrete wavelet transform, respectively. Moreover, the Cramér-Rao lower bound for joint AoA and ToF estimation is derived for performance evaluation. Finally, Wi-Fi measurements are taken in indoor scenarios, and the numerical results show that the proposed method can reduce the estimation errors by more than 20% in comparison with existing algorithms.
2 December 2019 in RSC Advances
Selective and controllable cataluminescence (CTL) sensors for volatile organic compounds (VOCs) are significant for chemical safety, environmental monitoring, health effects on human beings, and so forth. Most of the exploited CTL-based sensors suffer relatively low response and poor selectivity because of their high sensitivities to interferential substances. In this investigation, dendritic fibrous nano-silica & titania (DFNST) spheres have been synthesized as novel sensing materials and the corresponding DFNST-based CTL sensor has been fabricated to detect diethyl ether with high selectivity via a method of utilizing one 440 nm bandpass filter. The as-prepared DFNST hybrids not only keep the excellent dendritic fibrous morphology but also bear ca. 21 wt% catalytic titanium oxide of anatase crystalline structure. The DFNST-based sensor exhibits extremely strong CTL emission at 440 nm toward diethyl ether against other VOCs like acetone, ethyl acetate, butanol, and so forth. The high response can be attributed to the unique architectural texture of DFNST. Under the optimum parameters, ether could be easily detected in a wide range from 2.0 to 40.0 mM with a fine detection limit of 1.55 mM (S/N = 3). Furthermore, the working life of this CTL sensor is satisfactory with outstanding stability and durability, far from damaging the morphology and activity of the DFNST sensing material. In conclusion, it is expected that this novel sensing material, the relevant CTL sensor, and the approach of employing the bandpass filter will be significant for the detection of diethyl ether in actual applications.
19 October 2023 in IEEE Communications Letters
Distance estimation using ultra-wideband transmissions suffers from significant accuracy loss in non-line-of-sight propagation environments. Prevailing machine learning-based error mitigation algorithms are often environment- or device-dependent. This letter proposes an unsupervised transfer learning method based on domain adversarial training and adaptive encoder-decoder, which can retain the ranging accuracy when the environment or the device changes with low data collection workload. Experimental results show that the proposed scheme can achieve a nearly supervised-level error mitigation performance in cross-environment and cross-device applications.
17 December 2024 in IEEE Communications Letters
The emerging B5G and 6G applications have brought forth the need for high-precision indoor localization. However, the complexity of indoor environments poses significant challenges to this goal, particularly due to the presence of non-line-of-sight (NLOS) conditions and multipath effects. This letter proposes an attention-based positioning network (ABPN) that exploits fine-grained features from MIMO channel state information (CSI) by spatial attention to combat the limited receptive field of traditional convolutional neural networks (CNNs) as well as channel attention to discriminate the importance of different wireless channels. Extensive experiments, conducted on two real-world datasets, demonstrate that the proposed ABPN outperforms the popular PirnatEco, AAresCNN, MIMOnet and CLnet with an average localization accuracy improvement of over 50%.