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.
Ultra-wideband technology has the merits of high temporal resolution and stability, and it has been widely used for high-accuracy localization and tracking. However, most ultra-wideband localization systems need multiple anchors for trilateration, which results in high system cost, large messages overhead, and insufficient extraction of information. In this paper, we propose a single-anchor localization (SAL) mehtod that achieves high-accuracy multi-agent localization with high efficiency. In the proposed method, the anchor broadcasts the first two messages and then each agent responds one message to the anchor (quasi-)simultaneously. Based on the received message with superpositioned agent responses, the time-of-flight and angle-of-arrival information from all agents to the anchor can be extracted altogether. We implement the localization system in two indoor environments, and show that the proposed method can achieve decimeter-level accuracy for multiple agents using three messages. Our method provides design guidelines for high-accuracy and high-efficiency multi-agent localization systems.
2 February 2024 in IEEE Transactions on Wireless Communications
Location awareness plays an increasingly important role in wireless network applications. However, accurate localization in complex indoor environments remains challenging for existing radio frequency (RF)-based systems, among which the ultra-wide bandwidth (UWB) technology ranks to be the most promising one due to its capability in providing channel information with fine time resolution. In this paper, we propose a multipath-assisted single-anchor localization framework that can provide high-accuracy positional information in complex indoor environments. Specifically, a deep variational learning method is proposed to produce calibrated estimates of position-related parameters, including distance, time-difference-of-arrival and angle-of-arrival, which are then fed into a multipath-assisted single-anchor localization algorithm. The proposed method is implemented on self-built UWB transceivers and assessed with real-world data from an indoor measurement campaign. Extensive experimental results show that the proposed method outperforms conventional machine learning-based error mitigation approaches and can achieve 0.15m root mean square position error in non-line-of-sight scenarios.
The next-generation communications impose requirements on integrated sensing and communication. However, the non-line-of-sight propagation in indoor complex environments poses great challenges to common localization techniques. In this letter, we propose a signal denoising network based on the transformer and temporal attention to improve the angle-of-arrival estimation accuracy. In the proposed network, the channel impulse response is denoised and reconstructed to mitigate errors. Then, two database are constructed based on self-built ultra-wideband transceivers in indoor environments for validation. Results show that the proposed network outperforms other machine learning methods in terms of angle-of-arrival estimation accuracy.
Device-free localization (DFL), an important aspect in integrated sensing and communication, can be achieved through exploiting multipath components in ultra-wide bandwidth systems. However, incorrect identification of multipath components in the channel impulse responses will lead to large angle-of-arrival (AoA) estimation errors and subsequently poor localization performance. This letter proposes a learning-based AoA estimation method to improve the DFL accuracy. In the proposed method, we first design a classifier to identify the multipath components and then exploit the phase-difference-of-arrival to mitigate the AoA estimation error through a multilayer perceptron. Our learning-based method is validated using the datasets collected by ultra-wide bandwidth arrays, which significantly outperforms conventional methods in terms of AoA estimation and localization performance.
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.