A Lightweight CIR-Based CNN With MLP for NLOS/LOS Identification in a UWB Positioning System

Abstract

Implementing line-of-sight (LOS) and none-line-ofsight (NLOS) identification in ultra-wideband (UWB) systems is crucial. Convolutional neural network (CNN) based identification methods can extract higher-level features automatically, but they are based on channel impulse response (CIR)-turned image ingested features that impose calculation complexity and do not make use of manual features due to the data inundation risk. In this letter, we propose a novel multilayer perceptron (MLP)based LOS/NLOS identification algorithm that can utilize both manually extracted features and feature from CNN based on raw CIR inputs with only 7.39% calculation complexity as compared to the traditional image-based CNN. Three experiments, at a teaching building, an office, and an underground mine, were conducted to verify the proposed method's performance. Our proposed features are conducive to LOS/NLOS identification, especially the proposed raw CIR-based feature from the CNN, achieving 26.9% improvement over existing manual features. Furthermore, the proposed method outperformed the traditional image-based CNN with an improvement of 44.16%.

PEiD: Precise and Real-Time LOS/NLOS Path Identification Based on Peak Energy Index Distribution

“…Distinct from the approaches based on statistical parameters, UWB-based solutions focus on extracting features from CIR [24][25][26][27][28][29]39]. Early works tend to assess the energy decay and time delay caused by NLOS obstructions.…”

“…Based on commercial UWB chips, Ferreira et al [29] further design thirteen features and select a subset to achieve the best performance. More recent works take advantage of deep neural networks, such as Multi-Layer Perceptions (MLP), Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM) networks, to learn more complex features from CIR [25,26,28]. However, the complex model training also raises the issue of laborious data collection.…”

“…This type of solution is simple and effective but requires reliable measurement of range, Time of Arrival (ToA), and Angle of Arrival (AoA). Another group of works exploits the high-precision Channel Impulse Response (CIR) provided by Ultra-Wideband (UWB) technology [24][25][26][27][28][29]. It offers a feature map of superior time resolution that contains rich Multi-Path Components (MPCs) of received signals.…”

Adaptive Extended Kalman Filter Position Estimation Based on Ultra-Wideband Active-Passive Ranging Protocol

“…A sizable amount of research has been conducted by using the raw CIR: NLoS detection via Capsule Networks [16], proposing an NLoS-induced outlier-aware positioning method based on multilayer perception [17], signal decomposition by One-Dimensional Wavelet Packet Analysis in conjunction with Convolutional Neural Networks (CNN) [18], Transformer deep learning model [19], combining the Multilayer Perceptron with CNN to reduce calculation complexity [20], overcoming the problem of site-specific models by conducting Long Short-Term Memory training to predict NLoS error magnitude and variance of measurements [21], to name a few of the latest. In addition to ML and deep learning, other methods utilizing the raw CIR are explored: NLoS detection using fuzzy comprehensive evaluation [22], a weighted particle filter based on probability density functions of Line-of-Sight (LoS)/NLoS correlation coefficients [23], and adaptively selecting the optimal anchors based on the channel quality indicators [24].…”

“…Moreover, employing ML models requires large amounts of high-quality training data, which makes the data-gathering process tedious, while the training and implementation of models could turn out computationally expensive [20], [25].…”