An Exhaustive Method of TOA-Based Positioning in Mixed LOS/NLOS Environments

Chengwen He, Jin Xiao, Liangchun Hua, Fei Ye, Xuelei LiElectronics 2025 · DOI

TL;DR

This paper introduces an "exhaustive method" (EM) for improving the accuracy and speed of wireless sensor network (WSN) localization using Time-of-Arrival (TOA) measurements in environments with both clear Line-of-Sight (LOS) and obstructed Non-Line-of-Sight (NLOS) paths. The method systematically explores combinations of base stations (BSs) to identify LOS-only subsets, leading to more robust and faster positioning without requiring prior knowledge of NLOS conditions.

What Problem Does This Paper Address?

Accurate localization is crucial for many applications, from industrial monitoring to rescue operations. Time-of-Arrival (TOA) based positioning is a popular technique, but its performance degrades significantly in real-world scenarios where signals are often blocked by obstacles, leading to Non-Line-of-Sight (NLOS) conditions. NLOS errors introduce inaccuracies in distance measurements, which can cause positioning failures. Existing methods often struggle to effectively mitigate these errors, especially without prior information about which paths are LOS or NLOS, or the magnitude of the NLOS bias. This paper aims to address this challenge by developing a method that can robustly identify LOS paths and achieve high positioning accuracy even in mixed LOS/NLOS environments, while also being computationally efficient.

Methods

The proposed "exhaustive method" (EM) operates in two main stages:

  1. Combination and Initial Solution Generation: The algorithm systematically considers all possible combinations of base stations (BSs) that could be used for positioning. For each combination, it generates an initial positioning solution and its corresponding residual vector. This process effectively partitions the BSs into those participating in the current combination and those that are not.
  2. LOS Identification and Refinement: A novel criterion is introduced to identify combinations that are likely composed solely of LOS BSs. This criterion leverages the characteristics of NLOS errors and compares residual vectors from different combinations. Solutions identified as LOS-only are then further processed. The final positioning solution is determined by taking the median of the coordinates from these selected LOS-only solutions. A key advantage is that this method does not require any prior knowledge of NLOS status, NLOS error amplitudes, or noise variances, and it only needs a minimum of three LOS BSs. The paper uses a least-squares (LS) approach for initial solution generation, but notes that other positioning algorithms could be adapted.

Datasets

Not discussed in the paper. The paper relies on simulated data generated for numerical experiments.

Key Insights and Ideas

  • Systematic Exploration is Key: The core idea of exhaustively exploring BS combinations, rather than relying on heuristics or prior knowledge, is a powerful way to overcome the ambiguity introduced by NLOS.
  • Residual Analysis for LOS Detection: The paper demonstrates that analyzing the residual vectors from different positioning subsets can effectively distinguish between LOS and NLOS paths. This is a crucial insight for robust localization.
  • No Prior Information Needed: The ability to achieve good performance without knowing NLOS status, error magnitudes, or noise variances is a significant practical advantage, as this information is often difficult or impossible to obtain in real-world deployments.
  • Median for Robustness: Using the median of selected LOS-only solutions helps to further improve robustness against any remaining minor inaccuracies.
  • Computational Efficiency: Despite the "exhaustive" nature, the method is shown to be significantly faster than many existing approaches, making it practical for real-time applications.

Key Results

  • Improved Accuracy: Numerical simulations show that the proposed EM method significantly improves positioning accuracy compared to state-of-the-art methods like SDR, SOCR, CWLS, and RSDP-New, especially when prior information is not available.
  • Reduced Computation Time: The EM method is demonstrated to be substantially faster than other compared methods, achieving an average computation time of 0.0084 seconds in one simulation, compared to others ranging from 0.3 to 0.94 seconds.
  • Robustness to NLOS: The method performs well across various scenarios with different numbers of NLOS paths (from 1 to 4 in simulations).
  • Minimum LOS Requirement: The method requires a minimum of three LOS BSs, which is a reasonable assumption for many practical scenarios.

Limitations and Future Work

  • Minimum LOS BSs: The paper acknowledges that the method's performance in scenarios with fewer than three LOS BSs needs further investigation.
  • Extreme Environments: The authors suggest that future work will focus on improving positioning performance under "extreme environments," implying scenarios with very high NLOS rates or severe signal obstructions.
  • Applicability to Other Modes: While the paper focuses on TOA, it suggests the method's core principle can be adapted to other positioning modes like AOA, TDOA, and PDOA, provided specific criteria for NLOS error characteristics in those modes are developed.

Summary updated

Abstract

This paper studies the problem of locating wireless sensor networks (WSNs) based on time-of-arrival (TOA) measurements in mixed line of sight/non-line-of-sight (LOS/NLOS) environments. To mitigate the impacts of NLOS and improve performance both in positioning accuracy and computation time, we hereby propose an exhaustive method (i.e., EM). The EM method mainly consists of two processes. In the first process, all BSs are arranged into various combinations. For each combination, a solution and its corresponding residual vector can be obtained. For each combination, all BSs can be divided into two categories: BSs that participate in positioning and BSs that do not. Therefore, the above residual vector can also be divided into two categories in each group. In the second process, combining the comparison results of two residual vectors and the characteristics of NLOS errors, we propose a new criterion to find out solutions with only LOS-BSs. Then the final solution can be obtained by further processing these solutions. This method does not require any prior information regarding NLOS status, NLOS amplitude, or noise variance, and only needs three LOS-BSs. Numerical simulation results shows that our method greatly improves the accuracy and reduces the computation time compared to state-of-art methods.
BibTeX
@inproceedings{he2025exhaustive,
  author = {Chengwen He and Jin Xiao and Liangchun Hua and Fei Ye and Xuelei Li},
  title = {{An Exhaustive Method of TOA-Based Positioning in Mixed LOS/NLOS Environments}},
  booktitle = {Electronics},
  year = {2025},
  doi = {10.3390/electronics14193764},
}
Figure