A Novel Robust Trilateration Method Applied to Ultra-Wide Bandwidth Location Systems

Jiahong Li, Xianghu Yue, Jie Chen, Fang Deng2017 · DOI

TL;DR

This paper introduces a novel robust trilateration method for ultra-wideband (UWB) localization systems, specifically addressing challenges like non-line-of-sight (NLOS) and multipath fading. By incorporating a confidence-based approach to determine intersection points and a distance compensation mechanism for non-intersecting circles, the method significantly improves localization accuracy and robustness compared to conventional techniques.

What Problem Does This Paper Address?

Accurate localization is crucial for many applications, from the Internet of Things (IoT) to indoor navigation. Traditional range-based localization methods, like trilateration, rely on measuring distances to anchor nodes to pinpoint a target's location. However, real-world wireless environments, particularly those using ultra-wideband (UWB) signals, are plagued by issues like non-line-of-sight (NLOS) propagation and multipath fading. These phenomena introduce significant errors in distance measurements, leading to a common problem: the circles representing potential locations from different anchors either don't intersect at a single point or don't intersect at all. This makes conventional trilateration unreliable. Existing methods often struggle with these uncertainties, leading to reduced accuracy or increased computational complexity.

Methods

The paper proposes a "confidence-based robust trilateration method" that tackles the aforementioned issues through two main innovations:

  • Confidence-Based Intersection Determination: Instead of relying on simple proximity to a third circle's center, this method uses a confidence interval derived from the Cramér-Rao lower bound (CRLB) of the time-of-flight (TOF) measurement. This allows for a more robust selection of the "best" intersection point among multiple possibilities, even when noise is present. The method also introduces a new criterion to handle all intersection scenarios, including those where circles might not intersect.
  • Confidence-Based Distance Compensation: To address the problem of non-intersecting circles, the authors propose expanding the radius of the anchor circles. This expansion is not arbitrary but is determined by a confidence interval calculated based on the CRLB, ensuring that the compensation is statistically grounded. This allows circles that would otherwise not intersect to do so, enabling localization even in challenging scenarios.

The overall positioning method involves selecting three anchor nodes, calculating intersection points, determining the optimal intersection using the new criterion, and then averaging multiple such combinations to get the final target coordinate.

Datasets

Not discussed in the paper. The experiments were conducted in a controlled environment with seven anchor nodes and one tag node.

Key Insights and Ideas

  • NLOS and Multipath are the Primary Culprits: The paper clearly identifies NLOS and multipath fading as the main drivers of localization errors in UWB systems, leading to the critical problem of non-intersecting circles.
  • Statistical Confidence is Key: The core idea is to move beyond deterministic geometric approaches and incorporate statistical confidence intervals (derived from CRLB) to guide both intersection point selection and distance compensation. This makes the method more resilient to measurement noise.
  • A Unified Approach to Intersection Problems: The proposed intersection determination method aims to handle all possible intersection scenarios (two points, one point, no points) in a consistent and robust manner, unlike some prior methods that might only address specific cases.
  • Distance Compensation Needs a "Smart" Approach: Simply extending distances is insufficient. The paper shows that by using a confidence-based approach for distance compensation, the method can effectively create intersections where none existed, thereby improving coverage and robustness.

Key Results

The paper demonstrates significant improvements in localization accuracy and robustness:

  • Improved Accuracy: Experimental results show that the proposed method achieves a much lower Root Mean Square Error (RMSE) compared to the conventional Least Squares (LS)-based trilateration method. For instance, with fewer than four sensors, the RMSE of the proposed method is less than 0.3, while the LS method's RMSE is considerably higher.
  • Enhanced Robustness: When tested with reduced anchor ranges and the introduction of an anchor that might not intersect due to NLOS, the proposed method's RMSE is reduced, indicating its ability to compensate for NLOS and multipath effects. This robustness is superior to both LS-based and RANSAC-based trilateration methods.
  • Efficiency: The authors claim the method reduces computation burden and improves communication rates and sensor survivability, though specific metrics for these are not detailed.

Limitations and Future Work

  • Computational Complexity: While the paper claims reduced computation burden, the use of CRLB and confidence intervals might inherently increase computational complexity compared to simpler geometric methods. The paper mentions that a previous confidence-based method increased computational complexity.
  • Assumptions: The method relies on assumptions about the channel being modeled as Additive White Gaussian Noise (AWGN) and bounded ranging errors satisfying Gaussian properties, which might not always hold true in highly complex environments.
  • Future Work: The paper doesn't explicitly detail future work, but the presented method opens avenues for further research in adaptive confidence interval adjustments, integration with other sensor fusion techniques, and testing in more diverse and challenging real-world scenarios.

Summary updated

BibTeX
@misc{li2017novel,
  author = {Jiahong Li and Xianghu Yue and Jie Chen and Fang Deng},
  title = {{A Novel Robust Trilateration Method Applied to Ultra-Wide Bandwidth Location Systems}},
  year = {2017},
  doi = {10.3390/s17040795},
}
Figure