The Application of Robust Least Squares Method in Frequency Lock Loop Fusion for Global Navigation Satellite System Receivers

Mengyue Han, Qian Wang, Yuanlan Wen, Min He, Xiufeng He2020 · DOI

TL;DR

This paper introduces a novel Frequency Lock Loop (FLL) tracking method for Global Navigation Satellite System (GNSS) receivers that combines a Vector FLL (VFLL) with a robust least squares approach. This fusion significantly improves tracking accuracy and robustness, especially in challenging environments with weak or noisy satellite signals, by adaptively weighting signals based on their quality.

What Problem Does This Paper Address?

Global Navigation Satellite Systems (GNSS) are crucial for positioning, navigation, and timing. A key component of GNSS receivers is the Frequency Lock Loop (FLL), responsible for tracking the carrier frequency of satellite signals. However, traditional FLLs struggle with accuracy and reliability in complex environments characterized by signal fading, multipath, and interference. This leads to degraded positioning performance. Existing methods often focus on improving FLL structures or using FLLs to assist other loops, but there's a need for a more integrated approach that can dynamically adapt to varying signal qualities and mitigate the impact of poor-quality signals on overall performance.

Methods

The core innovation of this paper lies in the fusion of a Vector FLL (VFLL) with a robust least squares method. Here's a breakdown:

  • VFLL: Unlike traditional scalar FLLs, VFLLs process signals from multiple channels simultaneously, leveraging correlations between them for improved dynamic performance and the ability to re-acquire weak signals.
  • FLL Segmentation: The authors first analyze the probability density function of the FLL discriminator output under different signal-to-noise ratios (SNRs). This analysis allows them to define distinct "segments" for strong, general, and weak signals.
  • Robust Least Squares: This method is applied to assign weights to different satellite signals. Instead of using a fixed weighting scheme, the weights are determined at the signal level based on the segmentation results. This means signals with better quality (stronger SNR) are given higher importance.
  • IGGIII Weight Function: A modified IGGIII weight function is used, adapted to the signal level segmentation. This function assigns weights based on the Continuous Wave Noise Ratio (CNR) of each channel.
  • Kalman Filter Integration: The VFLL, combined with the robust least squares weighting, is integrated into a Kalman filter framework. This allows for accurate estimation of carrier frequency and other parameters, with the equivalent weight matrix dynamically adjusted based on signal quality.

Datasets

Not discussed in the paper. The authors conducted simulations using MATLAB software and generated synthetic satellite data.

Key Insights and Ideas

  • Signal-Level Weighting is Crucial: The paper strongly advocates for determining signal weights at the signal level rather than the observation level. This allows for more precise adaptation to varying signal strengths and a better distinction between good and bad signals before they impact parameter estimation.
  • FLL Segmentation as a Foundation: The concept of segmenting the FLL based on signal strength is a novel way to categorize signals and forms the basis for adaptive weighting. This segmentation effectively identifies "good," "suspicious," and "gross error" signals.
  • Robustness Against Gross Errors: The proposed VFLL model, by adaptively down-weighting or even eliminating signals with poor quality or gross errors, demonstrates significantly improved robustness compared to traditional methods.
  • Synergy of VFLL and Robust Least Squares: The combination of VFLL's dynamic tracking capabilities with the robust least squares' ability to handle noisy data and assign intelligent weights creates a powerful tracking system.

Key Results

  • Improved Tracking Accuracy: In open environments, the proposed VFLL model achieved frequency errors around 0.2 Hz, a significant improvement over traditional SFLL and VFLL.
  • Enhanced Robustness in Complex Environments: Under partial occlusion and with gross errors, the proposed VFLL controlled positioning errors within 0.3 m, while traditional SFLL failed to converge and the original VFLL had errors around 1 m.
  • Effective Handling of Weak Signals: The model successfully mitigated the impact of low-quality signals, preventing them from degrading the overall positioning accuracy.
  • Superiority of Signal-Level Weighting: Comparisons showed that signal-level weighting led to more accurate tracking parameters and positioning results than observation-level weighting.

Limitations and Future Work

  • Simulation-Based: The study relies on simulations, and real-world testing with actual GNSS receivers would be beneficial to fully validate the findings.
  • Specific Environment Focus: While tested in simulated complex environments, further validation across a wider range of real-world challenging scenarios (e.g., urban canyons, jamming) would be valuable.
  • Future Work: The authors suggest that this method has promising application perspectives in complex and harsh environments, implying potential for further refinement and deployment in operational GNSS receivers.

Summary updated

BibTeX
@misc{han2020application,
  author = {Mengyue Han and Qian Wang and Yuanlan Wen and Min He and Xiufeng He},
  title = {{The Application of Robust Least Squares Method in Frequency Lock Loop Fusion for Global Navigation Satellite System Receivers}},
  year = {2020},
  doi = {10.3390/s20041224},
}
Figure