A Theoretical and Empirical Integrated Method to Select the Optimal Combined Signals for Geometry-Free and Geometry-Based Three-Carrier Ambiguity Resolution

Dongsheng Zhao, Gethin Wyn Roberts, Lawrence Lau, Craig M. Hancock, Ruibin Bai

TL;DR

This paper introduces a novel method to select the best combinations of GPS signals for precise positioning using three-frequency data. By integrating theoretical analysis with real-world GPS data from various baseline lengths, they demonstrate that their optimized signal combinations can improve the success rate of ambiguity resolution by up to 8%, leading to more reliable and accurate positioning.

What Problem Does This Paper Address?

Precise positioning using GPS relies on accurately resolving carrier phase ambiguities. While triple-frequency GPS signals (L1, L2, L5) offer the potential for faster and more reliable ambiguity resolution (AR) through techniques like Three-Carrier Ambiguity Resolution (TCAR), a key challenge remains: identifying the optimal combination of these signals for both geometry-free (GF) and geometry-based (GB) AR modes. Existing methods often rely solely on theoretical predictions or simulated data, which may not accurately reflect real-world conditions like varying ionospheric and tropospheric biases across different baseline lengths. This can lead to suboptimal signal choices and reduced AR performance in practice.

Methods

The authors propose a two-stage approach:

  1. Theoretical Selection: They first develop a theoretical framework to identify promising signal combinations. This involves analyzing several criteria:
  2. Wavelength: Longer wavelengths generally lead to better AR.
  3. Ionospheric Scale Factor (ISF) and Measurement Noise: Lower values here indicate less susceptibility to ionospheric biases and higher precision.
  4. Wavelength-to-Total-Noise Ratio: A metric combining wavelength and noise to assess overall signal quality.
  5. Theoretical Success Rate: An estimation of AR performance based on statistical models.

    For GF TCAR, they systematically evaluate signal pairs, refining them based on the sum of ISFs and measurement noise, then by the wavelength-to-total-noise ratio and theoretical success rate. For GB TCAR, they build upon the GF TCAR selection, considering additional criteria like the independence of signals and the impact of tropospheric and orbit errors.

  6. Empirical Refinement: The theoretically selected signal combinations are then validated and refined using real triple-frequency GPS data collected from eleven baselines of varying lengths. This empirical testing allows for a more realistic assessment of AR performance under actual atmospheric conditions. They employ a Piecewise Cubic Hermite Interpolating Polynomial (PCHIP) method to interpolate AR success rates across different baseline lengths, visualizing how performance changes.

Datasets

The study utilizes real triple-frequency GPS data collected from nine stations in southwest Australia. This data was recorded using Trimble NetR9 receivers in static mode with a 30-second interval. Eleven baselines of varying lengths were specifically selected for the experiment. The data was processed without atmospheric corrections to directly assess the performance of different signal combinations in mitigating these biases.

Key Insights and Ideas

  • Bridging Theory and Reality: The paper's core contribution is the integration of theoretical signal selection with empirical validation using real data. This approach is crucial because theoretical models, while informative, can sometimes oversimplify real-world error sources.
  • Baseline Length Matters: The optimal signal combination is not universal. The performance of different signal pairs varies significantly with baseline length, primarily due to the changing influence of ionospheric and tropospheric biases.
  • Beyond Single-Metric Optimization: Instead of relying on a single criterion (like just the longest wavelength), the authors consider a multi-faceted approach, balancing wavelength, noise, and bias mitigation.
  • GF vs. GB TCAR Nuances: The optimal signal selection strategies differ between GF and GB TCAR, reflecting their distinct error handling mechanisms. GF TCAR focuses on eliminating biases, while GB TCAR aims to mitigate them.

Key Results

  • Improved AR Success Rates: The proposed integrated method leads to improved AR success rates. Specifically, in GF TCAR, improvements of up to 3% were observed, while in GB TCAR, gains of up to 8% were achieved at certain baseline length intervals.
  • Optimal Signal Identification: The study successfully identified specific signal combinations that perform better than traditional ones for both GF and GB TCAR across different baseline lengths. For instance, in GF TCAR, certain WL-single signal pairs showed significant improvements for shorter and longer baselines. In GB TCAR, specific NL signals demonstrated more stable performance.
  • Validation of Theoretical Assumptions: The empirical results generally supported the theoretical predictions, though some discrepancies were noted, highlighting the complexity of real-world error modeling.

Limitations and Future Work

  • Assumptions in Theoretical Models: The authors acknowledge that the theoretical analysis might involve slightly incorrect assumptions about error sources and model imprecision, which could explain some discrepancies between theoretical and real-world success rates.
  • Correlation in GB TCAR: For GB TCAR, a strong correlation between theoretical wavelength-to-total-noise ratio and real-world success rate was not always evident, suggesting that some potentially high-performing signals might have been overlooked by the theoretical selection criteria alone.
  • Future Research Directions:
    • Further research is needed to better understand the discrepancies between theoretical and empirical results, potentially by refining error source modeling.
  • Developing a theoretical method that can directly demonstrate GB TCAR success rates would be beneficial for discovering all high-performance signals.
  • Investigating the impact of other error sources like multipath, which were not explicitly modeled, could further enhance AR performance.

Summary updated

Abstract

Twelve GPS Block IIF satellites, out of the current constellation, can transmit on three-frequency signals (L1, L2, L5). Taking advantages of these signals, Three-Carrier Ambiguity Resolution (TCAR) is expected to bring much benefit for ambiguity resolution. One of the research areas is to find the optimal combined signals for a better ambiguity resolution in geometry-free (GF) and geometry-based (GB) mode. However, the existing researches select the signals through either pure theoretical analysis or testing with simulated data, which might be biased as the real observation condition could be different from theoretical prediction or simulation. In this paper, we propose a theoretical and empirical integrated method, which first selects the possible optimal combined signals in theory and then refines these signals with real triple-frequency GPS data, observed at eleven baselines of different lengths. An interpolation technique is also adopted in order to show changes of the AR performance with the increase in baseline length. The results show that the AR success rate can be improved by 3% in GF mode and 8% in GB mode at certain intervals of the baseline length. Therefore, the TCAR can perform better by adopting the combined signals proposed in this paper when the baseline meets the length condition.
BibTeX
@inproceedings{zhao2016theoretical,
  author = {Dongsheng Zhao and Gethin Wyn Roberts and Lawrence Lau and Craig M. Hancock and Ruibin Bai},
  title = {{A Theoretical and Empirical Integrated Method to Select the Optimal Combined Signals for Geometry-Free and Geometry-Based Three-Carrier Ambiguity Resolution}},
  booktitle = {Sensors},
  year = {2016},
  url = {https://openreview.net/forum?id=jRDGtTC3fD},
}

Loading related content...

Figure