TL;DR
This paper investigates the "ill-posedness" problem in satellite navigation ambiguity resolution, particularly for China's BeiDou Navigation Satellite System (BDS). They find that BDS's unique mix of high-orbit satellites (GEO/IGSO) with slower angular velocities leads to significantly more challenging ambiguity resolution compared to GPS and GLONASS, requiring longer observation times. However, fusing BDS with GPS or GLONASS dramatically improves performance, enabling near-instantaneous fixes.
What Problem Does This Paper Address?
Satellite navigation systems like GPS, GLONASS, and BDS rely on precise measurements from satellites to determine a receiver's position. A crucial step in achieving high accuracy is "ambiguity resolution" (AR), which involves determining the integer number of carrier wave cycles between the satellite and receiver. This process can be mathematically challenging, especially when the system of equations becomes "ill-posed." Ill-posedness means that small errors in the input data can lead to large errors in the solution, or that the solution is highly sensitive to the specific geometry of the satellites and receivers.
The BeiDou Navigation Satellite System (BDS) presents a unique challenge because, unlike GPS and GLONASS which primarily use Medium Earth Orbit (MEO) satellites, BDS also includes Geostationary Orbit (GEO) and Inclined Geosynchronous Orbit (IGSO) satellites. These high-orbit satellites have much slower angular velocities relative to the Earth. This slower movement means the geometric relationship between the satellite and the receiver changes less over time. In the context of AR, this leads to stronger correlations between measurements from consecutive time epochs, exacerbating the ill-posedness of the mathematical problem and making it harder and slower to resolve ambiguities accurately. This paper aims to quantify and analyze this ill-posedness in BDS and explore solutions.
Methods
The core of the paper's methodology involves analyzing the "condition number" of the variance-covariance matrix, a standard metric for quantifying ill-posedness in linear systems. A higher condition number indicates a more ill-posed problem.
The authors employ the following key techniques:
- Double-Differential (DD) Ambiguity Resolution Model: They use the standard DD carrier phase observable model, which combines measurements from two receivers and two satellites to eliminate common errors like clock biases and atmospheric delays.
- Least-Squares Adjustment: This is used to obtain an initial "float solution" for the ambiguities and their covariance matrix.
- Least-Square Ambiguity Decorrelation Adjustment (LAMBDA) Method: This is a widely used algorithm for searching and fixing integer ambiguities from the float solution and its covariance matrix.
- Condition Number Analysis: The condition number of the normal matrix (N) is calculated at each epoch to track the evolution of ill-posedness during the AR process.
- Experimental Design: They conduct various experiments using different combinations of BDS satellites (GEO, IGSO, MEO), as well as GPS and GLONASS satellites, to compare their AR performance. They also perform fusion experiments combining BDS with GPS and/or GLONASS.
Datasets
The experiments were conducted using real-world data collected on August 29th, 2013, in Nanjing, China. The data consisted of short-distance (approximately 3m baseline) observations from BDS, GPS, and GLONASS satellites. The observation interval was 1 second, and data was collected for 20-minute sessions (1200 epochs) to ensure sufficient data for ambiguity resolution. The sky plot of visible satellites for the experiment is provided.
Key Insights and Ideas
- BDS's High-Orbit Satellites are the Culprit: The slower angular velocities of BDS's GEO and IGSO satellites are the primary drivers of its increased ill-posedness in AR. This is a direct consequence of their orbital characteristics.
- Ill-Posedness Directly Impacts AR Performance: The paper clearly demonstrates a strong correlation between the condition number (a measure of ill-posedness) and the time required for successful ambiguity resolution (FixTime). Higher condition numbers directly translate to longer FixTimes.
- Fusion is the Key to Overcoming BDS Ill-Posedness: While BDS alone struggles with AR, combining it with GPS and/or GLONASS significantly mitigates the ill-posedness. This is because the faster-moving MEO satellites from other systems help to decorrelate the observations and stabilize the solution.
- MEO Satellites Improve AR: Even within BDS, experiments show that increasing the number of MEO satellites (and decreasing GEO/IGSO) leads to a weaker ill-posedness and faster AR.
Key Results
- BDS AR is Significantly Slower: Experiments showed that for BDS alone, even with optimal satellite combinations, the time to fix ambiguities (FixTime) could be up to 120 seconds (Experiment 1). In contrast, GPS and GLONASS achieved fixes in 52 and 45 seconds, respectively.
- Fusion Achieves Near-Instantaneous Fixes: When fusing all three systems (GPS+GLONASS+BDS), the FixTime was reduced to just 2 seconds, representing an almost instantaneous ambiguity resolution.
- Condition Numbers Reflect Performance: The condition numbers for BDS experiments were orders of magnitude higher (e.g., 1.948e+008 for Experiment 1) compared to GPS (4.437e+005) and GLONASS (3.802e+005). Fusion experiments also showed significantly lower condition numbers.
- High-Orbit Only BDS is Very Challenging: When only high-orbit BDS satellites were used (Experiment 1), float ambiguities remained unstable and could not be resolved to within 1 or 0.5 cycles within the 20-minute observation period.
Limitations and Future Work
- Short Baseline Assumption: The study focused on short-baseline data (3m). The authors acknowledge that the performance might differ for longer baselines, where atmospheric effects become more dominant.
- Specific BDS Constellation: The analysis is based on the current BDS constellation. Future changes to the constellation (e.g., more MEO satellites) could alter the ill-posedness characteristics.
- Focus on Angular Velocity: While the paper highlights angular velocity as a key factor, other aspects of BDS (e.g., signal structure, inter-system biases) could also influence AR performance and were largely excluded to isolate the effect of angular velocity.
The authors suggest that for current BDS, especially when only high-orbit satellites are visible, fusing with GPS or GLONASS is the preferred approach for effective ambiguity resolution. Future work could explore the impact of longer baselines and evolving BDS constellations on these findings.
Summary updated