Abstract
Cycle slip detection and repair are crucial steps in achieving high accuracy in Global Navigation Satellite System (GNSS) data processing. The use of Global Positioning System (GPS) and BeiDou Navigation Satellite System (BDS) triple frequency observations allows for more accurate detection and repair of cycle slips compared to single or dual frequency. This study presents a moving window global search method by selecting three sets of combined coefficients to construct geometry-free (GF) models to minimize the influence of the ionosphere, using a moving window to update the standard deviation of cycle slip estimation, applying the "3 \(\upsigma\)" criterion to constrain the range, and utilizing a global search method to detect and repair triple-frequency cycle slips. Through five sets of 1 Hz GNSS data experiments, the results demonstrate the effectiveness of this method in determining the position and size of triple-frequency cycle slips while avoiding multi-value problems. The detection success rate for GPS ranges from 98.0 to 100.0%, while BDS ranges from 92.0 to 100.0%. On average, GPS achieves a detection rate of 99.2%, and BDS reaches 96.7%, which is 0.8% and 1.8% higher than the direct rounding method, respectively. Compared to existing methods, it is also effective for the vast majority of small cycle slips within 2 cycles.
Similar content being viewed by others
Introduction
GPS in the United States and BDS in China are currently the most advanced GNSS systems in the world. During the GNSS observation process, the carrier signal may be obstructed or jammed, resulting in an interruption of the phase observation. When the receiver regains the signal, the slip from the previous signal is known as a cycle slip1. The detection and repair of GNSS cycle slips is a crucial issue in high-precision positioning research, as it directly affects the reliability of positioning results2,3. Therefore, it is essential to accurately detect and repair cycle slips. With the rapid development of multi-frequency satellites, there are now more combined models of multi-frequency in GNSS positioning, providing greater opportunities for cycle slip detection and repair4,5. However, current research on cycle slip detection and repair primarily focuses on single-frequency, dual-frequency, and triple-frequency systems6, with limited discussion on five-frequency systems7.
Single frequency cycle slip detection and repair commonly use single difference (SD), double difference (DD), and triple difference (TD) methods8,9, as well as fitting methods10,11. The difference method is suitable for detecting large cycle slips and is not sensitive to small ones. The fitting method has strict requirements for the order of fitting and the amount of data, and is prone to overfitting. Statistical testing and robust testing methods12,13 are used to extract the size of cycle slips by directly rounding after residual processing using the least squares method. While these methods can detect large cycle slips, they may not be sensitive enough to detect small cycle slips, particularly when they are close to the residual size. Huang and Huo 14,15 employed wavelet analysis for cycle slip detection. Although this method can detect most cycle slips, the selection of wavelet base can be challenging since different wavelet base functions have varying detection effects. For dual-frequency cycle slip detection, Sunil B. Bisnath16 proposed using the inter-station DD or TD method, which requires at least two receivers and is not suitable for single-station positioning. Banville and Langley17 propose a ground-based (GB) cycle slip detection method and apply it to different ionospheric environments. However, this method requires a significant amount of external information, such as precise orbit parameters and clock deviation parameters. As a result, it is more suitable for post-processing than real-time processing.Zhao et al.18 propose a three-step detection method for ultra-wide, wide, and narrow lanes. The method relies on ionospheric prediction, which is suitable for environments with low ionospheric activity or sampling rates but is not suitable for single-point precision positioning (PPP). Currently, the TurboEdit method19,20,21,22,23 is a widely used dual-frequency cycle slip detection and repair method that uses Melbourne-Wübbena (MW) and Geometry-Free (GF) models. Although this method is widely used in the practical world of engineering, it is not without its own problems. For instance, the dual-frequency combination wavelength is still relatively short, and at times, the pseudo-range error can significantly affect the detection outcome. At the same time, it is insensitive to cycle slips of the same size or special combination cycle slips on the dual-frequency, and it is prone to missed or false detections. Recently, scholars24 have made new attempts to detect cycle slips through graphical structural constraints. With the development of navigation systems, many studies have been conducted on the combination of GPS and BDS multi-frequency25. Currently, the combination coefficient method is widely used to detect triple frequency cycle slips. The selection of combination coefficients must ensure the integer characteristics of cycle slips, such as long wavelengths, low noise impact, and minimal influence from the ionosphere. Dai et al.26 propose a dual GF phase model for detecting and repairing triple-frequency cycle slips. The model utilises the least squares method to remove ambiguity correlation and determine the optimal value of cycle slips. While the model can detect most cycle slips, there are still a few insensitive ones that cannot be detected or repaired. Some scholars27,28,29,30,31,32 use a combination of pseudo-range-phase and GF models to detect triple-frequency cycle slips in BDS. The optimal combination coefficient is selected and possible cycle slips are searched for. Although most cycle slips can be detected, small cycle slips may be difficult to detect due to observation noise or special combination cycle slips.
This article presents a novel moving window global search algorithm for detecting and repairing GPS/BDS triple-frequency cycle slips. The algorithm utilizes three sets of frequency coefficient combinations, implements a moving window approach, applies integer constraints to cycle slip estimation, updates its range, and employs the global search method to simultaneously detect and repair triple-frequency cycle slips. Compared to other methods, this approach offers a simpler model and superior detection capabilities. One key advantage of this method is its ability to eliminate the need to consider the integer characteristics of combined cycle slips. It also effectively separates small cycle slips from observation noise, accurately constrains the real-time standard deviation of cycle slip sequences, and determines the position and size of each frequency's cycle slip without external conditions. Furthermore, it avoids the issue of multiple values for cycle slips.
Method
Cycle slip valuation
Given the triple-frequency \({f}_{i}\left(i=\mathrm{1,2},3\right)\) of GNSS, the equations for pseudo-range and phase observations can be expressed as follows:
where, \({P}_{i}\) represents pseudo-range observation value, \({\lambda }_{i}\) represents carrier wavelength, \({\varphi }_{i}\) represents the carrier phase observation, \(\rho\) represents the geometric range, \({\varepsilon }_{{P}_{i}}\) and \({\varepsilon }_{{\varphi }_{i}}\) represent the observation noise of pseudo-range and phase respectively, \({N}_{i}\) is the integer ambiguity, \({\eta }_{i}={f}_{1}^{2}/{f}_{i}^{2}\), \({I}_{1}\) is the first-order ionospheric term parameters of \({f}_{1}\).
By combining Eqs. (1) and (2) and ignoring the influence of the ionosphere, we can calculate the difference between adjacent epochs \({t}_{1}\) and \({t}_{2}\). This will allow us to obtain the estimated cycle slip at epoch \({t}_{2}\):
where, \({\Delta N}_{i }={N}_{i }\left({t}_{2}\right)-{N}_{i }\left({t}_{1}\right)\), \({\Delta \varphi }_{i }={\varphi }_{i }\left({t}_{2}\right)-{\varphi }_{i }\left({t}_{1}\right)\), \({\Delta P}_{i }={P}_{i }\left({t}_{2}\right)-{P}_{i }\left({t}_{1}\right)\).
If the receiver receives pseudo-range \({P}_{i}\) on \({f}_{i}\left(i=\mathrm{1,2},3\right)\), it can take the average of the received multi-frequency pseudo-range observations instead of their respective \({P}_{i}\). If the receive can receive the GNSS P-code, replace the pseudo-range with the P-code. The triple-frequency cycle slip estimation represented by Eq. (3) can be expressed as:
Combined GF model
Linear combinations are employed in multi-frequency GNSS observations to mitigate the impact of certain parameters. The combination can be represented based on the triple-frequency phase observations:
where \({\upalpha }\) represents the combination coefficient. Expand the above Eq. (5) as:
If \({\alpha }_{1}+{\alpha }_{2}+{\alpha }_{3}=0\), the influence of geometric distance \(\rho\) can be eliminated, and the difference between adjacent epochs can be obtained as:
If \(({\lambda }_{1}{\eta }_{1}+{\lambda }_{2}{\eta }_{2}+{\lambda }_{3}{\eta }_{3})\Delta {I}_{1}\approx 0\), the influence of ionospheric residuals can be ignored. Combining the integer feature of cycle slips, the cycle slip estimation in Eq. (4) can be used to calculate the integer solution that satisfies Eq. (7), it is the cycle slip value.
\({\varvec{\upalpha}}\) coefficient optimization
In order to eliminate the influence of geometric distance and weaken the influence of ionospheric residuals, multiple sets of coefficients \({\upalpha }\) can be found within a certain range, meeting the following conditions:
To prevent error amplification, select \({\upalpha }\in \left[-10 10\right]\) as the range for combination, and through comparison, obtain the relevant information of the combination coefficients of GPS and BDS triple-frequency phase observations, as shown in Tables 1 and 2. The table shows that the combination coefficient can eliminate the distance term and reduce the influence of the ionosphere by up to 80% compared to its original effect \(\Delta {I}_{1}\). Based on the selected coefficients, cycle slip detection Eqs. (9) and (10) can be established, where \(j\) in \(\Delta {\varnothing }_{\alpha }^{j}\) represents the combination coefficient number.
Moving window global search method
The cycle slip will cause an outlier in \({\Delta \varnothing }_{\alpha }\), and the location of the outlier is the epoch of the suspected cycle slip. By constraining the value range of \({\Delta N}_{i}\), the integer solution satisfying Eqs. (9) and (10) can be calculated to obtain the cycle slip value of GPS and BDS. Therefore, it is important to determine the location of the cycle slip in the sequence \({\Delta \varnothing }_{\alpha }\) and the value range of \({\Delta N}_{i}\).
Figure 1 illustrates the process of identifying outliers in a sequence X(n) of length n. The first step is to set a window of width w, starting from the first element of the sequence. This window consists of the first w elements of the sequence and is used to determine if the w + 1 element is an outlier. The window is then moved one element back, and the next w elements (2 ~ w + 1) are used to determine if the w + 2 element is an outlier, and so on. This method involves a global search, which involves locating outliers within the moving window, setting constraints on parameter values, and detecting and repairing triple-frequency cycle slips.
Cycle slip detection and repair process
The algorithm flowchart is shown in Fig. 2, and the specific steps are as follows:
-
(1)
Based on the selected coefficients \(\mathrm{\alpha }\), construct \({\Delta \varnothing }_{\alpha }^{j}\) and \({\Delta N}_{i}\) for GPS and BDS, and determine the moving window width \(w\), define the XZQZ[a, b] function (rounded towards the middle of two numbers a and b).
-
(2)
Calculate the standard deviations \({m}_{{\Delta N}_{i}}\) and \({m}_{\Delta \phi }^{j}\) of \({\Delta N}_{i}({1,2},3\ldots \ldots w)\) and \({\Delta \varnothing }_{\alpha }^{j}(\mathrm{1,2},3\ldots \ldots w)\).
-
(3)
For any epoch \(t\in (\mathrm{1,2},3\ldots \ldots w-1,w)\), if any of the three combinations \(\left|\Delta {\varnothing }_{\alpha }^{j}(t)\right|>3{m}_{\Delta \phi }^{j}\), it indicates that there is a suspected cycle slip at epoch \(t\). The integer solution satisfying Eqs. (11) and (13) below is the cycle slip. Detect and calculate \(t\) epoch \({\Delta N}_{i}\) according to Eqs. (11) and (13); Simultaneously repair \({\Delta N}_{i}(t)\) and \(\Delta {\varnothing }_{\alpha }^{j}(t)\) in the original sequence according to Eqs. (12) and (14); recalculate the standard deviations \({m}_{{\Delta N}_{i}}\) and \({m}_{\Delta \phi }^{j}\) of \({\Delta N}_{i}(\mathrm{1,2},3\ldots w)\) and \({\Delta \varnothing }_{\alpha }^{j}(\mathrm{1,2},3\ldots w)\) until \(t=w\).
$$\left\{\begin{array}{c}\begin{array}{c}{\Delta N}_{1}\in XZQZ\left[{\Delta N}_{1}(t)-3{m}_{{\Delta N}_{1}} {\Delta N}_{1}(t)+3{m}_{{\Delta N}_{1}}\right]\\ {\Delta N}_{2}\in XZQZ\left[{\Delta N}_{2}(t)-3{m}_{{\Delta N}_{2}} {\Delta N}_{2}(t)+3{m}_{{\Delta N}_{2}}\right]\end{array}\\ \begin{array}{c}{\Delta N}_{3}\in XZQZ\left[{\Delta N}_{3}(t)-3{m}_{{\Delta N}_{3}} {\Delta N}_{3}(t)+3{m}_{{\Delta N}_{3}}\right]\\ {\left[\begin{array}{c}-3{m}_{\Delta \phi }^{1}\\ -3{m}_{\Delta \phi }^{2}\\ -3{m}_{\Delta \phi }^{3}\end{array}\right]}_{GPS}<{\left[\begin{array}{c}\Delta {\varnothing }_{\alpha }^{1}(t)\\ \Delta {\varnothing }_{\alpha }^{2}(t)\\ \Delta {\varnothing }_{\alpha }^{3}(t)\end{array}\right]}_{GPS}-\left[\begin{array}{ccc}{\lambda }_{1}& -6{\lambda }_{2}& 5{\lambda }_{3}\\ {\lambda }_{1}& -5{\lambda }_{2}& 4{\lambda }_{3}\\ {\lambda }_{1}& -4{\lambda }_{2}& 3{\lambda }_{3}\end{array}\right]{\left[\begin{array}{c}\Delta {N}_{1}\\ \Delta {N}_{2}\\ {\Delta N}_{3}\end{array}\right]}_{GPS}<{\left[\begin{array}{c}3{m}_{\Delta \phi }^{1}\\ 3{m}_{\Delta \phi }^{2}\\ 3{m}_{\Delta \phi }^{3}\end{array}\right]}_{GPS}\end{array}\end{array}\right.$$(11)$$\left\{\begin{array}{c}\begin{array}{c}{\Delta N}_{1}(t)={\Delta N}_{1}(t)-{\Delta N}_{1}\\ {\Delta N}_{2}(t)={\Delta N}_{2}(t)-{\Delta N}_{2}\end{array}\\ \begin{array}{c}{\Delta N}_{3}(t)={\Delta N}_{3}(t)-{\Delta N}_{3}\\ {\left[\begin{array}{c}\Delta {\varnothing }_{\alpha }^{1}(t)\\ \Delta {\varnothing }_{\alpha }^{2}(t)\\ \Delta {\varnothing }_{\alpha }^{3}(t)\end{array}\right]}_{GPS}={\left[\begin{array}{c}\Delta {\varnothing }_{\alpha }^{1}(t)\\ \Delta {\varnothing }_{\alpha }^{2}(t)\\ \Delta {\varnothing }_{\alpha }^{3}(t)\end{array}\right]}_{GPS}-\left[\begin{array}{ccc}{\lambda }_{1}& -6{\lambda }_{2}& 5{\lambda }_{3}\\ {\lambda }_{1}& -5{\lambda }_{2}& 4{\lambda }_{3}\\ {\lambda }_{1}& -4{\lambda }_{2}& 3{\lambda }_{3}\end{array}\right]{\left[\begin{array}{c}\Delta {N}_{1}\\ \Delta {N}_{2}\\ {\Delta N}_{3}\end{array}\right]}_{GPS}\end{array}\end{array}\right.$$(12)$$\left\{\begin{array}{c}\begin{array}{c}{\Delta N}_{1}\in XZQZ\left[{\Delta N}_{1}(t)-3{m}_{{\Delta N}_{1}} {\Delta N}_{1}(t)+3{m}_{{\Delta N}_{1}}\right]\\ {\Delta N}_{2}\in XZQZ\left[{\Delta N}_{2}(t)-3{m}_{{\Delta N}_{2}} {\Delta N}_{2}(t)+3{m}_{{\Delta N}_{2}}\right]\end{array}\\ \begin{array}{c}{\Delta N}_{3}\in XZQZ\left[{\Delta N}_{3}(t)-3{m}_{{\Delta N}_{3}} {\Delta N}_{3}(t)+3{m}_{{\Delta N}_{3}}\right]\\ {\left[\begin{array}{c}-3{m}_{\Delta \phi }^{1}\\ -3{m}_{\Delta \phi }^{2}\\ -3{m}_{\Delta \phi }^{3}\end{array}\right]}_{BDS}<{\left[\begin{array}{c}\Delta {\varnothing }_{\alpha }^{1}(t)\\ \Delta {\varnothing }_{\alpha }^{2}(t)\\ \Delta {\varnothing }_{\alpha }^{3}(t)\end{array}\right]}_{BDS}-\left[\begin{array}{ccc}{\lambda }_{1}& 2{\lambda }_{2}& -3{\lambda }_{3}\\ {\lambda }_{1}& 3{\lambda }_{2}& -4{\lambda }_{3}\\ {\lambda }_{1}& 4{\lambda }_{2}& -5{\lambda }_{3}\end{array}\right]{\left[\begin{array}{c}\Delta {N}_{1}\\ \Delta {N}_{2}\\ {\Delta N}_{3}\end{array}\right]}_{BDS}<{\left[\begin{array}{c}3{m}_{\Delta \phi }^{1}\\ 3{m}_{\Delta \phi }^{2}\\ 3{m}_{\Delta \phi }^{3}\end{array}\right]}_{BDS}\end{array}\end{array}\right.$$(13)$$\left\{\begin{array}{c}\begin{array}{c}{\Delta N}_{1}(t)={\Delta N}_{1}(t)-{\Delta N}_{1}\\ {\Delta N}_{2}(t)={\Delta N}_{2}(t)-{\Delta N}_{2}\end{array}\\ \begin{array}{c}{\Delta N}_{3}(t)={\Delta N}_{3}(t)-{\Delta N}_{3}\\ {\left[\begin{array}{c}\Delta {\varnothing }_{\alpha }^{1}(t)\\ \Delta {\varnothing }_{\alpha }^{2}(t)\\ \Delta {\varnothing }_{\alpha }^{3}(t)\end{array}\right]}_{BDS}={\left[\begin{array}{c}\Delta {\varnothing }_{\alpha }^{1}(t)\\ \Delta {\varnothing }_{\alpha }^{2}(t)\\ \Delta {\varnothing }_{\alpha }^{3}(t)\end{array}\right]}_{BDS}-\left[\begin{array}{ccc}{\lambda }_{1}& 2{\lambda }_{2}& -3{\lambda }_{3}\\ {\lambda }_{1}& 3{\lambda }_{2}& -4{\lambda }_{3}\\ {\lambda }_{1}& 4{\lambda }_{2}& -5{\lambda }_{3}\end{array}\right]{\left[\begin{array}{c}\Delta {N}_{1}\\ \Delta {N}_{2}\\ {\Delta N}_{3}\end{array}\right]}_{BDS}\end{array}\end{array}\right.$$(14) -
(4)
For any epoch \(t\in (w+1,w+2,w+3\ldots n)\), calculate the standard deviations \({m}_{{\Delta N}_{i}}\) and \({m}_{\Delta \phi }^{j}\) of \({\Delta N}_{i}(t-w,t-w+1,t-w+2\ldots t-1)\) and \({\Delta \varnothing }_{\alpha }^{j}(t-w,t-w+1,t-w+2\ldots t-1)\), if any of the three combinations \(\left|\Delta {\varnothing }_{\alpha }^{j}(t)\right|>3{m}_{\Delta \phi }^{j}\), it indicates that there is a suspected cycle slip at epoch \(t\). Detect and repair cycle slip according to Eqs. (11) to (14) until \(t=n\).
Data testing and analysis
Data source and experimental description
To validate the GPS/BDS triple-frequency cycle slip detection and repair method described in this article, we utilized data from two Hong Kong CORS stations HKKT and HKMW, which were observed on the 1st and 54th days of 2023, respectively. The HKKT station was observed for 5 h from 0:00 to 5:00, while the HKMW station was observed for 8 h from 5:00 to 13:00 with a sampling rate of 1 Hz. Table 3 displays the essential information of the satellites used in the experiment, and the data can be downloaded from ftp://ftp.geodetic.gov.hk/. The moving window width was set to w = 30 epochs, which is equivalent to 30 s. The experiment was divided into five groups, as shown in Table 4. During data processing, P-code was used to replace pseudo-range observations in GPS, and the average of triple-frequency pseudo-range was used to replace pseudo-range observations in BDS.
Experiment 1
The GPS and BDS data utilize different combination coefficients. Specifically, GPS uses [1 −6 5] [1 −5 4], and [1 −4 3], while BDS uses [1 2 −3], [1 3 −4], and [1 4 −3]. Furthermore, 100 sets of random cycle slips were artificially added within the range of [−10, 10]. Figures 3, 4, 5 and 6 compare the \({\Delta \varnothing }_{\alpha }^{j}\) and \({\Delta N}_{i}\) before and after cycle slips were added to the G10, C04, C12, and C16. The cycle slip detection method is compared with the direct rounding method in Table 5.
Figures 3, 4, 5 and 6 demonstrate that cycle slips have a significant impact on the amplitude of \({\Delta \mathrm{\varnothing }}_{\alpha }^{j}\) and \({\Delta N}_{i}\). The experiment achieved a detection success rate of over 94.0%, but there were some cases of false and non-detections. The analysis shows that the BDS system has more false detections than GPS due to the use of pseudo-range for cycle slip estimation calculation, while GPS uses P-code. Furthermore, it is important to note that small cycle slips may be obscured by observation noise. The experimental results show that the average success rate is 1.3% higher than the direct rounding method used.
Experiment 2
The experiment used the same original observation data and combination coefficients as Experiment 1 for each satellite, but with added cycle slip sizes of [−10 −4] and [4 10] cycles. Figures 7, 8, 9 and 10 compared the G10, C04, C12 and C16 before and after adding cycle slips. Table 6 shows the detection effect of this experiment.
From Figs. 7, 8, 9 and 10, it can be seen that before and after adding cycle slips, the amplitudes of \({\Delta \mathrm{\varnothing }}_{\alpha }^{j}\) and \({\Delta N}_{i}\) significantly increase. However, GPS has a higher success rate than BDS in terms of detection effectiveness, with BDS experiencing false detections, non-detections, and multiple detections. Despite this, the overall detection success rate is over 97.0%. The experimental results demonstrate that this method has an average success rate 1.6% higher than the direct rounding method.
Experiment 3
The experiment used the same original observations and combination coefficients for each satellite as in Experiment 1, but with the addition of cycle slip sizes of [−4 −2] and [2 4] cycles. Figures 11, 12, 13 and 14 compare the G10, C04, C12 and C16 satellites before and after adding the cycle slips. The detection results are presented in Table 7.
Figures 11, 12, 13 and 14 show that the addition of cycle slips significantly affects the change in \({\Delta \varnothing }_{\alpha }^{j}\) with a relatively large amplitude, \({\Delta N}_{i}\) shows a slight increase in amplitude. Undetected situations were observed for multiple system satellites in terms of detection effectiveness. It is analysed that the added cycle slip size is submerged by noise, resulting in false and multiple detections by BDS. The accuracy of the pseudo-range is significantly lower than that of the P-code. However, the overall success rate exceeds 96.0%. The experimental results demonstrate that this method has an average success rate 0.7% higher than that of the direct rounding method.
Experiment 4
To test the effectiveness of the method described in this article in detecting and repairing small cycle slips, cycle slip sizes of [−2 2] cycles were added in the experimental design. The observation data and combination coefficients used for each satellite in this experiment are identical to those used in Experiment 1. Figures 15, 16, 17 and 18 compare the G10, C04, C12, and C16 satellites before and after adding the cycle slips. The detection effect of this experiment is shown in Table 8.
From Figs. 15, 16, 17 and 18, it can be seen that before and after adding cycle slips, \({\Delta \varnothing }_{\alpha }^{j}\) changes significantly and the curve has sharp parts, the change in \({\Delta N}_{i}\) is gentle, with almost the same shape before and after. From the perspective of detection effectiveness, the method of using P-code to constrain GPS cycle estimation for the added small cycle slips is significantly higher than that of BDS using pseudo-range average constraint. In terms of the number of undetected cycles, BDS satellites are significantly higher than GPS satellites. The overall success rate is over 92.0%. From the experimental results, it can be seen that the average success rate of this method is 2.5% higher than that of the direct rounding method.
Experiment 5
To assess the detection and repair effects of three types of satellite cycle slips in the BDS system, this experiment analysed 4 h (09:00–13:00) of observation data from C04, C12 and C13. For comparative analysis, 100 sets of fixed epoch fixed cycle slips within the range of [-10, 10] were added. Figures 19, 20 and 21 compare \({\Delta \mathrm{\varnothing }}_{\alpha }^{j}\) and \({\Delta N}_{i}\) before and after the addition of cycle slip of C04, C12 and C13. The detection effect of this experiment is shown in Table 9.
Figures 19, 20 and 21 show that the same size cycle slip method with fixed epochs did not result in any false or multiple detections. However, there were still some cases of undetected detections. Experiment 5 demonstrates that the detection success rate of this method is consistent across different types of BDS satellites, with a success rate of over 95.0%. The experimental results demonstrate that this method has an average success rate 2.0% higher than the direct rounding method.
Analysis
(1) The results of Experiments 1 to 5 demonstrate that the moving window global search method is effective in detecting and repairing cycle slips within 10 cycles, 4–10 cycles, 2–4 cycles, and within 2 cycles, with average detection success rates of 97.2%, 98.8%, 97.5%, 95.0% and 97.0%, respectively. No issues with multiple values were observed. (2) The average detection success rate across all five experiments was 97.1%, which is 1.5% higher than the success rate of the direct rounding method at 95.6%. (3) The success rate of GPS cycle slip detection using P-code constraints is higher than that of BDS cycle slip detection using pseudo range constraints, with an average increase of about 1.8%. (4) Experiments 1 to 4 showed that using P-code instead of pseudo-range for GPS cycle slip detection resulted in an average success rate of 99.2%, which is 0.7% higher than the direct rounding method of 98.5%. The detection effect is equivalent. Additionally, using the average of triple-frequency pseudo-range instead of BDS for cycle slip detection resulted in an average success rate of 96.6%, which is 1.8% higher than the direct rounding method of 94.8%. (5) In general, the moving window global search method has a higher success rate than the direct rounding method.
Conclusion
Based on the moving window global search method, the following conclusions were drawn from the experimental analysis:
-
(1)
The GPS and BDS systems both utilize different combination coefficients to form the GF model can effectively reduce the impact of ionospheric interference. By implementing a moving window and constraining the search range using the repaired sequence standard deviation, cycle slips can be accurately detected and repaired. The success rate range from 92.0% to 100.0%, demonstrating its feasibility and effectiveness.
-
(2)
Applying the "3σ" criterion to constrain the search range in a moving window can effectively avoid the multi value problem in cycle slip detection process.
-
(3)
The moving window global search method is able to detect and repair both small cycle slips within 2 cycles and larger cycle slips exceeding 2 cycles.
The detection and repair of cycle slips in BDS with more than triple-frequency, dynamic modes, and different sampling conditions will be our future focus of research.
Data availability
All data and materials can be consulted with the corresponding author. The original data can be downloaded publicly, the download website is: ftp://ftp.geodetic.gov.hk/. Other materials can be obtained by contacting the corresponding author D.H. (email: dwhuang81@163.com).
References
Miao, Y., Sun, Z. W. & Wu, S. N. Error analysis and cycle-slip detection research on satellite-borne GPS observation. J. Aerosp. Eng. 24(1), 95–101 (2011).
Feng, S., Ochieng, W., Moore, T., Hill, C. & Hide, C. Carrier phase-based integrity monitoring for high-accuracy positioning. GPS Solut. 13, 13–22 (2009).
Lee, H. K. & Park, W. Y. High accuracy of GPS/pseudolite/INS integration: carrier phase measurement processing Issues. KSCE J. Civ. Eng. 9, 261–270 (2005).
Gu, X. & Zhu, B. An improved ambiguity resolution of three carriers in precise point positioning. IEEE Access 6, 3951–3961 (2018).
Qin, H., Liu, P., Qu, J. & Cong, L. Optimal carrier-phase integer combinations for modernized triple-frequency BDS in precise point positioning. IEEE Access 7, 177449–177459 (2019).
Yoon, Y. M., Lee, B. S. & Heo, M. B. Multiple cycle slip detection algorithm for a single frequency receiver. Sensors 22(7), 2525 (2022).
Yuan, H. et al. Real-time cycle slip detection and repair method for BDS-3 five-frequency data. IEEE Access 9, 51189–51201 (2021).
Chen, D. et al. A double-differenced cycle slip detection and repair method for GNSS CORS network. GPS Solut. 20, 439–450 (2016).
Farooq, S. Z., Yang, D. & Ada, E. N. J. Survey of cycle slip detection & correction techniques for single frequency receivers. In 2018 IEEE 18th International Conference on Communication Technology (ICCT). 957–961 (IEEE, 2018).
Liu, N., Xiong, Y. & Xu, S. Detection and repair of cycle slips using improved turboedit algorithm and Chebyshev polynomial method. Geom. Inf. Sci. Wuhan Univ. 36(12), 1500–1503 (2011).
Li, B., Liu, T., Nie, L. & Qin, Y. Single-frequency GNSS cycle slip estimation with positional polynomial constraint. J. Geodesy 93, 1781–1803 (2019).
Kirkko-Jaakkola, M., Traugott, J., Odijk, D., Collin, J., Sachs, G., & Holzapfel, F. A RAIM approach to GNSS outlier and cycle slip detection using L1 carrier phase time-differences. In 2009 IEEE Workshop on Signal Processing Systems. 273–278. (IEEE, 2009).
Fujita, S., Saito, S. & Yoshihara, T. Cycle slip detection and correction methods with time-differenced model for single frequency GNSS applications. Trans. Inst. Syst. Control Inf. Eng. 26(1), 8–15 (2013).
Dingfa, H. & Jiancheng, Z. Wavelet analysis for cycle slip detection and reconstruction of GPS carrier phase measurements. Acta Geod. Cartogr. Sin. 26(4), 353–359 (1997).
Guo, H. & Lingjuan, M. Cycle-slip detection of GPS carrier phase with methodology of SA4 multi-wavelet transform. Chin. J. Aeronaut. 25(2), 227–235 (2012).
Bisnath, S. B. Efficient, automated cycle-slip correction of dual-frequency kinematic GPS data. In Proceedings of the 13th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 2000). 145–154 (2000).
Banville, S. & Langley, R. B. Mitigating the impact of ionospheric cycle slips in GNSS observations. J. Geod. 87, 179–193 (2013).
Zhao, Q. et al. Real-time detection and repair of cycle slips in triple-frequency GNSS measurements. GPS Solut. 19, 381–391 (2015).
Blewitt, G. An automatic editing algorithm for GPS data. Geophys. Res. Lett. 17(3), 199–202 (1990).
Wu, J., Shi, C. & Fang, R. improving the single station data cycle slip detection approach TurboEdit. Geom. Inf. Sci. Wuhan Univ. 36(1), 29–33 (2011).
Cai, C. et al. The cycle-slip detection and repair of BDS based on improved TurboEdit algorithm. Geom. Inf. Sci. Wuhan Univ. 41(12), 1632–1637 (2016).
Xu, X., Nie, Z., Wang, Z. & Zhang, Y. A modified TurboEdit cycle-slip detection and correction method for dual-frequency smartphone GNSS observation. Sensors 20(20), 5756 (2020).
Zhang, X., Zeng, Q., He, J. & Kang, C. Improving turboedit real-time cycle slip detection by the construction of threshold model. Geom. Inf. Sci. Wuhan Univ. 42(3), 285–292 (2017).
Suzuki, T. GNSS odometry: Precise trajectory estimation based on carrier phase cycle slip estimation. IEEE Robot. Autom. Lett. 7(3), 7319–7326 (2022).
Zhang, X. & Li, P. Benefits of the third frequency signal on cycle slip correction. GPS Solut. 20, 451–460 (2016).
Dai, Z., Knedlik, S., & Loffeld, O. Real-time cycle-slip detection and determination for multiple frequency GNSS. In 2008 5th Workshop on Positioning, Navigation and Communication . 37–43. (IEEE, 2008).
Huang, L. et al. A new triple-frequency cycle slip detecting algorithm validated with BDS data. GPS Solut. 20(4), 761–769 (2016).
Li, B., Qin, Y., Li, Z. & Lou, L. Undifferenced cycle slip estimation of triple-frequency BeiDou signals with ionosphere prediction. Mar. Geod. 39(5), 348–365 (2016).
Cui, L., Chen, D., Wang, X. & An, J. C. BDS three-frequency pseudo-range/carrier linear combination cycle-slip detection and correction. Eng. Surv. Mapp. 27(12), 1–3 (2018).
Zhao, D., Roberts, G. W., Hancock, C. M., Lau, L. & Bai, R. A triple-frequency cycle slip detection and correction method based on modified HMW combinations applied on GPS and BDS. GPS Solut. 23, 1–14 (2019).
Li, D., Dang, Y., Yuan, Y. & Mi, J. A new cycle-slip repair method for dual-frequency BDS against the disturbances of severe ionospheric variations and pseudoranges with large errors. Remote Sens. 13(5), 1037 (2021).
Zhang, C., Dang, Y., Wang, X., Quan, J. & Yu, Z. A real-time cycle slip detection and repair method based on ionospheric delay prediction for undifferenced triple-frequency BDS signals. IEEE Access 9, 69999–70013 (2021).
Acknowledgements
This study was financially supported by Subject Construction Fund of Yunnan College of Business Management (NO: 2022XKJS06) . The authors would like to thank the anonymous reviewers for their helpful comments.
Author information
Authors and Affiliations
Contributions
J.W. completed the design and writing of the calculation program, compiled the calculation results, and wrote the manuscript; D.H. provided guidance on the process of the paper; All authors discussed the results and reviewed the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Wang, J., Huang, D. GPS/BDS triple-frequency cycle slip detection and repair based on moving window global search method. Sci Rep 14, 6615 (2024). https://doi.org/10.1038/s41598-024-57063-5
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-024-57063-5
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.