An Improved Hatch Filter Algorithm towards Sub-Meter Positioning Using only Android Raw GNSS Measurements without External Augmentation Corrections
Round 1
Reviewer 1 Report
The literature review must includes the Why the work done, the What (claim of the paper reviewed) and How the work achieved. The LR is fragmented with mostly WHAT being claimed, and to be frank, a bit think considering that this type of filters has been around and in abundance.
The English needs to improve with many typos that should not be there. The content is useful for publications and stimulating future research, especially that which combines the signals of more than one localisation transmission. Accuracy comparison can be improved with more inclusive work (think of a solution rather than a filer).
Author Response
Response to Reviewer 1 Comments
Dear Reviewer:
Thanks for your comments concerning our manuscript entitled “An Improved Hatch Filter Algorithm towards Sub-meter Positioning using only Android Raw GNSS Measurements without external augmentation corrections” (ID: remotesensing-530654). Those comments are all valuable and very helpful for revising and improving our paper. We have studied comments carefully and have made corrections which can hopefully meet your request. Revised portions are marked in red in the paper. The main corrections in the paper and the responses to the reviewer’s comments are as flowing:
Point 1: “The literature review must includes the Why the work done, the What (claim of the paper reviewed) and How the work achieved. The LR is fragmented with mostly WHAT being claimed, and to be frank, a bit think considering that this type of filters has been around and in abundance.”
Response 1: Thank you for your suggestions. According to your opinion, we have rearranged and organized the literature review part of the introduction in our paper to illustrate Why the work is done, the What (claim of the paper reviewed) and How the work achieved in detail. We reunite the structure of the literature review so that it can be clear. The detailed explanation will be as follows:
Firstly, we have combined the first two paragraphs into one paragraph to illustrate Why the work is done clearly. The accuracy of the GPS modules embedded in smartphones and the need for sub-meter-level positioning are introduced respectively in the beginning. Then we introduce the possibility of processing raw measurement with the release of Android 7 and the huge number of smartphones in the mass market, based on which we made relative research about sub-meter-level positioning by making use of raw Android-device GNSS measurements. That’s why we do our research and organize our paper.
Secondly, we introduce several works that mainly focus on traditional PPP (Precise Point Positioning) and some relative research work in the second paragraph. However, a high-precision position obtained by the PPP method needs an online statement and more electric power. Therefore, we illustrate the purpose of our study for the reason that a high-precision position without any additional corrections for Android-device is worthy of research. The relative description has been written in our paper.
Thirdly, we have cited the other three following papers in the first paragraph to illustrate the smartphone GNSS performance under different conditions:
a) Warnant, René, De Vyvere, Laura Van & Warnant, Quentin. (2018), Positioning with Single and Dual Frequency Smartphones Running Android 7 or Later, ION GNSS+ 2018, Miami, Florida, September 2018, 284-303.
b) Gao, Han, and Groves, Paul. (2018), Environmental Context Detection for Adaptive Navigation using GNSS Measurements from a Smartphone, Journal of The Institute of Navigation, 65(1), 99-116.
c) Specht, Cezary & Dąbrowski, Paweł & Pawelski, Jan & Specht, Mariusz & Szot, Tomasz. (2019). Comparative Analysis of Positioning Accuracy of GNSS Receivers of Samsung Galaxy Smartphones in Marine Dynamic Measurements. Advances in Space Research. 63. 3018-3028.
Fourthly, we have discussed the previous work about Hatch filter algorithm and its improvements on phase-smoothed pseudorange about geodetic receivers. Based on the above research work, we propose an enhanced algorithm based on the Hatch filter and the Kalman filter algorithms to achieve sub-meter-positioning accuracy towards Android-devices to explain how the work achieved, which is to explain how the work achieved.
Fifthly, we briefly introduce our research process in the fourth paragraph. We first briefly review the release of GNSS observations after Android N and the respective data collection application. Then we introduce our improvements combined with classical Hatch filter and Kalman filter. Meanwhile, we carried on the relative validations and the quality assessments. Finally, the experiments, discussions and conclusions are presented. That’s to explain how the work achieved.
As your opinion and the third paragraph shows, we have found that many types of Hatch filter and its improvements presented in previous work. However, due to lots of cycle slips, outliers in raw GNSS measurements from Android devices and clock bias inconsistency between phase and code observations towards raw Android-device GNSS measurement different from that of geodetic receivers, previous Hatch filter and its improvements are not applicable to raw Android-device GNSS measurement. Meanwhile, few relative research work about Hatch filter and its improvements for Android-device GNSS measurement has been proposed. Therefore, in our paper, the improved Hatch filter algorithm presented is mainly towards single-frequency Android devices and can improve positioning accuracy compared with GNSS chipset solution.
Point 2: “The English needs to improve with many typos that should not be there.”
Response 2: Thank you for your serious review. We have revised the whole manuscript carefully and tried to avoid any grammar or syntax error. In addition, we have asked several colleagues who are skilled authors of English language papers to check the English. We believe that the language is now acceptable for the review process.
Point 3: “The content is useful for publications and stimulating future research, especially that which combines the signals of more than one localisation transmission.”
Response 3: Thank you for your affirmation!
Point 4: “Accuracy comparison can be improved with more inclusive work (think of a solution rather than a filer).”
Response 4: Thank you for your comment. According to your opinion, accuracy comparison with other practical solutions will be more inclusive work. Therefore, we add the relative description and explanation after Topic Results. The added description is as follow:
“In this part, firstly we make algorithm validation with survey-grade receivers to verify the specific suppression effect of the three threshold detections on the ionosphere delay cumulative error, cycle slip and outliers. Secondly, we make a quality assessment for Nexus 9 Raw GNSS data to verify that pseudorange noise of the Nexus 9 tablet can be reduced by phase-smoothed pseudorange. Finally, GNSS chipset solution inner Nexus 9 tablet, single point positioning solution by software RTKLIB and traditional Hatch filter solution as comparisons are used to compare the accuracy of the result accessed by TT-SD Hatch filter algorithm in static test and kinematic test, respectively.
”
That means that to compare the accuracy of the result accessed by our improved Hatch filter algorithm, we choose GNSS chipset solution inner Nexus 9 tablet, single point positioning solution by software RTKLIB and traditional Hatch filter solution as comparisons. The reason why we choose solutions above is that we have made comparisons with the most widely used solutions (GNSS chipset solution) and classical solutions (single point positioning solution and traditional Hatch filter solution). Without external augmentation corrections, the comparison is described in our paper as follow:
“…The static experiment shows that the horizontal and vertical position errors of TT-SD Hatch filter solution are about 0.6 and 0.8 m in terms of RMS, respectively, after taking a few minutes to convergence. Conversely, the horizontal and vertical positioning errors of chipset solution are approximately 2.6 m and 2.5 m and varies with time. Both the SPP solution and the traditional Hatch filter solution have the position RMS exceed 10 m in horizontal and vertical components….”
“…The kinematic experiment shows that the TT-SD Hatch filter solution can converge after a few minutes, and the 2D error is about 0.9 m, which is about 64%, 89% and 92% lower than that of the chip solution, the traditional Hatch filter solution and SPP solution, respectively….”
Author Response File: Author Response.docx
Reviewer 2 Report
General comments:
The author proposed the improved Hatch filter with single difference and three thresholds from smartphone raw GNSS measurements. It can achieve better positioning performance in both static and dynamic scenarios than smartphone chipset solution. The paper overall is well structured and worth publishing after minor correction.
Detailed comments:
1. Some words in the tiles are in capital while some are not.
2. Page 1, Line 40. “GPS-friendly conditions” is too vague. You actually means the open conditions that most received satellite signals are LOS. Cite the following papers regarding to the smarphone GNSS performance under different conditions:
a) Warnant, René, De Vyvere, Laura Van & Warnant, Quentin. (2018), Positioning with Single and Dual Frequency Smartphones Running Android 7 or Later, ION GNSS+ 2018, Miami, Florida, September 2018, 284-303.
b) Gao, Han, and Groves, Paul. (2018), Environmental Context Detection for Adaptive Navigation using GNSS Measurements from a Smartphone, Journal of The Institute of Navigation, 65(1), 99-116.
c) Specht, Cezary & Dąbrowski, Paweł & Pawelski, Jan & Specht, Mariusz & Szot, Tomasz. (2019). Comparative Analysis of Positioning Accuracy of GNSS Receivers of Samsung Galaxy Smartphones in Marine Dynamic Measurements. Advances in Space Research. 63. 3018-3028.
3. Page 4, Equation 5. Explain how to select the reference satellite.
4. Page 5, Line 171. “the ionosphere delay errors are gradually accumulated with time.” The words were stated again in Line 175.
5. Page 5, Line 181. What is error propagation law? How does it apply into determining the threshold?
6. Equation 12 and 13, suggest bold symbols for vectors and matrixes throughout the paper to avoid misleading.
7. The setting of state error covariance matrix of Kalman filter is missing.
8. Line 274, suggest to add reference for “Klobuchar and Saastamoinen models”.
9. Figure 3, in determining whether the user is static or not, how did you do that? Using Doppler or velocity solution? Any thresholds have been applied? These should be stated in the paper.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
This paper presents a "hands-on" improved navigation solution for mobile devices - with raw measurements support. The use of Hatch filter combined with Kalman filter is a natural and efficient way to improve raw measurements.
The paper discusses an important topic and is "reader-friendly" - rather easy to follow.
The result shows improvement over existing "mobile-device-level" positioning accuracy.
Major comments:
Being a "hands on paper" the authors should experiment also newer Android devices see: https://developer.android.com/guide/topics/sensors/gnss
Dual Frequency GNSS: (L1+L5) are available for over a year now: e.g., mid-range devices such as Xiaomi Mi 8 supports dual frequency (L1+L5): see:
https://www.gsa.europa.eu/newsroom/news/world-s-first-dual-frequency-gnss-smartphone-hits-market,
see the following list for additional info: https://www.xda-developers.com/dual-frequency-gnss-important-location-feature-your-phone-probably-missing/
Trying to improve results such as: https://www.mdpi.com/2079-9292/8/1/91 seems highly interesting and doable.
Minor remarks:
1. Where is the Tablet GNSS-antenna located? it might affect the results in +-10 cm easily.
2. Having the raw measurements available as a benchmark - would suggest an interesting contribution.
3. Comparring the results with the new F9 Ublox GNSS receiver would be interesting to you can actually compare the existing positioning solution (single device no RTK) with your improved filters.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 3 Report
Please consider publishing a benchmark with the GNSS raw measurements and the improved (filtered) pseudo ranges.
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.