A Real-Time Linear Prediction Algorithm for Detecting Abnormal BDS-2/BDS-3 Satellite Clock Offsets

Round 1

Reviewer 1 Report

This paper first discusses some typical abnormal types of satellite clock offsets, and then proposes a real-time linear prediction algorithm for detecting abnormal BDS-2/BDS-3 satellite clock offsets, and compares the experimental results with grey model and ARIMA model from two aspects of clock detection quality and clock prediction performance. Finally, the kinematic PPP with the real-time estimated satellite clock products is carried out to verify the effectiveness of the proposed method. The topic has the value of scientific research and engineering application. The algorithm flow is reasonable, and the experimental results are credible, which have certain reference significance. On the whole, this is a high-quality academic paper with standard writing, smooth language and clear logic, but there are also some points to be improved and perfected. Suggestions for modification are as follows:

1、Page2, Line87. Please delete “Research manuscripts ... to publication”. This sentence has nothing to do with this article.

2、Page3, Line119. Please revise “and is ... frequency data” to ”and n is ... frequency data”. Note that the symbol interpretation should be complete.

3、Page4, Line161. Please delete “To verify ... in detail”.

Page7, Line247. Please delete “The frequency ... corresponding outliers”.

These two sentences are repeated with the above content.

4、Page8, Line262. Please delete “The abnormal ... proposed algorithm”.

Page8, Line279. Please delete “The abnormal ... detection module”.

Page9, Line303. Please delete “Compared to ... orbit types”.

Sentences of similar meaning have appeared above, and these sentences should not be placed in the picture description.

5、Page5, Line199. This statement “The accuracy ... orbit maneuver” is wrong.

6、Fiugre4, Fiugre5, Fiugre6. Please supplement the units of the ordinate “Frequency”.

7、Fiugre10. Please supplement the abscissa.

8、Table1. Please supplement the average STD accuracy before and after the abnormal clock detection, and the improvement in Table 1 should be mentioned in the interpretation of the results results in the following text.

9、Table2. Please supplement the average RMS accuracy improvement. The average RMS accuracy in the following results interpretation cannot be directly drawn from Table 2, so appropriate explanation is required.

10、Keywords. The keyword ”Real-time” is lack of pertinence, so it is recommended to replace.

11、Introduction. The grey model and ARIMA model are selected as control experiments in the verification part, while the introduction part lacks the introduction of grey model.

12、Experiments, Results, and Discussion. Chapter 3 is excessive, while the first section introduces the algorithm proposed in this paper, suggesting as a separate chapter.

13、Performance of the real-time linear clock prediction algorithm. Please explain briefly why the 20-minute clock offset data is used in the fitting model.

 

Additional suggestions:

1、This paper discusses the detecting outliers of BDS satellite real clock offsets. The authors proposed a linear moving prediction model and a new method to detect the "abnormal value" of the satellite clock error in the original satellite clock offsets and clock frequency. Based on detecting predicted epochs one by one, to finish the delete "abnormal value" work. But, there are some questions need to be explained explicitly.

2、In line 283. the improved STD accuracy of the real-time estimated satellite clock offsets is assessed in this section after the abnormal clock values are removed. STD indicates the degree of dispersion, however, smaller STD does not represent the high accuracy of clock offsets. In contrast, RMS is more suitable as the evaluation index of model performance.

3、In Figures 4 and 6, The numbers in the picture are not completely displayed. Please update the picture again. It is worth mentioning that the words in the figures are too large, please reduce the size of the word in the pictures appropriately, which may be more beautiful.

4、In figure 7, “DOY” is the first time used in this paper, you should explain what is this in the article.

5、As to the detection of the "abnormal" values, epoch-difference algorithm is also be used. Why the so-called "linear moving short-term prediction model" is presented in this paper?

6、In line 175. " The clock difference time series of the C05, C06, and C24 satellites from GEO, IGSO, and MEO, respectively is first presented in Figure 1." What mean of the "difference"? Does the difference denote the difference between the final products and satellite-based RTS? In BDS real-time service (RTS), the satellite clock correction is computed based on the SSR and Broadcast ephemeris, which is obtained using L-band two-way satellite time-frequency transfer (TWSTFT) and exhibits poor self-consistency. If the difference denotes the difference between the final products and satellite-based RTS, the so-called "outliers" may be normal. Besides, the clock difference algorithm should list a reference.

7、Figure 5 show the BDS C02 has phase modulation operating. I wonder whether the observation data and positioning are available during the period.

 

Comments for author File: Comments.pdf

Author Response

Thanks for your questions and comments. We have revised the manuscript according to your suggestions. Besides, we are very happy to make a further revision if we fail to answer some questions and comments.

Author Response File: Author Response.docx

Reviewer 2 Report

1、The hour-boundary errors indeed are different for MEO, GEO and IGSO. How much errors for MEO, GEO and IGSO respectively?

2、From Figure 2, we see the STD accuracy of the IGSO satellite is worse than that of MEO. The  in Equ(6) should be different for IGSO and MEO?

3、Line 200, “The accuracy of the real -time clock is related to the satellite orbit type due to poor orbital error, such as GEO satellite orbit maneuver.” This may be and please correct.

4、In PPP validation, the satellite clock is the estimated clock or the predicted clock?

5、In Figure 13, how many satellites are eliminated from the UTC time 14:00 to 17:00?

6、This paper mainly uses the clock difference products of WHU for analysis and verification. It is recommended to give a more detailed description of the generation strategy of the WHU clock product.

7、It is suggested that the deficiencies of the proposed algorithm are also explained so that readers can understand and use the method more fully.

8、There are some mistakes in english expression which should be corrected:

①        p.2, l.81: is should be are

②        Equ (1), y shoule be yi

③        p.2, l.88-92: These descriptions seem irrelevant.

④        p.3, l.114:  “the epoch” should be “the epoch i”.

⑤        p.3, l.119:  should be “n is the total …”, n is missing.

⑥        p.4, l.137:  “Where mf will be identified an abnormal frequency point if satisfy the above condition. ” should be modify.

⑦        p.4, l.157:  needs should be need

⑧        p.2, l.161-164: repeated !

 

Comments for author File: Comments.pdf

Author Response

Thanks for your comments. We have answered and finished the revision carefully according to your suggestions and comments. It’s very helpful to improve the quality of the paper. We have replied item by item. If we fail to answer some questions and comments, please point them out and give us a chance to revise.

Author Response File: Author Response.docx

Reviewer 3 Report

General remark: it would be worth specifying the results for BDS-2/3.

Lines 87-91: this information should not be here.

Lines 107-109: prediction is based on the use of data from previous epochs and forward interpolation. Is data from previous epochs checked/controlled in any way? It must be excluded that these data themselves have no errors (gross/outliers) that will be used later for interpolation.

Line 119: should be “n is the total number of frequency data”?

Lines 125-127: should explain which values of n_orbit for which type of satellites are set. Where do these values come from? Some reference?

Lines 127-131: since real-time products can be different (e.g. IGS-RTS, ultra-rapid) it should be explained here for which products (with what time span) these formulas can be used since they depend on the parameter 'hour-boundary bias'. For example, the ultra-rapid products themselves have a different update period: ultra-rapid clocks from WUHAN are updated every hour, but most are every 6 hours, e.g. CODE, GFZ and IGS. Will the same algorithms work for both? Have such analyses been performed?

Line 155: change the name of the subsection.

Line 166: why this period was chosen?

Line 196-197: the sentence 'The abscissa is the satellite PRN and the ordinate is the average STD of the clock difference.' should be removed.

Line 227: ‘grey model’ - uniformly use the abbreviation 'GB' throughout a manuscript that has previously been explained.

Line 231: some reference?

Lines 242-245: for satellite C24 are also big boundary jumps: around 13 and 14 o'clock.

Line 310: ‘table’ with a capital letter.

Line 311: for the lines on the figure, I would use the same color as the mark or one black color for all marks.

Line 344: in Figure 11 it would be better to show the results for each satellite type (GEO, IGSO, and MEO).

Lines 363-364: what software and what settings/models were used for the calculations? Just show the table.

Author Response

Thanks for your comments and suggestions. We have answered and finished the revision carefully according to your suggestions and comments. It’s very helpful to improve the quality of the paper. We have replied item by item as follows. If we fail to answer some questions and comments, please point them out and give us a chance to revise. 

Author Response File: Author Response.docx

Reviewer 4 Report

 

The topic is interesting especially for real-time precise positioning application based on RT-clock.

There are some comments as follows:

1. Figure 3 should be replaced with a clearer picture.

2.L184,  In the process of precision clock estimation, the orbit, station coordinates, etc. are generally fixed, and the troposphere may also be fixed. However, "...tropospheric ambiguity..." there are misunderstandings and please explain.

3.please add the signal combination used in precise clock estimation in section 3.1.

4. One reference about effects on clock of BDS GEO and non-GEO orbit can be cited in L214.

5. Please add some explaination into manuscript about how to obtain the percentage in Figure 9.

6. Line 360, The PPP model and parameters 'are displaied'  in Table 2.

7. Figure 13, Consider the use of horizontal error and vertical error expression. There are positive and negative values in NEU, and the expression rate of improvement is not obvious.

8.Please modify the format of references according to the MDPI requirements.

like "Weng, D.; Ji, S.; Lu, Y.; Chen,W.; Li, Z. Improving DGNSS Performance through the Use of Network RTK Corrections. Remote Sens. 2021, 13, 1621. [CrossRef]"

Author Response

Thanks for your comments. We have answered and finished the revision carefully according to your suggestions and comments. It’s very helpful to improve the quality of the paper. We have replied item by item as follows. If we fail to answer some questions and comments, please point them out and give us a chance to revise. Besides, we use italics font to distinguish our corrections from the original text.

Author Response File: Author Response.docx