Chaotic Coding for Interference Suppression of Digital Ionosonde

Round 1

Reviewer 1 Report

This paper presents a method for suppressing interference using chaotic coding by a set of Bernoulli map sequences. The authors calculate PSL, ISL, NS, ACF of four simulated codes and show that Bernoulli map codes has better performance. The experimental testing showed that using Bernoulli map coding could remove interference and improve SNR.

However, there are many confusing statements and misprints in the manuscript. The simulations are not enough to prove that Bernoulli map codes has better performance than 40-bit bark-like code. And the simulation didn’t show the interference occurred near 70km and at 500km virtual height. Is this interference common or is this an exception? How does it come about? And where is Figure 7? For various reasons, the current manuscript is not recommended for publication.

Detailed comments:

Line 87. Is it should be n rather than N?

Figure 1. The max value in (d) is 2 but it is 1 in other subgraphs, why?

Line 104. It seems to be “(g) and (h)” but not “(g) and (f)”.

Line 106. How does this conclusion come about?

Line 126. Form Figure 1 and 2, It's difficult to reach that conclusion.

Line 181. Spelling mistakes: “occurred” but not “occured”.

Line 198. Where is Figure 7?

Line 212. There seems to be notes in Chinese. Maybe authors forget to delete. Please check it.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Notes:

1. The article presents a comparison only with Barker-like signals. However, various M-sequences, Legendre sequences have been widely used for a long time. M-sequences have a better balance of zeros and onesthan the modified Barker codes. As a result, it is desirable to explain why Bernoulli map sequences are better than M-sequences.

2. Question to Table 2. Experience shows that when you win some, you can simultaneously lose some. The authors have not yet identified the challenging aspects that appear when using the proposed codes. Are these codes the best in any aspect?

Notes on the English language.

Some revisions are required. The main errors are as follows:
1. Mistypes (e.g., lines 27, 40, Table 3, 161-162, 164);

2. Wrong verb forms (e.g., lines 26, 53, 83, 120);

3. Punctuation marks: missing commas, excessive commas and full stops (e.g., lines 85, 150, 168, 197, 216-217);

4. Revise Item 5.5 of MDPI Style Guide (e.g., lines 44, 55, 79);

5. Usage of prepositions (e.g., lines 49, 164);

6. Meaning of some sentences is unclear (e.g., lines 146-149, 226-228);

7. "Where" should be placed after equations (e.g., line 77).

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The article is devoted to the analysis of application the various methods for encoding ionosonde pulses. It is approved that the using 40-bit Bernoulli map coding gives the best results. Nevertheless, there are several questions.

1.      The authors do not explain why is 40-bit code chosen, and, for example, not 35 or 45.

2.      A few questions about the experimental testing and its results.

- The experiment was carried out using a vertical ionosonde in Yinchuan, Ningxia Hui Autonomous Region, China. It is specified that the experiment was limited to comparing only two 40-bit codes in an apparently quiet ionosphere at the same time of day. While, the authors assumed the same noisy environment.

- The authors do not provide the original ionograms without coding.

- The critical frequencies are not observed in the ionograms shown in Fig. 8. The reasons of this situation are not specified. Meanwhile, the absence of critical frequencies on ionograms may be a consequence of the coding method used.

- There is no indication on the ionograms which time is used LT or UT.

- There were no original (without coding) ionograms and ionograms using pulse coding under conditions of weak and strong disturbances in the ionosphere (F-spread, geomagnetic storms) or strong interference.

3.      The authors simultaneously use in article the terms “Bark-like” and “Barker-like” (206, 211 and 19, 32…), “Baker” (34); there are some errors with dimensions (ms instead of μs, see lines 158, 159).

4.      The given links [1-3] do not open.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Review of the manuscript submitted for publication in Remote Sensing:

Title: Chaotic Coding for Interference Suppression of Digital Ionosonde

Authors: Sijia Han, Wei Guo *, Peng Liu, Te Wang, Caiyun Wang, Qingyu Fang,
Jian Yang, LingLing Li, Dapeng Liu, Jianping Huang

Special issue: Advanced Radar Signal Processing and Applications.

The paper under review discusses the use of the chaotic sequences as the waveforms in the Chaotic Coding Digital Ionosonde (CCDI). To deal with the interference and the resulting false signals occurred in the ionograms, the authors introduce a set of 40-bit Bernoulli chaos sequences that virtually suppress the coherent interference.

The chaotic coding employing the Bernoulli map provides the almost perfect ambiguity function and good statistic performance. In the paper, the comparison of the various chaotic waveforms was performed by simulation of the echo signals by adding noise with Rayleigh distribution. As a result, it turned out that the Bernoulli map codes have a better signal-to-noise ratio than Barker and complementary codes.

The manuscript provides a description and technical details of the CCDI ionosonde and examples of ionograms obtained using various chaotic coding waveforms. The comparison of the ionograms showed that the SNR can be improved up to more than 5dB, and the coherent interference can be almost entirely removed using the Bernoulli map codes.

Thus, the article convincingly substantiates and demonstrates the advantage of Bernoulli map codes for use in the CCDI.

 One serious note.

Equation (9) for the autocorrelation function with an interference effect is formally incorrect. The absolute value of the sum is generally not equal to the sum of the absolute values of the terms. You must write "less than or equal to" instead of an equals sign. Only if the second term tends to zero does the inequality turn into equality. Obviously, this is exactly what the authors mean, but the formalities must be obeyed. Please correct.

 Some typos

In line 41 “code” -> “codes”

In line 53 “partial” -> “partially”

In line 55 “statistical” -> “statistically”

In line 119 “signals is” -> “signals are”

In lines 159, 160 “ms” -> microsec

Finally, I believe the manuscript is worth publishing after minor revision considering the remarks above.

Some typos noticed are enumerated in the review

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The revised manuscript represents a significant improvement. In the experiment test, Bernoulli map coding show a good performance. The current manuscript is recommended for publication. But I still have two small questions.

1.      Is code length same as sequence length? In equation (5), the authors states that N represents the length of the sequence, but in table 2, it seems to be code length rather than sequence length used in calculation of NS.

2.      From Fig.1 and Fig.2, It is not clear that Bernoulli map codes has better performance than 40-bit bark-like code, especially in noise suppression, but why did it work in experiment?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

References [1-3] are in Chinese.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf