A Survey on the Use of Deep Learning Techniques for UAV Jamming and Deception

Round 1

Reviewer 1 Report (Previous Reviewer 1)

Well Written

Author Response

Thank you very much for your review.

Reviewer 2 Report (Previous Reviewer 4)

Thank you very much for your efforts. The quality of revised draft is highly improved. The authors have addressed my concerns well. I have two minor comments for this revised draft.

1. There are some abbreviations which are not defined fully. So, I suggest to add a Table in Appendix. 

2. Still the quality of some figures is not good. Again I suggest to provide all figures in high resolutions. For instance, compare the quality of Figure 11 (a) and 11 (b). You can find the difference easily.  

Author Response

Thank you very much for your review. 

Added explanatory notes for undefined abbreviations.

Poor quality figures were replaced.

Reviewer 3 Report (Previous Reviewer 2)

This is a very good piece.

Author Response

Thank you very much for your review.

Reviewer 4 Report (New Reviewer)

The article addresses the topic of jamming and deception of UAVs in a very broad manner. In the introduction, it presents the possibilities of using UAVs for various activities, including those that may raise various questions in both the civilian and military fields. Also in the introduction, the Authors cite publications on relevant research and reviews on various UAV-related applications, divided into the following categories: UAV Related Application, Wireless Networks, UAV Jamming, UAV Protection, UAV Spoofing, Machine Learning, and Deep Learning. This part of the article is quite synthetic and could even be expanded a bit. Section 2 deals with the topic of signals and communication protocols used in UAVs. Basic information is presented in general, but due to the number of solutions presented, despite the rather concise description, this part of the work takes up quite a lot of space in the whole article. The same is true of Chapter 3, which discusses UAV interference issues. In Chapter 4, the Authors move on to the issue of machine learning, where they summarize the most important information in this area. Chapter 5 is a continuation relating to deep learning. Both Chapters 4 and 5 contain tabular summaries of what the Authors consider to be the most important work in the areas covered. Chapter 5 also concludes the descriptive, overview section of the paper. In Section 6, the Authors go on to describe a signal modulation classifier. In my opinion, this is the most interesting part of the work, especially for those previously familiar with the whole of the issues presented in the paper. The results are presented in a very accessible way, but it would be advisable to expand the discussion a bit.

The article is characterized by a very large number of pages. I do not regard this as its disadvantage, however, it might be worth considering dividing its content. Especially since the first 20-plus pages are typically of an overview nature. Dividing it into sections could even allow to introduce readers even more thoroughly to the issues covered, if only the Authors decided to present the information in Chapters 2 and 3 in more detail. The same could be done with Chapters 4 and 5 with regard to UAVs themselves. I would see as a separate article the content of Chapter 6, preceded, of course, by an appropriate introduction, appropriate both in content and length. I encourage the Authors to prepare a book publication complementarily treating the issues raised on the basis of the content of the article in its present form, of course, after further detailing. 

It is rare that an article at the beginning of the review process does not raise objections to the editorial side. I only noticed:

Since most UAVs have some autopilot feature (such as the "go home" feature mentioned above), GPS jamming is one way to neutralize drones without destroying the.
->
Since most UAVs have some autopilot feature (such as the "go home" feature mentioned above), GPS jamming is one way to neutralize drones without destroying them.

RNN IS used when the data have a sequential character
->
RNN is used when the data have a sequential character

Author Response

We very much appreciate your comments.
The errors you mentioned have been corrected.
After discussion among the authors, the structure of the paper as a whole remains unchanged, although we agree with most of your comments. The aim of our work, as mentioned in the introduction, is primarily to provide insight into the issues under study and to link the mentioned areas, as we have not been able to find a similar publication so far. At the same time, we want to encourage further research on the described issue. Initially, Section 6 was intended as a separate article. However, a review article alone, without an experimental section, may not be considered by all as worthy of publication. At the same time, this section serves as an example of the application of the discussed problem, hence we decided to add the section to this article ( in the first submission this section was omitted). The authors are planning further publications in this area, so we will certainly take your comments that we have not used in correcting this article into account in future publications. Once again, thank you very much for your comments.

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.

Round 1

Reviewer 1 Report

This paper is well written. I agree with the words "However, no detailed research has been done so far on the use of these techniques for jamming and deception of UAVs". However, I still have some major comments:

1. It should be a table or a figure (map) to explain the relationship of reviewed literature.
2. In fig.7, it should use vivid figures to depict different cyberattacks.
3. 4.2.3. Semi-supervised Learning – SSL and 5.1. Deep Learning Models. They are too simple. This content needs to be strengthened. 
4.  Some important literature is missing, listed as follows: 

   S. Fang, G. Chen and Y. Li, “Joint optimization for secure intelligent reflecting surface assisted UAV networks”, IEEE Wirless. Commun. Lett., vol. 10, no. 2, pp. 276-280, Feb. 2021.

   A. Saeed, G. Chen, Y. Rahulamathavan, G. Zhang, B. AsSadhan and S. Lambotharan, “Trusted UAV Network Coverage Using Blockchain, Machine Learning, and Auction Mechanisms”, in IEEE ACCESS, vol. 8, pp. 118219-118234, 2020.

   J. Tang, G. Chen and J. P. Coon, ''Secrecy Performance Analysis of Wireless Communications in the Presence of UAV Jammer and Randomly Located UAV Eavesdroppers", in IEEE Trans. Inf. Forensics and Security, vol. 14, no. 11, pp. 3026-3041, Nov. 2019.

   Please properly comment on them in this paper. Moreover, the delayed information induced by the mobility of UAV should be considered. Please comment on the following papers 1) Full-Duplex Relay with Delayed CSI Elevates the SDoF of the MIMO X Channel 2) On Secure Degrees of Freedom of the MIMO Interference Channel with Local Output Feedback

Author Response

1. Added Table 1 with an overview of recent publications and commentary. Also added section 6. Discussion that compares recent review articles with our publications, their limitations, etc.
2. Fig. 7 essentially serves as a table of contents for the following paragraphs. Not edited in any way.
3. Both sections (4.2.3 Semi-supervised Learning and 5.1 Deep Learning) have been expanded. In particular, a commentary on Table 4 (Applications of DL algorithms in signal processing and UAV-related applications) has been added at the end of Section 5.1.
4. From the papers you suggested, some of them were used as references. However, I have not discussed them in detail, as the article does not aim to cover all topics in great detail. Nevertheless, thank you very much for the suggested literature, it partly overlaps with future planned research, so definitely some publications will be used later.

Reviewer 2 Report

Unmanned aerial vehicles (UAVs) can be used for a variety of illegal activities (e.g., industrial espionage, smuggling, terrorism).

Given their growing popularity and availability, and advances in communications technology, more sophisticated ways to disable these vehicles must be sought. As a rule, various variants of jamming are used to disable drones, however, more advanced techniques such as deception and UAV takeover are considerably difficult to implement and there is a large research gap in this area. Currently, machine and deep learning techniques are very popular and are also used in various drone-related applications. However, no detailed research has been done so far on the use of these techniques for jamming and deception of UAVs.

The authors proposed a  paper on exploring the current techniques in the area of jamming and deception.

They proposed a survey on the use of machine or deep learning specifically in UAV related applications is also conducted.

The outcome of their paper provides insight into the issues described and encourages more detailed research in this area.

The manuscript is very interesting and written with enthusiasm. There is a lot of information useful and encouraging feature studies.

With pure academic spirit I propose the following improvements:

1. L’abstract può essere migliorato riassumendo meglio le varie sezioni. L’obiettivo è quasi alla fine
2. As aim you say “This article is aimed to fill a gap in current research and should serve as a brief insight into the mentioned issues and provide the basic knowledge necessary for orientation in these above-mentioned areas.” I suggest beeing more concise
3. Insert a discussion where you in particular can discuss the limitations of the cited studies and of your study and perhaps the perspectives
4. The last paragraph is “summary and conclusions”. Change it into “conclusions”, you already have a summary… the abstract.

Author Response

Thank you very much for your review report. Regarding the proposed improvements:

1. Added Table 1 with an overview of recent publications and commentary. Also added section 6. Discussion that compares recent review articles with our publications, their limitations, etc.
2. Fig. 7 essentially serves as a table of contents for the following paragraphs. Not edited in any way.
3. Both sections (4.2.3 Semi-supervised Learning and 5.1 Deep Learning) have been expanded. In particular, a commentary on Table 4 (Applications of DL algorithms in signal processing and UAV-related applications) has been added at the end of Section 5.1.
4. From the papers you suggested, some of them were used as references. However, I have not discussed them in detail, as the article does not aim to cover all topics in great detail. Nevertheless, thank you very much for the suggested literature, it partly overlaps with future planned research, so definitely some publications will be used later.

Reviewer 3 Report

The article is a large literature review. The works of other authors are examined in detail. No scientific research has been done.  However, a review of the literature, however comprehensive it may be, is not a scientific work. The article uses paintings from other research papers. Images from which source the image was copied are not listed. The impression is that the painting was created by the authors themselves.

Author Response

Thank you for your rewiev report.

Regarding the pictures, all pictures were drawn by the author. However, it is true that several of them are based on pictures from other publications. Appropriate citations have been added. In addition, other significant changes have been made as recommended by other reviewers.

Reviewer 4 Report

The writing and presentation quality of this article is good. A very precise and clear idea about this work is presented in both Abstract and conclusions. In this study, the authors focus on exploring the current techniques in the area of UAV jamming and deception. They have conducted a survey on the use of machine learning and deep learning specifically in UAV-related applications. It is a timely survey as there is a research gap in this specific research topic. While reviewing this interesting article, I have found a few issues which must be addressed to improve the quality of this work.

1. It is suggested to provide a Table in Section 1 and compare your work/contributions with relevant reported Surveys or Reviews.
2. In Keywords, what is UAS? Line 76-77, what is AM, FM, PM, ASK, FSK, PSK? Authors should carefully check each abbreviation must be defined fully in the first place of appearance to enhance understanding and better readability.
3. Section 1 Introduction: Authors should add more significant discussion in this section. There is a complete lack of discussion and reference literature in this section.
4. The quality of Figures 2 (b), and 9 (b) is not good. Some figures do not look attractive. It is highly suggested to improve figures quality and must provide in high resolution for better readability.
5. At some places authors have placed Figures without proper analysis and discussion. Each figure must be discussed thoroughly and sufficient information must be given to increase understanding for readers.
6. Authors have provided insufficient information for THSS.
7. Sub-section 2.1.5, what is QAM, QPSK? As the authors did not define several abbreviations so it is suggested to add Nomenclature.
8. Line 161, Error! Reference source not found. Make relevant corrections.
9. The discussion for SSL in Sub-section 4.2.3 is not insightful.
10. The summarization of Table 2 and Table 3 is not insightful. The authors are recommended to present detailed information from related papers rather than one-sentence summary. 
11. Authors have designed all these figures or some figures have been taken from other resources? In such conditions, authors must acknowledge those studies and must cite relevant studies properly.
12. Authors should add more significant discussion about DL Models from relevant research studies as provided discussion seems insufficient.
13. In the final section, the authors should shed some light on possible research directions for future works which might be helpful for the relevant research fraternity as there is some research gap in these areas.
14. Reference section is updated with very recent research contributions. I have found some interesting studies.

In my opinion, this article is well-organized. The English writing and presentation quality is good. As it is a timely survey and focused on a hot research topic so it will be interesting for the relevant research community

Author Response

Děkuji moc za podrobnou revizi. Ohledně problémů, které jste zmínil:

1. Tabulka 1 přidána i s komentářem, zároveň přidána sekce 6. Discussion porovnávající citované publikace s naší prací.
2. UAS vymazáno. Tato zkratka znamená Unmanned Aerial System. AM, FM, PM jsou zkratky pro amplitodovou, frekvenční a fázovou modulaci. ASK, FSK, PSK jsou zkratky pro amplitudové, frekvenční a fázové klíčování. Tyto zkratky nejsou v článků rozebrány jelikož je z části zaměřen do signálové domény a modulací a je předpoklad, že tyto zkratky jsou obecně známy v této oblasti.
3. Doplněno viz. bod 1.
4. Zmíněné obrázky upraveny a změněny.
5. Mělo by být vyřešno.
6. Doplněny informace k THSS.
7. Stejný případ jako v bodě 2. QAM a QPSK jsou známé a často používané modulaci. 
QAM = kvadraturní amplitudová modulace
QPSK = kvadraturní fázové klíčování
8. Vymazáno.
9. Doplněny informace k SSL.
10. Popis jednotlivých publikacích v příslušných tabulkách rozveden.
11. Autor sice kreslil obrázky sám, ale některé jsou založeny na obrázkách z jiných publikací. U těchto obrázků přidány dané citace.
12. Rozšířen komentář v této sekci, především u tabulky s přehledem literatury v této sekci.
13. Přidám stručný souhrn plánovaného výzkumu.

Round 2

Reviewer 1 Report

Good Revision, No Comments