Optical FBG Sensor-Based System for Low-Flying UAV Detection and Localization

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors propose the detector for low-flying UAV based on the optical fiber Bragg gratings. They demonstrate the efficient detection of the drone’s localization for the heights from 30 to 120 cm. The results obtained demonstrate high efficiency, and the topic is highly relevant today and can attract the interest of community. However, some details should be clarified and specified prior to publication, namely:

1. I have a few questions related to Eq. (1). What is the unit of coefficient A? Please discuss under which conditions the second term becomes negligible with respect to the first one, and vice versa.
2. What is the value of the initial strain wavelength? What are the typical changes of temperature?
3. Please add more details on the measurements of wavelength shift. What is its resolution?
4. Table 1 providing the comparison between different systems, results and methods strengthens the manuscript. Can the authors briefly discuss the perspectives of using more advanced microstructured fibers, e.g. [Laser & Photonics Reviews, 10(6), 922-961 (2016); Applied Physics Reviews, 10(1), 011401 (2023)]? In other words, can the sensitivity be further improved?

Author Response

Thank you for your suggestions!

Please see attachment or responses provided below.

Comment 1: I have a few questions related to Eq. (1). What is the unit of coefficient A? Please discuss under which conditions the second term becomes negligible with respect to the first one, and vice versa.

 

Response 1: “We appreciate this suggestion! The unit of coefficient A is με^-1, which is an inverse microstrain. It is indicated in the publication in line 145. Microstrains are just the ratio between the difference in length of the object under stress ∆ L, and the initial length of the object under stress. In this context, the first term is never negligible, as high differences in temperature will lead to a high wavelength shift in the sensor. The second term becomes negligible when the change in temperature is less than a tenth of a degree, as the first term will be orders of magnitude larger than the second term. During the experiments, the temperature was stable within +/- 0.1 degrees. We have expanded the text to include this suggestion (lines 194-196).”

 

Comment 2: What is the value of the initial strain wavelength? What are the typical changes of temperature?

 

Response 2:

“Thank you for this comment. Initial wavelengths are in the C band from 1520 to 1565 nm. (FBG1 = 1520 nm; FBG2 = 1525 nm; FBG3 = 1530 nm; FBG4 = 1535 nm; FBG5 = 1540 nm; FBG6 = 1545 nm; FBG7 = 1550 nm; FBG8 = 1555 nm; FBG9 = 1560 nm; FBG10 = 1565 nm) The experiments were conducted indoors, so the temperature was stable at 22 degrees +/- 0.1 degrees. We've added more information to the text (line 187).”

 

Comment 3: Please add more details on the measurements of wavelength shift. What is its resolution?

 

Response 3:

“We are grateful for this comment! The device specification lists the resolution as 1 pm, however, the saved data files show a precision of 1 fm. Notable parameters of the spectrometer are a scanning frequency of 5 kHz and an 80 nm wavelength range in the C band, 1510 – 1590 nm. We've added this information to the manuscript (lines 198 – 199 and 204).”

https://www.sylex.sk/product/scn-80-s-line-scan-800/

 

 

Comment 4: Table 1 providing the comparison between different systems, results and methods strengthens the manuscript. Can the authors briefly discuss the perspectives of using more advanced microstructured fibers, e.g. [Laser & Photonics Reviews, 10(6), 922-961 (2016); Applied Physics Reviews, 10(1), 011401 (2023)]? In other words, can the sensitivity be further improved?

 

Response 4:

“Thank you for the suggestion of considering advanced nanostructured fibers, as we have not explored this topic. Advanced nanostructures show great promise in increasing the light coupling efficiency and therefore the sensitivity of the FBG sensor network. While the possible benefits are clear, they do not yet outweigh the drawbacks, such as an increased manufacturing cost and complexity, and the general maturity of this technology, as one of the key advantages of the proposed system is the reuse of existing fiber technology. We have added discussion of such a perspective in the introduction of the manuscript (lines 104-109).”

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript proposes a CUAV concept using an FBG sensor array to infer UAV downwash for height estimation and coarse localization. The idea is interesting, but the framing, benchmarking, and evidence do not yet meet full rigor and state-of-the-art coverage.

1. Frame detection as fusion, not as “optical *or* acoustic … and radars.” In the Introduction, the sentence “Popular drone recognition methods use optical or acoustic sensors and radars” misleads by implying exclusivity. Reframe around multi-sensor fusion and cite representative fusion CUAV work (EO↔radar, RF↔acoustic).
   Examples:  LNCS fusion chapter (EO + low-RCS radars): https://link.springer.com/chapter/10.1007/978-3-031-13324-4_36 ;
   RF+acoustic fusion: https://www.mdpi.com/1424-8220/24/8/2427

2. Justify or generalize the “photonic radar” claim. The Conclusions state the prototype “has a high potential … as an augmentation to existing photonic radar systems.” Either justify why augmentation is specific to photonic radar, or generalize to radar/RF sensing (including passive or mmWave) to match the manuscript’s broader narrative.

3. Add a focused comparison to fiber-optic DAS for CUAV. A short subsection should position FBG arrays vs Distributed Acoustic Sensing (DAS): sensitivity to airborne acoustics, spatial sampling vs point FBGs, interrogator complexity/cost, coverage, and EM immunity. Useful examples: FOAS-enhanced DAS for drone detection: https://ieeexplore.ieee.org/document/9896929 ; recent DAS-based low-altitude UAV surveillance: https://ieeexplore.ieee.org/document/10667006 ; DAS + optical microphones: https://ieeexplore.ieee.org/document/9748183 .

4. Refresh the Related Work to 2025 CUAV SoTA. Integrate a recent comprehensive review (2020–2025) to align terminology (C-UAS/CUAV), layered architectures, and fusion trends:  https://www.sciencedirect.com/science/article/pii/S1874548225000058
   Use one form (CUAV or C-UAS) throughout; ensure consistent typography for FBG, DAS, mmWave, RCS.

5. Ground truth and protocol rigor. Height/localization validation relies on sections “visually noted” during flights; this is not an adequate GT for a sensing paper. Use RTK-GNSS, motion capture, or instrumented markers to provide positional/temporal GT, and separate training/calibration from testing. If it's impossible to get a better test campaign, state the limit as a clear LIMITATION. 

6. You report 150 readings/s per FBG with a 301-point Savitzky–Golay filter; discuss anti-aliasing, effective bandwidth, and justify why 150 Hz suffices for rotor-induced dynamics.

7. Report robustness to wind and background motion/noise, more flight profiles (non-perpendicular, variable speed), and different UAV classes (if not feasible report as LIMITATION or FUTURE WORK)

Author Response

Thank you for your suggestions!

Please see the attachment or comments below.

Comment 1: Frame detection as fusion, not as “optical *or* acoustic … and radars.” In the Introduction, the sentence “Popular drone recognition methods use optical or acoustic sensors and radars” misleads by implying exclusivity. Reframe around multi-sensor fusion and cite representative fusion CUAV work (EO↔radar, RF↔acoustic).

Examples:  LNCS fusion chapter (EO + low-RCS radars): https://link.springer.com/chapter/10.1007/978-3-031-13324-4_36;

RF+acoustic fusion: https://www.mdpi.com/1424-8220/24/8/2427

 

Response 1:

“The Reviewer is correct in pointing out that the sentence in question misleads by implying exclusivity. We have rephrased it as “Popular drone recognition methods use optical and acoustic sensors, and radars, as well as incorporate sensor fusion techniques to mitigate the shortcomings of any individual detection method.” with relevant references (lines 24-25). We have also added a detailed discussion of multi-sensor fusion later in the introduction of the manuscript, citing contemporary work (lines 81-93).”

 

Comment 2: Justify or generalize the “photonic radar” claim. The Conclusions state the prototype “has a high potential … as an augmentation to existing photonic radar systems.” Either justify why augmentation is specific to photonic radar, or generalize to radar/RF sensing (including passive or mmWave) to match the manuscript’s broader narrative.

 

Response 2:

“We have generalized this claim to include radar and other UAV detection systems. It is in our future plans to integrate the proposed system with a frequency modulated continuous wave (FMCW) radar system as we have indicated in the revised manuscript (lines 334-338).”

 

Comment 3: Add a focused comparison to fiber-optic DAS for CUAV. A short subsection should position FBG arrays vs Distributed Acoustic Sensing (DAS): sensitivity to airborne acoustics, spatial sampling vs point FBGs, interrogator complexity/cost, coverage, and EM immunity. Useful examples: FOAS-enhanced DAS for drone detection: https://ieeexplore.ieee.org/document/9896929 ; recent DAS-based low-altitude UAV surveillance: https://ieeexplore.ieee.org/document/10667006 ; DAS + optical microphones: https://ieeexplore.ieee.org/document/9748183 .

 

Response 3:

“We have added a general overview of the DAS in the manuscript, as well as a comparison between DAS and FBG-based approaches to C-UAS on the points of system design and deployment, cost and capabilities (lines 40-48).”

 

Comment 4: Refresh the Related Work to 2025 CUAV SoTA. Integrate a recent comprehensive review (2020–2025) to align terminology (C-UAS/CUAV), layered architectures, and fusion trends:  https://www.sciencedirect.com/science/article/pii/S1874548225000058

Use one form (CUAV or C-UAS) throughout; ensure consistent typography for FBG, DAS, mmWave, RCS.

 

Response 4:

“We have added an additional analysis of contemporary work on sensor fusion techniques (lines 81-93, 60-64) as well as included state-of-the-art related work to the sections of the introduction, discussing acoustic, optical and radar UAV detection techniques. We have aligned the terminology in the manuscript (C-UAS).”

 

Comment 5: Ground truth and protocol rigor. Height/localization validation relies on sections “visually noted” during flights; this is not an adequate GT for a sensing paper. Use RTK-GNSS, motion capture, or instrumented markers to provide positional/temporal GT, and separate training/calibration from testing. If it's impossible to get a better test campaign, state the limit as a clear LIMITATION.

 

Response 5:

 

“We appreciate this comment! The height of the UAV was checked using the internal height measurement of the DJI AVATA UAV, and the localization was done by using the downward-facing camera of the same UAV. In our opinion, this relatively small-scale drone localization architecture would not benefit from an RTK-GNSS motion capture system, but we will indicate this as a limitation. The explanation for localization is added to the text (lines 213-215), and it was indicated as perspective in future works (line 335).”

 

Comment 6: You report 150 readings/s per FBG with a 301-point Savitzky–Golay filter; discuss anti-aliasing, effective bandwidth, and justify why 150 Hz suffices for rotor-induced dynamics.

 

Response 6:

 

“We thank the reviewer for noticing this. The Savitzky-Golay filter is used for anti-aliasing purposes. In the text, it is mentioned that the filter smooths out sudden and abrupt changes in reflected wavelength for a more continuous profile, but the reviewer is correct that anti-aliasing is a better term to be used here. We believe that this type of combination suffices for our purposes, as it provides what we deem as high fidelity of distinction. Savitzky-Golay filters can be tuned to use higher or lower degree polynomials, but lower degree polynomials would smooth the curves too much and make flight profiles blend together, while higher degree polynomials would leave too much noise for accurate distinction. In light of this comment, we reran our post-processing scripts with a more comprehensive cosine similarity method and found that a smaller filter frame of 101 points yielded an increase in fidelity. We've expanded the text in accordance with this (lines 225-227).”

Comment 7: Report robustness to wind and background motion/noise, more flight profiles (non-perpendicular, variable speed), and different UAV classes (if not feasible report as LIMITATION or FUTURE WORK)

 

Response 7:

“Thank you for your observation. Testing of the system in various weather conditions such as rain, snow, and wind, as well as determining the detection and localization capabilities with various UAVs in different flight modes are in our future plans. We have included a paragraph in the conclusions of the manuscript detailing our future work (lines 334-338).”

 

 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript presents a system for UAV detection and localization using a Fiber Bragg Grating sensor array to measure aerodynamic downwash. The concept is pertinent, addressing a known gap in conventional radar and optical systems for low-altitude targets, particularly in environments with high electromagnetic clutter. While, the claims of novelty are not sufficiently supported by the presented work, the experimental validation is confined to idealized laboratory conditions, and the data analysis lacks statistical robustness. Significant revisions are required to elevate the manuscript from a preliminary report to a comprehensive scientific paper.

1. The manuscript's claim to novelty is overstated; electromagnetic immunity is an inherent FBG property, not a novel discovery of this work.

2.Claims of robustness to environmental factors are unsubstantiated, as experiments were conducted in a controlled indoor setting, explicitly minimizing external airflow.

3.The selection of the cosine similarity algorithm is not justified against other potential, and potentially more robust, pattern recognition methods.

4.The physical origins of the observed downwash pressure patterns, particularly the complex multi-peak structure shown in Figure 3, are not adequately explained.

5.The system's performance is validated on only a single UAV model, which severely limits the generalizability of the findings to drones of different sizes and rotor configurations.

6.The procedure for creating the baseline flight matrix from "average peak wavelength shifts" is vaguely defined and lacks the detail required for reproducibility.

7.There is a significant discrepancy between the conclusive accuracy figures presented in the abstract and the more variable, and sometimes lower, results detailed in the text.

8.The fundamental relationship between the measured FBG strain and the airflow velocity is established via an un-detailed calibration, which is a critical step that requires more thorough explanation.

Author Response

Thank you for your suggestions!

Please see the attachment or comments below.

Comment 1: The manuscript's claim to novelty is overstated; electromagnetic immunity is an inherent FBG property, not a novel discovery of this work.

 

Response 1:

 

“Thank you for this comment! We agree with the statement that electromagnetic immunity is not a novelty discovered by us. Our aim was to highlight that a detection method with this specific property is uncommon and a great advantage. We have made this clearer in the text (line 15).”

Comment 2: Claims of robustness to environmental factors are unsubstantiated, as experiments were conducted in a controlled indoor setting, explicitly minimizing external airflow.

 

Response 2:

“The Reviewer is correct in pointing out that the experiments were conducted in a controlled indoor environment. For this reason, testing the system in various weather conditions such as rain, snow, and wind is in our future plans. We have included a paragraph in the conclusions of the manuscript detailing our future work (lines 334-338).”

 

Comment 3: The selection of the cosine similarity algorithm is not justified against other potential, and potentially more robust, pattern recognition methods.

 

Response 3:

 

“We appreciate this suggestion! There is a wide array of potential pattern recognition models, and it would not be feasible to compare every single one. In our work, we used the cosine similarity principle because it is a very simple method of vector comparison that does not require a large initial data set. From our research, we concluded that this pattern recognition model will be the best match for our experiment. We have also added this explanation in text (line 256).”

Comment 4: The physical origins of the observed downwash pressure patterns, particularly the complex multi-peak structure shown in Figure 3, are not adequately explained.

 

Response 4:

“Thank you for noticing this! The multi-peak downwash pattern is visible because there are 2 lines of propellers with their induced vortices combining near the hull of the UAV, while separating further away from the UAV, where these vortices lose intensity. This phenomenon is also visible in other publications, which are cited as 41 and 42 in our text. We've added a better explanation in our manuscript (lines 235-236).”

 

Comment 5: The system's performance is validated on only a single UAV model, which severely limits the generalizability of the findings to drones of different sizes and rotor configurations.

 

Response 5:

 

“We appreciate this comment! Unfortunately, we only had one specific model of drone available. Our aim was to show this system as proof of concept for future works that could potentially lead to an integrated drone localization and classification system using this method. It is in our future plans to determine the detection and localization capabilities with various drones, as well as to demonstrate the proposed system’s integration with a frequency modulated continuous wave (FMCW) radar system. We've noted this in the text (lines 334-337).”

 

Comment 6: The procedure for creating the baseline flight matrix from "average peak wavelength shifts" is vaguely defined and lacks the detail required for reproducibility.

 

Response 6:

“Thank you very much for this observation! In light of this comment, we reran our post-processing scripts with a more comprehensive cosine similarity method by comparing a larger dataset from the already available flights. We hope that a thorough explanation, which is now available in the manuscript, will be clear enough (lines 256-270).”

 

Comment 7: There is a significant discrepancy between the conclusive accuracy figures presented in the abstract and the more variable, and sometimes lower, results detailed in the text.

 

Response 7:

 

“Thank you for this remark! As mentioned in comment 6, due to concerns about accuracy, a more conclusive model of measurement was tried with the same data set and yielded higher fidelity results. We hope that these results are now up to standard. The results listed in the abstract are just the average fidelity across all height or localization measurements and were not meant to mislead. We've added this clarification to the abstract (line 14).”

 

Comment 8: The fundamental relationship between the measured FBG strain and the airflow velocity is established via an un-detailed calibration, which is a critical step that requires more thorough explanation.

 

Response 8:

 

“Thank you for this comment! The relationship of wavelength shift in the sensor to airflow speed is determined by using an airflow module, which was measured with an anemometer to be able to induce airflow of up to 8 m/s. It was measured at what distance from this module, whole integer air flow speeds from 1 to 8 m/s were observed. Placing the module at these distances from the sensors, their wavelength shift readings were noted. The overpassing UAV-induced wavelength shift was then interpolated onto these values. This has been added to the text (lines 220-222).”

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have referred to all my comments and improved the manuscript. I suggest it for the publication in Applied Sciences journal.

Reviewer 2 Report

Comments and Suggestions for Authors

Thank the authors for having accepted and implemented the proposed revisions.
In this reviewer’s opinion, the paper now meets the standards required for publication.

The implementation of the Optical FBG Sensor-based System is promising, and I hope it will have positive implications for C-UAV systems such as those developed within project such as SPS (AI4CUAV.org) among othres, which remain under active development (tested here: https://link.springer.com/chapter/10.1007/978-3-031-86162-8_20)

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript  can be published