Improving Recognition of Road Users via Doppler Radar Data and Deep Learning Convolutional Networks
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis study proposes a new method for recognizing and distinguishing objects using radar signatures and a specialized convolutional neural network. This method was validated with real 24 GHz radar measurements, proving effective in distinguishing pedestrians, cyclists, and cars in an urban environment. In general, the topic is meaningful, but the writing and novelty need improvement. The reviewer provides the following detailed comments:
1:The abstract needs to be concise; please avoid repeating similar sentences multiple times.
2:The introduction and related work sections should summarize the novelties, advantages, and disadvantages of existing research and compare them with the proposed method to emphasize the innovations of this paper.
3: The motivation and underlying principles of the proposed method, which can enhance performance, should be explained in greater detail.
4: Is there a relationship between f(t) in Equation 1 and the signal in Equation 2? If so, please express it using f(t) in Equation 2.
5:Performance differences among various waveforms should be analyzed and explained. If possible, it is recommended to illustrate these differences through simulations using different waveforms.
The abstract needs to be concise; please avoid repeating similar sentences multiple times.
Author Response
Comments 1: The abstract needs to be concise; please avoid repeating similar sentences multiple times.
Response 1: Thank you for your valuable feedback regarding the abstract. We have revised it to ensure that it is more concise and free from the repetition of similar sentences. Your observation was indeed correct, and we appreciate your guidance in improving the clarity of the manuscript.
Comments 2: The introduction and related work sections should summarize the novelties, advantages, and disadvantages of existing research and compare them with the proposed method to emphasize the innovations of this paper.
Response 2: Thank you for your insightful feedback regarding the introduction and related work sections. In response to your comments, we have incorporated a thorough summary that highlights the novelties, advantages, and disadvantages of existing research in relation to our proposed method. This enhancement emphasizes the innovations presented in our paper. We appreciate your attention to this aspect, which has significantly improved the overall clarity and focus of our manuscript.
Comments 3: The motivation and underlying principles of the proposed method, which can enhance performance, should be explained in greater detail.
Response 3: Thank you for your thoughtful comment. We have taken your suggestion to heart and have enriched the manuscript with a more detailed explanation of the motivation and underlying principles of the proposed method. We appreciate your guidance in helping us enhance the clarity of our work.
Comments 4: Is there a relationship between f(t) in Equation 1 and the signal in Equation 2? If so, please express it using f(t) in Equation 2.
Response 4: There is no relationship between f(t) in Equation 1 and the signal in Equation 2. The confusion may arise from the overlap of the radar's operating frequency fc​ with the time variable t. Thank you for your insightful observation.
Comments 5: Performance differences among various waveforms should be analyzed and explained. If possible, it is recommended to illustrate these differences through simulations using different waveforms.
Response 5: Thank you for your insightful comment. As you pointed out, the uRAD radar emits an electromagnetic wave with a frequency that varies over time, following the available types of modulation. The most commonly employed modulations include Continuous Wave (CW), Sawtooth, Triangular, and Dual Rate. In this study, we have chosen to focus on sawtooth modulation for both simulations and measurements, as this type of signal was found to be the most advantageous from our perspective, based on the radar parameter table.
Your observation is particularly interesting, prompting us to consider our research from a broader perspective and will serve as a valuable direction for future studies. We appreciate your understanding and valuable feedback.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe manuscript addresses the issues associated with road users recognition using a Doppler radar and aims to improve the performance of existing systems by integrating deep learning convolutional networks. The manuscript is interesting and contains relevant simulation and experimental results, justifying its publication in Electronics.
Please find my observations below:
1. “The method, previously published by the authors, has been validated using real measurements in an urban environment with a frequency-modulated continuous wave radar operating at 24 GHz.” I found this sentence somehow confusing as it is not clear which part of the method is found in an article “previously published by the authors”, and which is the novelty of this work. In my opinion, if part of the method is found in a different publication, this should be better clarified in the introduction section, and it should be more clearly explained which is the novelty of this new manuscript. On the other hand, I consider that it is not very appropriate to refer to a previous work in the abstract.
2. Paragraph between lines 33-36 could be supported by at least one reference.
3. CNN has been defined more than once within the manuscript.
4. Line 83-92: It seems for me that it is not very clear from this paragraph what are the differences between this work and the previous manuscript [6]. In my opinion, these paragraphs are not able to clearly explain what has been made, and how is this work continuing the previous article. More details should be provided.
5. In lines 200-205 the authors explain how their work is different from the literature review. In my opinion, the authors should also explain how and why is their approach important and useful.
6. Lines 207-210 should be changed/rephrased.
7. Fig. 5 would be more clear if the authors would add the titles of the events in the block diagram.
8. Titles for blocks should be also inserted in Fig. 6. Additionally, quality of this fig. should be improved.
9. Lines 365-368: “Authors should discuss the results and how they can be interpreted from the perspective of previous studies and of the working hypotheses. The findings and their implications should be discussed in the broadest context possible. Future research directions may also be highlighted.” – this should be removed.
10. The research experiment was conducted on 10,000 images (6 classes of objects with 1,000 images each). It is not clear how 6 classes of objects with 1,000 images each result in 10,000 images.
11. Line 383 “The results of the simulations presented above are the best among many obtained outcomes.” It is not very clear what the authors want to say. Additional info should be provided.
12. Additional comments/discussions should be provided for Fig. 8. Lines 390-394 are not sufficient.
13. In Fig. 11 the authors provide an analysis comparing the method proposed in this work and the results presented in [24] and [25]. I think the authors should firstly provide some details concerning [24] and [25], detailing the methods as well as the testing conditions.
14. The authors should also comment on why do they consider that the proposed method is better. They could also try to explain which is the component that made this method better.
15. It would be useful if the authors could optimize the presentation style throughout the manuscript.
Comments on the Quality of English LanguageIt would be useful if the authors could optimize the presentation style throughout the manuscript.
Author Response
Comments 1: “The method, previously published by the authors, has been validated using real measurements in an urban environment with a frequency-modulated continuous wave radar operating at 24 GHz.” I found this sentence somehow confusing as it is not clear which part of the method is found in an article “previously published by the authors”, and which is the novelty of this work. In my opinion, if part of the method is found in a different publication, this should be better clarified in the introduction section, and it should be more clearly explained which is the novelty of this new manuscript. On the other hand, I consider that it is not very appropriate to refer to a previous work in the abstract.
Response 1:
Thank you very much for your valuable feedback regarding the clarity of our manuscript. We have conducted a thorough revision of the abstract, introduction, and related works, which automatically incorporates your insightful comments. In the updated version, we have clearly indicated which elements of the method are derived from our previous work and which represent the novelty of the current study. Once again, we appreciate your assistance in strengthening our manuscript.
Comments 2: Paragraph between lines 33-36 could be supported by at least one reference.
Response 2: Thank you for your suggestion. We have now added a relevant reference to support the paragraph between lines 33-36. Please let us know if any further clarification is needed
Comments 3: CNN has been defined more than once within the manuscript.
Response 3: Thank you for your insightful observation regarding the redundancy in defining CNN within the manuscript. We agree that this is a valid point that enhances the clarity of the article. We have eliminated the repetitive definitions to ensure a more concise and coherent presentation of the content. We appreciate your feedback, which has contributed to improving our work.
Comments 4: Line 83-92: It seems for me that it is not very clear from this paragraph what are the differences between this work and the previous manuscript [6]. In my opinion, these paragraphs are not able to clearly explain what has been made, and how is this work continuing the previous article. More details should be provided.
Response 4:
Thank you for your valuable comment. We have addressed your valid concern by removing the redundant content. Additionally, we have included the following key properties of the new approach to clarify the differences between this work and the previous manuscript [6]:
- Detection of a broader class of objects, such as two pedestrians walking simultaneously or the concurrent movement of a pedestrian and a cyclist in open-space conditions.
- Feature extraction and classification of images represented as spectrograms for multiple object classes through a specialized CNN.
- A specialized CNN capable of achieving very high object detection accuracy (> 90%).
- A CNN that is resistant to overfitting, which occurs when the CNN model fits the training data too well but loses the ability to generalize to new, unseen data.
- A solution based on numerical simulations using a simulated dataset in the form of spectrograms (for radars operating at frequencies of 24 GHz and 77 GHz).
- Validation of the numerical solution based on our own dataset from an available FMCW radar operating at a frequency of 24 GHz.
- The use of data augmentation techniques to overcome the challenges and time-consuming nature of processing data from a single FMCW radar measurement.
While we acknowledge the drawbacks of our solution, such as the requirement for a large input dataset and the time-consuming process of building the neural network structure to achieve the highest possible accuracy without utilizing transfer learning techniques, we believe these enhancements significantly strengthen our manuscript.
Comments 5: In lines 200-205 the authors explain how their work is different from the literature review. In my opinion, the authors should also explain how and why is their approach important and useful.
Response 5: Thank you for this valuable feedback. We agree that it is important to explain the significance and usefulness of our approach. In the revised version of the manuscript, we have added detailed explanations regarding why our method is relevant in the context of current radar technologies and what benefits it can bring to practical applications. We believe that this additional information will strengthen our work and help to better convey its value.
Comments 6: Lines 207-210 should be changed/rephrased.
Response 6: Thank you for your valuable feedback. We have made the necessary changes to the specified lines to enhance clarity and precision. We appreciate your input in strengthening our manuscript.
Comments 7: Fig. 5 would be more clear if the authors would add the titles of the events in the block diagram.
Response 7: Thank you for your observation. We have added the titles of the events in the block diagram of Fig. 5 as requested. Please let us know if any further adjustments are needed.
Comments 8: Titles for blocks should be also inserted in Fig. 6. Additionally, quality of this fig. should be improved.
Response 8: We appreciate your feedback. Titles for the blocks have been inserted in Fig. 6, and we have improved the quality of the figure. We hope this version meets your expectations.
Comments 9: Lines 365-368: “Authors should discuss the results and how they can be interpreted from the perspective of previous studies and of the working hypotheses. The findings and their implications should be discussed in the broadest context possible. Future research directions may also be highlighted.” – this should be removed.
Response 9: Thank you for your valuable feedback regarding lines 365-368. We have rephrased the content of these lines to better address the concerns raised. Your input has been instrumental in enhancing the clarity of our manuscript. Thank you once again for your insightful suggestion.
Comments 10: The research experiment was conducted on 10,000 images (6 classes of objects with 1,000 images each). It is not clear how 6 classes of objects with 1,000 images each result in 10,000 images.
Response 10: Thank you for pointing out this inconsistency. We appreciate your keen observation. To clarify, the research experiment was conducted on a total of 6,000 images, with each of the 6 classes containing 1,000 images. This change has already been implemented in the revised manuscript for better clarity. Thank you once again for your valuable feedback.
Comments 11: Line 383 “The results of the simulations presented above are the best among many obtained outcomes.” It is not very clear what the authors want to say. Additional info should be provided.
Response 11: Thank you for your valuable feedback regarding the clarity of our statement in line 383. We appreciate your comments and understand that the phrasing could be improved for better understanding.
We would like to clarify that the results referenced come from numerous time-consuming simulations conducted to fine-tune the neural network model and optimize its hyperparameters. These simulations were crucial in achieving the accuracy and performance levels presented in our findings.
We would like to inform you that in the revised version of the manuscript, we have made the necessary adjustments to ensure that this information is presented more clearly, emphasizing the effort invested in the simulation process and how it contributes to the significance of our results.
Once again, thank you for your insightful comments, which have undoubtedly strengthened our manuscript.
Comments 12: Additional comments/discussions should be provided for Fig. 8. Lines 390-394 are not sufficient.
Response 12: Thank you for your insightful feedback regarding Figure 8. We appreciate your observation that the current discussion in lines 390-394 may not sufficiently address the key points related to the figure.
In response, we have expanded our discussion to provide more detailed commentary on the results depicted in Figure 8. This includes a deeper analysis of the implications of the findings, as well as their relevance to our overall research objectives. We believe that these additions will enhance the clarity and depth of our manuscript.
Thank you once again for your valuable input, which has undoubtedly strengthened our work.
Comments 13: In Fig. 11 the authors provide an analysis comparing the method proposed in this work and the results presented in [24] and [25]. I think the authors should firstly provide some details concerning [24] and [25], detailing the methods as well as the testing conditions.
Response 13: Thank you for your comment. In response, we have provided additional information regarding the methods in [24] and [25], along with the testing conditions. Specifically, the techniques outlined in [24] and [25] were implemented in the Matlab environment. The first paper frames the multiuser automatic modulation classification (mAMC) of compound signals as a multi-label learning problem, which aims to identify the modulation type of each component signal in a compound signal [24]. The second paper introduces a semantic-based learning network (SLN) that simultaneously performs modulation classification and parameter regression for frequency-modulated continuous-wave (FMCW) signals [25]. Both methods were adapted to allow the recognition of 6 object categories. Figure 11 illustrates the performance of the three approaches in identifying pedestrians, cyclists, and vehicles. The experiment was carried out on a dataset of 6,000 images (1,000 images for each of the 6 object categories). The images were randomly split into three sets: training data (70% of the images) were used to determine the neural network’s weights, validation data (15%) were used to evaluate the trained network, and test data (15%) were employed to assess the network’s functionality after training.
Comments 14: The authors should also comment on why do they consider that the proposed method is better. They could also try to explain which is the component that made this method better.
Response 14: Thank you for your insightful comment. We appreciate the opportunity to clarify our rationale behind considering the proposed method as superior to existing approaches. Our belief stems from several key factors related to the design and implementation of our method. Firstly, we have conducted a thorough analysis of various neural network architectures before selecting the one that best aligns with the specific requirements of our task. The chosen structure not only accommodates the complexities of the data but also enhances the network's ability to generalize well to unseen instances. As noted in the literature, the selection of an appropriate network architecture is crucial; it can significantly reduce training time and improve final outcomes by ensuring that the model is well-suited to the nuances of the problem being addressed. Additionally, we have implemented a carefully crafted learning process tailored for our input data. This process involves the optimization of hyperparameters and the incorporation of techniques such as data augmentation, which collectively contribute to a more robust learning experience. By addressing these critical aspects, we believe that our method is able to achieve higher accuracy and reliability. In summary, we attribute the enhanced performance of our proposed method to the strategic selection of the neural network structure and the meticulous design of the learning process. We are confident that these elements play a pivotal role in delivering the superior results observed in our experiments. Thank you again for your valuable feedback, and we hope this explanation addresses your concerns.
Comments 15: It would be useful if the authors could optimize the presentation style throughout the manuscript.
Response 15: Thank you for your valuable feedback regarding the presentation style of our manuscript. We appreciate your suggestion to optimize the overall presentation, as clarity and coherence are crucial for effectively conveying our research.
In response to your comment, we have thoroughly revised the manuscript to improve its presentation style. This includes enhancing the organization of content, refining the language, and ensuring a more consistent formatting approach throughout the document. We believe these changes will significantly enhance the readability and professionalism of our work.
Thank you once again for your constructive input, which has helped us strengthen our manuscript.
Reviewer 3 Report
Comments and Suggestions for AuthorsDear authors,
The article provides valuable insights into object detection using different sensor approaches. It is well-written and easy to read. However, I have identified some questions that could enhance the quality and content of the article. Please refer to the comments below:
Abstract:
Please add the accuracy value of the convolutional neural network in distinguishing between the objects.
Line 17 – 18: This sentence has already been mentioned before.
Experiment and results:
Line 425 – 436: This information is already mentioned in the manuscript. I suggest that this paragraph be included in the material and methods section.
Discussion:
The discussion section is too weak. I recommend discussing the limitations and future applications of the results and approach. For example, how can the proposed equipment/technology handle noise data?
The affirmation …”can be used in traffic monitoring systems in cities, where radar can automatically recognize pedestrians, cyclists, and vehicles in real-time, supporting traffic light management and improving safety at crosswalks..” dealing with radar signals can be challenging in real-world scenarios due to the presence of multiple objects, artifacts, and noise, which can result in similar radar signals and compromise the correct object classification.
Will the results benefit from using a well-known object detection model such as YOLO, Mask2former, etc?
Author Response
Comments 1: Abstract: Please add the accuracy value of the convolutional neural network in distinguishing between the objects.
Response 1: Thank you for your valuable comment. We have added the accuracy value of the convolutional neural network in distinguishing between the objects to the abstract, highlighting that the developed solution achieves an accuracy of over 95%. Your feedback has significantly improved the clarity of our research.
Comments 2: Line 17 – 18: This sentence has already been mentioned before.
Response 2: Thank you very much for your valuable observation. We truly appreciate your feedback. In response, we have made substantial revisions to the entire abstract to enhance clarity and eliminate redundancy.
Comments 3: Line 425 – 436: This information is already mentioned in the manuscript. I suggest that this paragraph be included in the material and methods section.
Response 3: We sincerely appreciate your insightful suggestion. We have made the necessary adjustments as recommended. Thank you for your valuable feedback!
Comments 4: The discussion section is too weak. I recommend discussing the limitations and future applications of the results and approach. For example, how can the proposed equipment/technology handle noise data?
Response 4:
Thank you for your valuable feedback regarding the discussion section. We have made significant improvements to address this issue.
In the revised manuscript, we have elaborated on the limitations of our approach, particularly the necessity for a large input dataset and the time-consuming process of constructing the neural network to achieve optimal accuracy without employing transfer learning techniques.
We are aware of potential future applications, emphasizing how the proposed radar technology can be adapted to handle noisy data through advanced signal processing techniques and noise filtering methods.
The use of signal processing algorithms, such as wavelet transforms or adaptive filters, allows for the effective separation of useful signals from noise. This enables better analysis of radar data in the presence of noise.
The implementation of noise filtering algorithms, such as Wiener filtering or frequency-domain filtering, allows for the reduction of noise impact on radar analysis results. This approach provides higher object detection rates in challenging conditions, which is crucial for many industrial and commercial applications.
With these techniques, the proposed radar technology will not only be able to effectively identify and differentiate objects but also meet the challenges associated with processing noisy data, significantly enhancing its usability in practical applications.
However, this knowledge and the wide spectrum of research that would need to be conducted go beyond the scope of the work in this article. Nevertheless, it represents a further direction for our research.
Thank you for your insightful suggestion, which has undoubtedly strengthened our work.
Comments 5: The affirmation …”can be used in traffic monitoring systems in cities, where radar can automatically recognize pedestrians, cyclists, and vehicles in real-time, supporting traffic light management and improving safety at crosswalks..” dealing with radar signals can be challenging in real-world scenarios due to the presence of multiple objects, artifacts, and noise, which can result in similar radar signals and compromise the correct object classification.
Response 5: Thank you for this valuable comment, which indeed highlights significant challenges related to processing radar signals in real-world scenarios. We agree that the presence of multiple objects, artifacts, and noise can affect the accuracy of object classification, which poses a critical issue in the context of traffic monitoring.
Consequently, in the revised version of the article, we have thoroughly discussed the potential limitations associated with applying radar technology in real-world conditions. We are aware of methods for dealing with noise and disturbances, such as advanced signal processing techniques and filtering algorithms. Therefore, as a further direction for research, we should implement and test radar technology that can be supported by various signal processing algorithms, aimed at improving signal quality and enabling more effective object identification, even in challenging conditions.
Our future research will also focus on testing and optimizing algorithms to enhance the system's resilience to disturbances and improve classification accuracy in complex urban environments.
Thank you for bringing this issue to our attention, which has undoubtedly contributed to strengthening our approach to the topic.
Comments 6: Will the results benefit from using a well-known object detection model such as YOLO, Mask2former, etc?
Response 6: Thank you for your valuable comment. Indeed, utilizing well-known object detection models such as YOLO or Mask2former could benefit our research by enhancing the efficiency and accuracy of object classification.
Our primary goal has been to use radar as a sensor. Radar is inherently more discreet than cameras, especially in applications involving close proximity to people. It also offers a significantly greater range than the aforementioned optoelectronic sensors. This allows for the use of technologies like Synthetic Aperture Radar (SAR) to image terrain both during the day and at night, as well as in conditions with clouds, fog, or smoke. Furthermore, radar can detect moving objects and has the capability to identify concealed objects, such as those hidden by vegetation or leaves.
However, the echo received by the radar contains noise that complicates accurate detection. There is also the phenomenon of multipath propagation and false alarms. From a military perspective, radar is easy to detect by adversaries due to the emission of electromagnetic waves. Additionally, radar imagery, such as that from SAR, can be challenging to interpret and requires skilled analysts.
We plan to explore the possibility of integrating these advanced models with our radar technology in future work. These models are known for their high performance in detecting objects in complex environments, which could significantly improve our results, especially in real-world conditions where there are many interferences and similar signals.
Our future research will focus on assessing how these models can be adapted to the specifics of radar signatures and how their application could impact the accuracy of object detection in dynamic urban settings.
Once again, thank you for this suggestion, which will undoubtedly contribute to the further development of our approach to the topic.
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsI have no additional detailed comments at this time. However, it is recommended to show the revised sections in the response and provide more detailed explanations of the changes made.
Comments on the Quality of English LanguageN/A
Author Response
Comments 1: I have no additional detailed comments at this time. However, it is recommended to show the revised sections in the response and provide more detailed explanations of the changes made.
Response 1: Thank you for your constructive feedback. We appreciate your suggestion and have made the necessary revisions to the manuscript. In our response, we have included detailed explanations of the changes made to each relevant section. Additionally, we have attached a file with color-coded markings to highlight the specific modifications. We believe these revisions enhance the clarity and quality of the manuscript. If you have any further comments or suggestions, please feel free to share them.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsAuthors have addressed all my observations, further improving the manuscript.
Please find below few minor recommendations:
1. title of section 3 could be further detailed. A section title entitled "Method" seems not attractive in my opinion.
2. I would still optimize the introduction of Section 3: "The paragraph mainly focuses on the uRAD USB v1.2 radar, which functions at a frequency of 24 GHz, along with the specialized software used for its operation. It also explains how raw radar echo signals are transformed into spectrograms through the application of the STFT." "paragraph" could be replaced by section.
3. Introduction of section 4 must be changed: "Authors should analyze the results and how they can be interpreted in light of previous studies and the working hypotheses. The findings and their implications should be discussed" - this seems a recommendation for the authors, not something that should appear in the manuscript.
Comments on the Quality of English LanguageEnglish is mostly fine.
Author Response
Comments 1: title of section 3 could be further detailed. A section title entitled "Method" seems not attractive in my opinion.
Response 1: Thank you for your valuable feedback regarding the section title. We have taken your suggestion into account and have changed the title of Section 3 to "System Design and Implementation." We believe this new title more accurately reflects the content of the section and enhances its appeal.
Comments 2: I would still optimize the introduction of Section 3: "The paragraph mainly focuses on the uRAD USB v1.2 radar, which functions at a frequency of 24 GHz, along with the specialized software used for its operation. It also explains how raw radar echo signals are transformed into spectrograms through the application of the STFT." "paragraph" could be replaced by section.
Response 2: Thank you for your suggestion regarding the introduction of Section 3. We have revised the content to enhance clarity and detail. The new text is as follows:
The section mainly focuses on the uRAD USB v1.2 radar, which operates at a frequency of 24 GHz, making it suitable for various applications, including short-range surveillance and target detection. This radar system is designed to provide high-resolution data, enabling accurate measurements of distance and speed. In addition to discussing the radar's specifications, the section elaborates on the specialized software that facilitates its operation. This software plays a crucial role in configuring the radar parameters, collecting data, and processing the received signals. It provides users with an intuitive interface for real-time monitoring and control, ensuring optimal performance during radar operations. Furthermore, the section explains the process of transforming raw radar echo signals into spectrograms using the Short-Time Fourier Transform (STFT). This technique involves analyzing the radar signals in small time windows, allowing for a detailed frequency representation of the signals over time. By applying STFT, the section illustrates how the radar can effectively visualize the frequency components of the received signals, which aids in identifying and distinguishing between different targets within the radar's operational environment. This transformation is essential for enhancing the interpretability of radar data, ultimately leading to improved decision-making in various applications.
We believe this revision addresses your concern by replacing "paragraph" with "section" and by providing a more comprehensive overview of the radar system and its software.
Comments 3: Introduction of section 4 must be changed: "Authors should analyze the results and how they can be interpreted in light of previous studies and the working hypotheses. The findings and their implications should be discussed" - this seems a recommendation for the authors, not something that should appear in the manuscript.
Response 3: Thank you for your valuable comment. We agree that the introduction of Section 4 requires modification. The new text will read:
"In this section, the authors analyzed the results and their interpretation in light of previous studies and working hypotheses. The findings and their implications have been discussed in the broadest possible context. Furthermore, future research directions have been highlighted."
We believe that this change will enhance the coherence of the manuscript and be more appropriate in the context of the discussion of the results. Thank you for your suggestion!
Author Response File: Author Response.pdf