Effectiveness of Data Augmentation for Localization in WSNs Using Deep Learning for the Internet of Things

Round 1

Reviewer 1 Report

The background and related work section needs to be relevant and comprehensive to backup the proposed approach.

The introduction and the related work sections should cover relevant and comprehensive references to justify the proposed work. There are a number of basic introductions of basic concepts e.g. intro to ANN.

The presentation of data augmentation section is very specific examples and are not general. The equation (equation. 9) describing the training data size (D_t) is not described as to how the relations between the variables are established.

The claimed contribution of this paper is increasing the amount of training datasets by creating a multi-virtual anchor around the actual anchors.

Methodology, used to create virtual anchors coordinates is basically regressing the coordinate of a real anchor and doesn’t really create another reference point for localization purposes rather than random probability given by Gaussian variations.

Comments for author File: Comments.pdf

Several grammatical errors. The paper surely requires extensive English language review. 

Author Response

Thank you so much for reviewing [Effectiveness of Data Augmentation for Localization in WSNs using Machine Learning for the Internet of Things] and providing valuable feedback. I genuinely appreciate your thoughtful insights and constructive comments. Your input has been incredibly helpful in refining my work, and I'm grateful for your expertise in guiding me toward improvement.

• I have enhanced both the related work and the introduction, providing additional information on the following aspects:
• I elaborated on fundamental concepts such as Artificial Neural Networks (ANN) with more detailed explanations and discussed their application in Section 4
• I introduced the concept of data augmentation in general, followed by a detailed explanation of how I utilized it to increase the input data for my work in Section 3.
• I provided detailed information on the creation of virtual anchors, explained Equation (9) in greater detail, and described the format of the augmented input data for the Deep Neural Network (DNN).
• Algorithm 1 elucidated how the variables were established in the data augmentation process.
• Recently, data augmentation has been employed to enhance the performance of machine learning models by expanding the dataset. Following a similar approach mentioned in the reference in Section 3, we utilized data augmentation to augment the dataset. This involves generating multiple copies of virtual anchors for each real anchor around its position, as detailed in Section 3.
• Adding a virtual anchor involves placing it close to the real anchor. For instance, in Figure 2, Anchor-1 is surrounded by five virtual anchors (details on generating the coordinates for the five virtual anchors). I used a span (distance between the real anchor and virtual anchor) of 3m. Subsequently, the span is multiplied by a Gaussian distribution to obtain five coordinates. Further details are explained in Section 3, specifically in Step 6 and Step 6 of Algorithm 1.

Author Response File: Author Response.pdf

Reviewer 2 Report

Overall, the paper introduces a comprehensive solution that integrates deep learning techniques with data augmentation strategies to address the challenges of node localization in WSNs, emphasizing its potential impact on IoT applications. However, the following issues were spotted and must be considered: 

The presentation of the paper is very poor and it was hard to read it. It consists of many small subsections. It was hard to tell when the introduction ended and when the literature, the proposed solution section, or the implementation section started. 

The abstract is poorly written and usually you do not start immediately with what the paper is proposing. You need to start with an opening sentence or background information about the topic, discuss the issue, what the paper is proposing, and after that talk a little bit about your achieved results. Try to enrich your abstract because it is 142 words you can include more details to reach the 200 words. 

The introduction is very short as well as other sections and does not prepare the reader well for the main issue of concern and what is the motivation of the work. Also, at the end of your introduction, you needed to state your paper structure. 

You do not have to define the acronym of the WSN every time it is mentioned. Since you mentioned it first in the abstract there is no need to write the full name every time in the following sections. Please check that for all acronyms.

In the problem formulation section, your questions section is not clear for the reader so please emphasize this section. Also, not all symbols used in the equations are defined! After that, you started straightforward to present your results. I struggled a lot in reading the paper. 

Figure 1 is not clear. 

Please ensure to refer to your figures inside the text e.g. figure 2. Check that for all figures. 

All figure qualities must be improved as they are very hard to read. 

The conclusion section needs to be rewritten so that it gives a good summary of the paper's contribution, achieved results, and future work. 

The list of references is very limited. Please enrich your list of references and use more work from the MDPI journals. 

The entire paper must be proofread and revised 

Author Response

Thank you so much for reviewing my [Effectiveness of Data Augmentation for Localization in WSNs using Machine Learning for the Internet of Things] and providing valuable feedback. I genuinely appreciate your thoughtful insights and constructive comments. Your input has been incredibly helpful in refining my work, and I'm grateful for your expertise in guiding me toward improvement. 

• I reviewed the figures and made the necessary corrections.
• I changed the structure of the abstract, incorporating background information about the topic and adding more sentences to increase its length.
• I included additional relevant information in the introduction. Instead of defining WSNs, I opted to use the acronym.
• I defined the symbols in all equations and highlighted this section.
• Before explanation of my results, I provided a brief description.

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript highlighted the effectiveness of the data augmentation strategy for range-free localization in WSN/IoT applications based on Dv-hop with the deep neural network. Some corrections are needed as listed:

1) minor English correction.

2) missing the definition of the term range-free; why use distance while it is range-free?

3) some references are outdated; please refer to the latest research.

4) all figures must be more readable (labels are too small) and blurry. 

5) please add the localization process flow.

6) is there any maximum communication range?

minor corrections

Author Response

• I have reviewed the grammatical errors in my work and corrected them.
• I added the definition of "range-free" in Section 1.
• All references and figures have been revised and updated.
• I included a localization process overview, as illustrated in Figure 1. Additionally, I changed the title of Section 1 to "Localization Process" (Thank you so much for this suggestion).
• In my work, I utilized an area for deploying the unknown sensor with a range of 100m, which aligns with other studies where the maximum range is 30m. If we increase it to 50m, it implies that the sensor will require more power, leading to an increase in the cost of the nodes.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Thank you for considering and responding to my comments. You have clarified a number of concepts in the paper. You have  improved the paper very well. I am not really convinced that the core idea presented really contribute to improving location accuracy via creating more datasets however the general idea of virtual anchors for localization is interesting. Your simulations suggest otherwise so i accept your arguments.

Still found some minor grammatical issues but nothing major. I careful read through the entire article should be sufficient. 

Reviewer 2 Report

The authors of the paper correctly corrected the paper, and they took into account all my remarks and included them in the correction of the paper.