Social Network Analysis Based Localization Technique with Clustered Closeness Centrality for 3D Wireless Sensor Networks

Round 1

Reviewer 1 Report

The manuscript is somewhat improved with respect to the previous round. However, its quality is still far from being acceptable.

Overall, the text is quite confusing, especially in the first two sections, needing for a restructuring of the presentation in a more schematic, readable fashion.

The Discussion is still unsatisfactory and needs integration in terms of comparison with existing studies in the related field.

The Conclusions should be better defined, with a stronger take-home message to the reader and an information about the possible applications of such approach in practice.

Overall, English language and grammar should be revised, and typos should be checked and corrected.

Author Response

Author’s reply to the reviewers’ comments of the manuscript Electronics-777908 entitled “Social Network Analysis Based Localization Technique with Clustered Closeness Centrality for 3D Wireless Sensor Networks”

We would like to thank the editor to give us a chance to revise our manuscript for possible publication in MDPI Electronics. We appreciate the valuable comments and constructive suggestions from all reviewers, and hereby we provide a detailed response to each comment from each reviewer and indicate the changes that have been made in our revised manuscript. All changes are also highlighted with underline and red color in our revised manuscript. The major changes are summarized as follows:  

1. We have added paragraph sentences in Section 1 and Section 2.
2. We have highlighted the key features of our proposed schemes and revised the manuscript as per the reviewer suggestion.  
3. We have revised the conclusion part according to the schemes discussed in the paper.
4. Extensive English editing has been done.

 

 

 

 

 

 

1. Reply to Reviewer 1
2. The manuscript is somewhat improved with respect to the previous round. However, its quality is still far from being acceptable.

Overall, the text is quite confusing, especially in the first two sections, needing for a restructuring of the presentation in a more schematic, readable fashion.

Authors’ reply: We thank the reviewer for his/her positive comments. We have rephrase and restructure the whole paper again as per reviewer comments. The first two sections are again edited and highlighted in red from pp. 1 to 3. We have added some necessary discussion in Section I and Section II.

 

2. The Discussion is still unsatisfactory and needs integration in terms of comparison with existing studies in the related field.

Authors’ reply: The complete paper in terms of idea and discussion is again revised carefully. The comparison part in Section 6.4 is enhanced as shown in pg.17.

 

3. The Conclusions should be better defined, with a stronger take-home message to the reader and an information about the possible applications of such approach in practice.

Overall, English language and grammar should be revised, and typos should be checked and corrected.

Authors’ reply: The conclusion part is revised. Pp. 17-18. We have restructured the text and typos are checked carefully in a revised version.

 

 

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors introduce a clustering mechanism in wireless sensor networks based on the closeness centrality metric and by exploiting the social network analysis.

The paper is overall well written and easy to follow. The provided analysis of the proposed clustering method is concrete and correct.

The proposed approach is very interesting based on its holistic nature.

However, there are some minor comments that the authors should consider towards improving the quality of their manuscript. The comments are listed as follows.

In Section 3, the authors should better clarify the adopted wireless communications model, e.g., Singhal, Chetna, and Swades De, eds. Resource allocation in next-generation broadband wireless access networks. IGI Global, 2017. The reviewer advises the authors to be specific in the formulation of the nodes’ channel gains, why the proposed models are adopted, i.e., free space loss models, what is the transmission power boundaries of the nodes, etc. Please provide the appropriate references wherefrom the proposed wireless communications model is adopted.

The authors exploit the social network analysis in order to conclude to the clustering mechanism based on the closeness centrality metric. How the exploitation of big data in social network analysis, e.g., Thai, My T., Weili Wu, and Hui Xiong, eds. Big Data in Complex and Social Networks. CRC Press, 2016, can improve the clustering mechanism introduced in this paper. The authors should enrich the provided literature review, which is currently very limited.

The authors should provide some comparative numerical results to other relevant research works from the recent literature in order to show the benefit of adopting their proposed framework. The presented numerical results show only the pure performance of the proposed framework, however without a comparative study to other approaches, the reader is not able to understand the benefits of the proposed framework.

There are other clustering mechanism proposed in the literature based on social and physical parameters of the examined system, e.g., Tsiropoulou, E., K. Koukas, and S. Papavassiliou. "A socio-physical and mobility-aware coalition formation mechanism in public safety networks." EAI Endorsed Trans. Future Internet 4 (2018): 154176. The authors should provide a thorough literature review, identify the existing research gap in the literature and better position their paper and their motivation regarding the proposed research. Section 2 needs to be revised accordingly.

The authors should devote a new section to explain and analyze the computational complexity of the proposed framework.

 

 

Author Response

Author’s reply to the reviewers’ comments of the manuscript Electronics-777908 entitled “Social Network Analysis Based Localization Technique with Clustered Closeness Centrality for 3D Wireless Sensor Networks”

We would like to thank the editor to give us a chance to revise our manuscript for possible publication in MDPI Electronics. We appreciate the value comments and constructive suggestions from all reviewers, and hereby we provide the detailed response to each comment from each reviewer and indicate the changes that have been made in our revised manuscript. All changes are also highlighted with underline and red color in our revised manuscript. The major changes are summarized as follows:  

1. We have added paragraph sentences in Section 1 and Section 2.
2. We have highlighted the key features formulation of the nodes’ channel gains.  
3. We have revised the conclusion part according to the schemes advised by the reviewer.

 

 

 

 

 

 

 

1. Reply to Reviewer 1
2. The authors introduce a clustering mechanism in wireless sensor networks based on the closeness centrality metric and by exploiting the social network analysis.

The paper is overall well written and easy to follow. The provided analysis of the proposed clustering method is concrete and correct.

The proposed approach is very interesting based on its holistic nature.

Authors’ reply: We thanks the reviewer for his/her positive comments.   

 

2. However, there are some minor comments that the authors should consider towards improving the quality of their manuscript. The comments are listed as follows.

In Section 3, the authors should better clarify the adopted wireless communications model, e.g., Singhal, Chetna, and Swades De, eds. Resource allocation in next-generation broadband wireless access networks. IGI Global, 2017. The reviewer advises the authors to be specific in the formulation of the nodes’ channel gains, why the proposed models are adopted, i.e., free space loss models, what is the transmission power boundaries of the nodes, etc. Please provide the appropriate references wherefrom the proposed wireless communications model is adopted.

Authors’ reply: We thanks the reviewer for his/her positive suggestions. The wireless signal model and channel gains explained in Section 3.1 [31]. However, we again added some statements in favor of our proposed model in Section 4 of our “System Model”. Refer to pg.4 and pg.7. Furthermore, the power boundaries of nodes is also explained in Section 6.2. Refer to pg. 14.    

[31] Halder, S. J., Giri, P., & Kim, W. (2015). Advanced smoothing approach of RSSI and LQI for indoor localization system. International Journal of Distributed Sensor Networks, 11(5), 195297.

3. The authors exploit the social network analysis in order to conclude to the clustering mechanism based on the closeness centrality metric. How the exploitation of big data in social network analysis, e.g., Thai, My T., Weili Wu, and Hui Xiong, eds. Big Data in Complex and Social Networks. CRC Press, 2016, can improve the clustering mechanism introduced in this paper. The authors should enrich the provided literature review, which is currently very limited.

Authors’ reply: We thank the reviewer for his/her positive comments. However, our main focus is to derive a new mechanism for 3D based localization algorithm. The main theme of this work is not a clustering mechanism. We are basically drawn a cluster upon which we are computing the sensor node position if it presented with in a cluster/ area. That is why we are not explaining the clustering review detailed in this work. This will mislead the readers and reviewers in two different directions.   

Refer to Figure 4 on pg. 9 v1,v2,v3, and v4 is a cluster. We are forming this cluster by making the shortest CC values explained in SNA. So, the primary task of this work is to compute the position of node N lies between v1,v2,v3,v4.

 

4. The authors should provide some comparative numerical results to other relevant research works from the recent literature in order to show the benefit of adopting their proposed framework. The presented numerical results show only the pure performance of the proposed framework, however, without a comparative study to other approaches, the reader is not able to understand the benefits of the proposed framework.

Authors’ reply: We thank the reviewer for his/her suggestions. We have revised the paper and add some numerical values in the contract to the other localization algorithms presented in the literature review. We have added some values in pp. 16-17.

 

5. There are other clustering mechanisms proposed in the literature based on social and physical parameters of the examined system, e.g., Tsiropoulou, E., K. Koukas, and S. Papavassiliou. "A socio-physical and mobility-aware coalition formation mechanism in public safety networks." EAI Endorsed Trans. Future Internet 4 (2018): 154176. The authors should provide a thorough literature review, identify the existing research gap in the literature and better position their paper and their motivation regarding the proposed research. Section 2 needs to be revised accordingly.

Authors’ reply: This work is completely focused on wireless sensor network localization. The use of clusters or formulation of the cluster is not a primary goal of this work.

 

6. The authors should devote a new section to explain and analyze the computational complexity of the proposed framework.

Authors’ reply: We thanks the reviewer for his/her positive comments. The complexity analysis is discussed in several part of the paper however, we have added a separate section on pg. 17.

 

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The paper was improved with respect to the previous version, but several typos and language mistakes are still present and need to be further revised.

Author Response

Dear Reviewer:

we have again change the contents, grammatical mistakes, and typos.

Thanks for the comments provided on this paper.

 

Regards

 

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors have performed a very poor revision of the manuscript and the reviewers’ comments have not really addressed, but instead they have been bypassed with some very general and vague statements.

Specifically, the adopted wireless communications model has not described in the revised manuscript. The authors have provided a general statement “The wirless signal model and channel gains for RSSI computation is taken from [31] as shown in section 3.1” (by the way, there is also a typo in this statement). After searching [31] reference there are no provided information regarding the wireless communications model, neither the formulation of the nodes’ channel gains. The authors should consider getting advice from a book, e.g., Singhal, Chetna, and Swades De, eds. Resource allocation in next-generation broadband wireless access networks. IGI Global, 2017, to better understand themselves what the wireless communications model is and then thoroughly present it in the paper.  The authors should elaborate more on why the proposed models are adopted, i.e., free space loss models, what is the transmission power boundaries of the nodes, etc.

Part of this paper is exploiting the concept of clustering, in order to create clusters upon which the authors compute the sensor node position if it is presented with in a cluster. However, the authors avoid to discuss the clustering concept and its benefits and drawbacks of an efficient clustering formation, upon which the authors base the rest of their analysis. Thus, the authors whould elaborate on how the exploitation of big data in social network analysis, e.g., Thai, My T., Weili Wu, and Hui Xiong, eds. Big Data in Complex and Social Networks. CRC Press, 2016, can improve the clustering mechanism used and build upon that in this paper. The authors should enrich the provided literature review, which is currently very limited.

The authors should provide some comparative numerical results to other relevant research works from the recent literature in order to show the benefit of adopting their proposed framework. The presented numerical results show only the pure performance of the proposed framework, however without a comparative study to other approaches, the reader is not able to understand the benefits of the proposed framework.

The authors should devote a new section to explain and analyze the computational complexity of the proposed framework. Currently, the authors just say how much the computational complexity is, without explaining the reasoning behind that and without providing the theoretical computational complexity.

 

 

Author Response

The comments file is attached

Author Response File: Author Response.pdf

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

I have previously reviewed this paper. I haven't seen my commnets being taken into account or any reply t them.

For example, authors did not respond to my previous comment 11: Lines 7-8 "so we propose to use closeness centrality (CC) as one of the simplest metrics in measurement to ...". It is not clear how closeness centrality is one of the simplest metrics since its complexity is O(V^3).

Furthermore, in Section 6.4 a discussion regarding comparison with existing methods is missing. Moreover, authors may consider compare their method to existing ones in terms of accuracy, time, memory and computational resources.

For more details please seemy previous comments. And English language should be improved.

Reviewer 2 Report

The authors have made minor changes to the previous version. Most of the concerns are not addressed. The answers to the reviewers have not been integrated into the paper. Unfortunately, there are still many issues that remain unclear.

The title of the work has still strange wording. The authors write "clustered based". It should be "cluster based"; note: the wording clustered based appears only some few times in the literature, in papers written by non-native English speakers.

The use of "Social Network Analysis" is still inappropriate in your case. If you have social agents that carry network nodes, SNA would make sense. However, this is not described in the paper. For unspecified networks, "standard" Network Analysis would be the method of choice. Network Analysis uses a variety of methods, including diverse definitions of centrality metrics. See, for example, the work by Barabasi et al (there are many papers about network analysis). See also the entry in Wikipedia https://en.wikipedia.org/wiki/Network_theory to get an impression about network theory. SNA is the application of network analysis to social relations. The papers discussing SNA in connection with mobile networks etc. have social agents wearing these or controlling these in some way. In your case, you do not mention any movements of social agents as a preliminary for your work (with the exception of the cattle-example, but no properties from the cattle-example are used in your proposed algorithm), thus the term "Social" is misleading. Further, nothing in your proposed algorithm depends on the "social"-aspect.

Figure 1: It is still not explained in the paper how this graph came along. Please describe how this graph was developed. In your answer, you wrote that you used an extensive literature review; this is fine (and should be written out in the paper), but you need to also explain why you grouped the technologies as you did. Further, in the figure, you could include the references to the technologies; also to the groups of technologies when other authors have classified these in a similar way.

Further, several of the terms are not properly defined, for example range-based, range-free, etc. (Of course, the reader could look this up, but in that case, your paper would not be self-contained). I see that the authors added more related work, which is good.

Line 74: Kalman with a capital K

lines 69-74: several techniques (e.g., fingerprinting) are used without explaining how these work or what they are based upon.

Section 4 is very short. Possibly, Sections 4 and 5 could be joined.

The new text in Section 5 is quite vague. Patient monitoring might include body area networks, which  you excluded in the responses to the reviewers. This is confusing. The example with cattle is unclear. More properties of these types of networks would be welcome, and how these properties would have an impact on your algorithms.

Section 5: The word "training phase" is used for machine learning, where a model is trained. In your case, you are not training a model, but you are initialising it. Therefore, the wording "Initialisation Phase" would be the right term.

The new lines 202-207 are not comprehensible, and the list of the four conditions does not contain complete sentences. Please revise and write some more explanations.

Although I see the value of randomising the sensor nodes, the line 4 does not belong to the algorithm, but to the preconditions of the algorithm. As you formulate this, Line 4 of your algorithm does not have any effect. Please rewrite your algorithm so that you have a precondition part (which would be valid for all the algorithms), and then the algorithm(s). It is till not made evident in the paper where your algorithm(s) run.

Section 5: It is unclear how the example you use for simulation relates to the scenarios you previously described (cattle, healthcare), including how the number of nodes and the area relate to these examples. In the simulation, new terms appear, such as power loss, path loss, etc; these are not part of your algorithms.

Inconsistency: p.7: you write outdoor, p11. you write indoor.

Please describe more in detail what the diverse Figures show. Also provide what the diverse scales are.

In Figure 6, it is not explained how the nodes are deployed. Is there a parameter that places most of them in the middle of this area? This would be an important parameter for the understanding of the results.

The localisation error of 0.32m / 1.35m is obviously a result. However, I doubt that this is true for all configurations. Did you simulate other setups, so that one could use the monte-carlo method, or is this only for this one simulation? There is no comparison to the localisation accuracy in other work using different methods. There should be a discussion section to compare this.

The results are not discussed. Please discuss the findings, also the result for node 7 is only explained as comment to the reviewer, but not discussed in the paper. What does it mean that one node has this high localisation error, e.g., when it is an anchor (the paper does not mention if this is an anchor or a "unknown node".

Figure 9 is incomprehensible. Is this a simulation result? What does it mean to have 1.5 anchors and a CDF of 0.5. In your simulation, you write about 100 anchors; does this mean that the CDF is 99.0 ? Please explain in the paper.

The energy-considerations from line 302ff come as a surprise. They are not mentioned in the abstract, there are no literature references to models how to calculate, and how you could compare transmission costs. What are the energy costs in your simulation? The energy part does not lead to any new result; only some formulae are presented without discussion. The same is for Section 6.3. There is no relation to the rest of the paper; there is no discussion about the results. Lines 338-348: Incomprehensible text. Please describe what you are doing. Why use different number of nodes from the previous simulation? Are there any differences from other, similar experiments from the literature? What are "difference distances" ?

line 340: do you mean "affects" or "has an impact on" ? Although I see the graph, it is unclear what it means and how it is to be interpreted. The distance is in meters? or something else? What does this "jittery" behaviour of multipath loss mean for the accuracy, as signal intensity has an impact in RSSI? It is unclear how Figure 10 and Figure 11 are in relation to each other. Please write in words what the blue line and the red line represent in Fig 10.

Section 6.4 contains only a drawing, no explanation. How did this graphs come along for the three other methods? Did you simulate this, with which setup? Or did you get the results from the literature?

There is no discussion section.

The algorithms and the results are not compared with findings from the literature.

The paper should include a paragraph about the research contributions in the introduction section, including some hypotheses or research questions that can be verified or falsified in the discussion section. For example, the authors could claim that their "cluster-based localisation algorithm using closeness centrality is more accurate and uses less energy than other algorithms, hereunder a and b and c". In the discussion section, you could then discuss for which properties your algorithm performs better, worse, undecided in relation to the other technologies. The new text introduced on p.7 is more a rationale, rather than a research question.

Further, the authors do not write anything in the paper about which information must be exchanged between which type of nodes. A sketch of some kind of protocol, and which calculation is calculated where would be welcome. In the notes to the reviewer, the authors mention something about this, but there is some tacit knowledge the authors may have about their method; this knowledge is not communicated to the reader. Please make this clearer in the paper.

The authors have still the "equal contribution" footnote, although the "author contributions" section suggests otherwise.

There is a variety of language issues, including incomprehensible sentences. Please, ask native English speakers to proofread.

line 53: the word "popular" is not a good word to use her. Please consider: "The currently used transmission technologies include 6LoWPAN, LoRA, ...." (note that the sentence in your version is incomplete).

Reviewer 3 Report

It seems that there are only 2 points in the plot of figure 9. Please add more points in this figure.

Recent publications in this topic may be considered to add and compare, such as

[1] Qian Huang, Yuanzhi Zhang, Zhenhao Ge, Chao Lu, "Refining Wi-Fi based indoor localization with Li-Fi assisted model calibration in smart buildings", 16th International Conference on Computing in Civil and Building Engineering, pp. 1358-1365, 2016.

[2] Jian Chen, Gang Ou, Ao Peng, Lingxiang Zheng, Jianghong Shi, "An INS/WiFi indoor localization system based on the weighted least squares", Sensor, vol. 18, no. 5, 1458, 2018.

[3] Zhongliang Deng, Xiao Fu, Qianqian Cheng, Lingjie Shi, Wen Liu, "CC-DTW: an accurate indoor fingerprinting localization using calibrated channel state information and modified dynamic time warping", Sensors, vol. 19, no. 9, 1984, 2019.

Reviewer 4 Report

The paper presents an interesting topic in the specific field. However, the quality of presentation is quite low, therefore it should be completely revised before further processing. Some tips are indicated below: 

- Abstract is badly written. It leaves no space for discussion and/or conclusion parts, with the results briefly placed at the end of the section. Please, re-write.

- Introduction is quite confusing both in terms of text organization and English clarity. Please, re-write.

- In Section 2, the text preceding Figure 1 poorly explains the figure itself, as the figure shows processes that are not cited in the text (see left part of Fig.1).

- A discussion is lacking.

- References are badly formatted. Please, check and revise.

- Several typos are present throughout the text. Please, revise.

Reviewer 5 Report

The paper uses Social Network inspired concept of closeness centrality for localization in 3D WSNs. The paper has many issues, some of which are discussed below.

The language in the paper is very weak, making the presentation of the paper and its reading quite difficult. Apart from many typos and misspellings, there is constant inconsistency in terms of used tense and misuse in sentence formation throughout the paper. A thorough and careful pass of the paper in terms of language is necessary. 

Parts of the text are highlighted in red colour which I fail to understand in a submitted version that is not a corrected version to previous submission.

In terms of scientific presentation, the article lacks significantly in depth, as it takes a long portion of the manuscript to discuss and describe ideas and concept that one would normally find in textbooks or introductory classes for WSNs and Complex Networks. I recommend the authors focus on the ideas and concepts absolutely necessary to draw the picture of their contributions.

In terms of contributions, the reviewer is skeptical about their significance and novelty. The idea of finding the closer neighbours for localizing devices is not a new one, and the way the measure of closeness centrality is presented and used does not add any value to the excessive body of literature that is in this direction. (A small note here on the related works section, this fails to discuss the state-of-the-art and place this work in the existing body of literature, answering the question "which research gap is attempting to fill"? - 3D vs 2D deployments could be a valid point, however, its current treatment does not explain the work, nor it is conducted in a rigorous and interesting way, in this reviewer's opinion.

The reviewer feels that the paper needs more novel ideas on the subject in order for the work to be consdered for publication. Perhaps the idea of using social network tools in WSNs for 3D deployments could be improved by considering high-dimensional spaces and rigorously defining the concepts of centrality for those spaces. 

Reviewer 6 Report

The proposed algorithm relies on the propagation model of RSSI. The major problem of the propagation model is that it cannot be used in real environment by the obstacles, multi-path problem, other radio signal interference and walls in indoor case. In Figure 11., the authors provide the pretty noise added signal propagation, but still it is never realistic. The author must implement real experiment or consider the realistic simulation setup.