kumoh national institute of technology
Networked Systems Lab.

RR. Ginanjar, A.P. Hermawan, J.M. Lee, D.S. Kim "UAV-Based Sensor Nodes Localization Scheme utilizing Shallow Neural Network" (R)
By : Ade Pitra
Date : 2020-03-13
Views : 670


Paper:IoT-9851-2020 UAV-Based Sensor Nodes Localization Scheme utilizing Shallow Neural Network
Authors: Prof. Dong-Seong Kim, Ginanjar, Rizki; Hermawan, Ade; Lee, Jae; Kim, Dong-Seong
Editor: Dr. Ing. Michele Nitti

Dear Prof. Kim,

I am writing to you concerning the above referenced manuscript, which you submitted to the IEEE Internet of Things Journal.

Although it has merit, based on the enclosed set of reviews this manuscript has been rejected for publication in the IEEE Internet of Things Journal.

You may wish to consider the comments of the reviewers and submit it to another journal.

If you have any questions regarding the reviews, please contact the Associate Editor: Dr. Ing. Michele Nitti michele.nitti.it@ieee.org. Any other inquiries should be directed to the Administrative Assistant: Mariola Piatkiewicz m.piatkiewicz@ieee.org

Thank you for considering IEEE Internet of Things Journal for publication of your work.

Prof. Honggang Wang
Editor-in-Chief, IEEE Internet of Things Journal
hwang1@umassd.edu, wanghbell@yahoo.com

• Associate Editor Comments, if any, are listed below:

Associate Editor: Nitti, Michele
Comments to Author:
(There are no comments. Please check to see if comments were included as a file attachment with this e-mail or as an attachment in your Author Center.)

• Reviewer Comments, if any, are listed below:

Reviewer: 1

Comments to the Author
This paper studies the problem of ground users localization using multiple UAVs as anchor nodes. While the authors investigate an important problem, this paper suffers from serious problems.

1- Some assumptions in the system model is not realistic. Since UAVs are floating in the air, obtaining the exact location of UAVs to use for the localization is not possible in most of the cases. For example, when the GPS is used for this, there is some error in the given location for UAVs. Furthermore, authors does not mention that how they obtain the exact location of UAVs. Furthermore, in realistic scenarios, ground users may have different altitudes. Therefore, the authors should solve the problem for 3D case.
2- The proposed method is too heuristic. There is not any theoretical analysis to prove the effectiveness of the proposed method. Thus, I believe that this paper may not be successful in opening new researches in this area.
3- The authors have not provided convincing reasons to use neural network for the localization. In fact, there are a lot of researches on cooperative sensor network localization with rigorous theoretical analysis and their merits is proved through simulations. Authors does not discuss the superiority of using neural networks in comparison with these methods. In fact, the motivation behind using the neural network is not clear.
4- Following the 3rd comment, I believe that the authors should compare their proposed method with other methods in the literature to show the effectiveness of their approach. For example, [14] is highly related to their work.

To sum up, unfortunately I recommend rejection.

Reviewer: 2

Comments to the Author
A WSN based node localization system using unmanned aerial vehicles and neural networks is proposed to reduce the required complexity.
My comments are as follows:
1) It is strongly recommended to review the whole paper for grammatical and editorial typos.
2) The figures are mostly explained after their appearance in the paper. For example, Figure 6 is on page 3, however, its explanation is come on page 4.
3) In page 3 section A it is mentioned that: The table of the saved RSSI values from one antenna of the UAV is shown in Fig. 5. In previous section it was mentioned that UAV has two antennas and transmits the beacon from both antennas. Please eplaine how it transmits and receives at the same time from one antenna?
4) Please explain how nodes with low transmit power have the ability to connect to UAV? The power consumption is important for both UAV and nodes. Considering the long distance between nodes and UAV, how the consumed power is managed?
5) The novelty of the paper should be illustrated more clearly in the introduction section by comparing with other papers, such as:
*Z. E. Khatab, A. Hajihoseini and S. A. Ghorashi, "A Fingerprint Method for Indoor Localization Using Autoencoder Based Deep Extreme Learning Machine," in IEEE Sensors Letters, vol. 2, no. 1, pp. 1-4, March 2018, Art no. 6000204.
**Hajihoseini Gazestani, A., Shahbazian, R. & Ghorashi, S.A. Decentralized Consensus Based Target Localization in Wireless Sensor Networks. Wireless Pers Commun 97, 3587–3599 (2017)
The introduction section should be rewritten to show the weak-points of the previous works and advantegous of your proposed method.
6) What is the difference between your method (ELM) and a one layer neural network? Where did you use the ELM, present the mathematical equations of this.
7) The scenario and its applications should be explained more clearly. Is the location of nodes is fixed? When the nodes know their location, is there any need to UAV? There are lot of questions about the details of the considered scenario that should be clearly explained.
8) In training stage, how the received information is mapped to the location (x,y) of node? Is it a regression or classification problem?
9) It is not enough to compare only with SVM and BP, other papers also should be seen and compared with.