kumoh national institute of technology
Networked Systems Lab.

Review Comment

NSL > Works@NSL> About Review> Review Comment
Rizki Rivai Ginanjar, Si Wan Kim, and Dong-Seong Kim, "Extreme Learning Machine-Based Real-Time Node Localization: UAV Approach", 2018 IEEE Global Communications Conference: Workshops: 9th International Workshop on Wireless Networking and Control for Unmanned Autonomous Vehicles , 09 - 13 December 2018, Abu Dhabi, UAE.
By :
Date : 2018-09-17
Views : 15

Dear Mr. Rizki Ginanjar,

Thank you for submitting your paper #1570475472 ("Extreme Learning Machine-Based Real-Time Node Localization: UAV Approach") to 2017 IEEE Globecom Workshops. This year, IEEE Globecom Workshops received over 600 submissions. All papers underwent a rigorous review process. As a result of the reviews and the discussion among the workshop co-chairs, we regret that your paper could not be accepted for inclusion in the IEEE 2018 Globecom Workshops program.

The reviews are available at http://edas.info/showPaper.php?m=1570475472, which we hope you will find useful to improve your paper. Thank you for submitting your paper to 2018 IEEE Globecom Workshops, and we hope that you will be able to attend the conference in Abu Dhabi in December 2018.

Sincerely,

IEEE GLOBECOM 2018 Workshop Chairs
Kumudu Munasinghe, Workshop Co-Chair
Jaafar Elmirghani, Workshop Co-Chair



================================================================================================
Reviews
3 Reviews
Review 1 (Reviewer A)
Relevance and timeliness Technical content and scientific rigour Novelty and originality Quality of presentation
Little (2) Questionable work with severe flaws. (1) It has been said many times before. (1) Readable, but revision is needed in some parts. (3)
Strong aspects (Comments to the author: what are the strong aspects of the paper)
The paper aims to achieve localization of nodes with unknown location.
Weak aspects (Comments to the author: what are the weak aspects of the paper?)
It is unclear how the proposed method performs in comparison to no localization at all. It appears that there is something wrong with the proposed approach.

When looking at Fig. 8, it is clearly visible that the localization error is around half the coverage radius. Predicting the resulting location using a uniform random distribution within the coverage radius will yield very similar results.
Recommended changes (Recommended changes. Please indicate any changes that should be made to the paper if accepted.)
The authors should depict the baseline that random location estimation within the coverage radius would deliver. (This would be a linear function y = r/2)
Based on this knowledge the authors should compare the results to a simple non-linear-least-squares approach using the RSSI and a simple free-space propagation model with correct parameters.
Submission Policy (Does the paper list the same author(s), title and abstract (minor wording differences in the abstract are ok) in its PDF file and EDAS registration?)
yes
Review 2 (Reviewer B)
Relevance and timeliness Technical content and scientific rigour Novelty and originality Quality of presentation
Excellent (5) Solid work of notable importance. (4) Some interesting ideas and results on a subject well investigated. (3) Readable, but revision is needed in some parts. (3)
Strong aspects (Comments to the author: what are the strong aspects of the paper)
The proposed approach to localize nodes from beacon signals transmitted by UAVs is timely. The used method based on a single hidden-layer feedforward network interesting and relevant.

In general the paper is well written.
Weak aspects (Comments to the author: what are the weak aspects of the paper?)
The localization accuracy seems modest, namely approx. 42-45% of the UAV coverage radius.

The text should be cross-checked by a fluent English speaker (see detailed comments below).
Recommended changes (Recommended changes. Please indicate any changes that should be made to the paper if accepted.)
The investigations focus on localizing in one block of Fig. 1. In the generalization to the multi-block scenario, the unknown nodes will have to make a selection among the beacon signals sent by all UAVs. It is not clear how this is dealt with.

The following is rather a remark: It is surprising and actually baffling that the input weight and bias coefficients can be arbitrarily (randomly) selected. Only the appropriate setting of the output weights matters. I am wondering whether some additional performance gain can be achieved by optimizing the input weights and biases as well.

Detailed comments:
There are here and there some incorrect or incomplete sentences. Here are some examples:
P1, C1, Intro., 2nd par.: infers should be replaced by refers.
P1, C2, 3rd par.: the construction which this algorithm shows is incorrect.
P2, Subsec. II.A, 1st par.: Assuming should be replaced by Assume.
P2, Subsec. II.A, last sentence The RSSI values on its side: Its meaning is unclear.
P3, Caption of Fig. 4: Delete from.

P2, Subsec. II.A, last paragraph: The problem referred to is an ambiguity problem (in determining the location from the RSSI values).
Fig. 2: The coverage radius r should be depicted in this figure, as well as W and L.
P3, C1, par. below (3): the term stage is inappropriate. I would replace it by something like time index. By the way, depends should be replaced by depending. The following sentence is also not correct.
P2, C2, 2nd par. below (6): It is not clear how the normalization of the input vector (claimed to be so that it lies in the range (-1,+1)) is performed.
P3, C1, 2nd line: N is the number of training samples; j is the index of said samples.
Submission Policy (Does the paper list the same author(s), title and abstract (minor wording differences in the abstract are ok) in its PDF file and EDAS registration?)
It does!
Review 3 (Reviewer D)
Relevance and timeliness Technical content and scientific rigour Novelty and originality Quality of presentation
Good (4) Valid work but limited contribution. (3) Some interesting ideas and results on a subject well investigated. (3) Readable, but revision is needed in some parts. (3)
Strong aspects (Comments to the author: what are the strong aspects of the paper)
The architecture design of this localization approach allows good scalability of the WSN.

Despite the localization approach is tested only in simulated scenarios, experiments are clear and well explained.

In general, the paper is well written and easy to understand.
Weak aspects (Comments to the author: what are the weak aspects of the paper?)
It is not clear whether the localization algorithm is based on supervised learning. If it is the case, it is not clear how the original coordinates x and y are included in the training dataset.

Localization algorithm is tested in a simulated environment. Due to environmental conditions of real scenarios are not stable over time, it would be useful to test the localization algorithm in some real scenario. Moreover, RSSI can be affected by some environmental condition leading to decrease the localization performance of the localization algorithm. Therefore, it is important to consider some strategy to deal with this problem. It is not mentioned in this work.

Despite most of the paper is easy to understand, there are some sentences that are not very clear (e.g., in page 2 'where W is the width of field ' What is the field?. Is it the same as block?)
Recommended changes (Recommended changes. Please indicate any changes that should be made to the paper if accepted.)
Please consider comments in weak aspects section.
Submission Policy (Does the paper list the same author(s), title and abstract (minor wording differences in the abstract are ok) in its PDF file and EDAS registration?)
Yes, they are the same.