A Practical and Sustainable Approach to Determining the Deployment Priorities of Automatic Vehicle Identification Sensors

Round 1

Reviewer 1 Report

First, the study is similar to that of the studies to find the shortest path having obstacles or constraints. Here, the author uses flow speed or ratio through the route. Then by a formula, they evaluate an error rate. Moreover, after finding a random path, it is claimed to be that an NP-Hard problem is told to be converted to a polynomially solvable problem. These claims are hard to believe and disputable.

It is also seen that Algorithms 1 and 2 are some sort of integration using the nearest neighbor edge due to a genetic algorithm or probabilistic way. Then the whole proposed algorithm is presented to be converting an NP-Hard problem to be solved in polynomial time.  However, there are many algorithms as the shortest path being NP-Complete (NPC).” Considering the complexity and time consuming of BPM, we developed a data-driven random work algorithm to simulate plenty of paths to decompose BPM into sub-BPMs, i.e., each path corresponds to a sub-BPM, degenerating BPM from NP-hard to polynomial solvable” is found to be extremely pretentious in lines 111-112. NP-Hard problems cannot be solved in polynomial time as in halting problems however, if it is proven to be NPC, then it needs strong and serious proof. Then for special cases or a small number of instances, you can find a path in polynomial time. It is believed to be it is not the case as stated in the statement “ each path corresponds to a sub-BPM, degenerating BPM from NP-hard to polynomial solvable.” in line 113.   Moreover, in Remark-1 in line171 “The AVI-LP is one of the variants of MILP and is NP-hard. To be specific, if the heterogeneous distribution of vehicle paths turns to homogeneous, the AVI-LP degenerates to MILP.” Mixed ILP problems are known to be a decision or feasibility problems, it is more correct to say that MILP is NP-complete as well as NP-hard. “AVI-LP degenerates to MILP” in line 172 is another issue to be examined.

In Figure 5, there seems to be no logical basis to find the identification errors. As understood the smaller the better, however, each path’s error rate is multiplied by the next one. However, if the number of the multipliers increases then the values will decrease and you will get a wrong number or evaluation. In other words, multipliers must be equal for each path, which seems unrealistic.  

Additionally, it is hard to determine the aim and the algorithm is like a multi-objective shortest path as in Eq.1-3. 

There are editorial mistakes as seen in the abstract.

On the other hand, there is no comparison with the algorithms studying the same problem. This is a major drawback for such a research paper.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear authors,

 

Your article is very interesting and topical.

In order to improve its scientific level, it is recommended the following:

 - I recommend explaining more clearly the contribution to sustainability.

- I recommend improving the discussion.

 

Sincerely

                           Reviewer

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Dear Authors

Thanks for your good paper. However, I have the following suggestions and observations that may improve your paper.

1. In line 18 , the word NP-hard may be expanded for the first time. Subsequently, We can use the abbreviated form.

2. In line 98, you have mentioned it is not practical in the real-world. The reasons in your comments need more explanation.

3. In line 108, the word AND may be merged with the previous sentence in the proper way.

4. In line 170, MILP may be replaced by MILPP as per your writing style.

5. In line 173, what do you mean by HMHLP? Is it clearly explained up to conclusion as your proposition?

6. In line 176, what is meant for realistic-sized instances? It is better to mention it.

7. In line 495, the word effective may be replaced by effectively.

8. In line 499, the word initial  may be replaced by any positive and moderate word to highlight your research contribution.

9. There are no references for all equations stated in the text. Are they not derived based on some references?

10. You have claimed deployment score as a novel indicator. It is better to give more explanation on it

11.  What is the reason to use Gourobi Solver to solve DBPMs? A short description on the merit of the solver for your work is expected in the text.

12. You have developed BPM to solve AVI-LP. Though you have mentioned novel indicator, more focus on the overall novel contributions of the research is appreciated. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

The author has done a study on Reconstructing vehicles’ paths is essential for traffic management and city governance, which traffic sensors can achieve. As a typical sensor, the Automatic Vehicle Identification (AVI) sensor can observe the whereabouts and movements of the vehicles. In this article, we introduced an indicator named deployment-score to present the deploying priorities of AVIs for better reconstructing vehicles’ paths. The deployment score was obtained by solving a Bi-level Programming Model (BPM) to maximise the accuracy of recurring vehicle’s path and minimise the amount of AVIs. The solving process is data-driven, and a random-work-based decomposition method was developed to simplify BPM, degenerating it from NP-hard to polynomially solvable. The solutions of the decomposed BPM proved to be approximate Pareto optimum in a data-driven view. Furthermore, a PageRank method was presented to integrate the solutions. Thus the deployment-score was obtained. The proposed method is validated by Chengdu City, whose results demonstrate the remarkable value of our approach.

 

Find below some aspects that must be addressed by the authors:

1.       The authors must clearly state the novelty and main contributions of this work when compared with a large amount of literature available over the last decades.

2.        Add some more literature based on the application of traffic management and others in the introduction section. Some of them are given below, and the DOI of the given article is given below.

[1]         doi: 10.3390/app10134490 [2] doi.org: 10.3390/s151129316 [3] doi.org/10.1155/2022/9221211

3.       Need to improve figure resolution.

4.       The Problem description and assumptions describes in Section 2 is relatively simple. Better justify the use of this particular model. Moreover, why you have ignored non linearity in the model.

5.       Describe the “Random-walking method” in terms of components and other properties related to the consider work.

6.       Clarify how the equations are assembled and solved. How about the numerical and computational aspects?

7.       Comment on the computational/experimental accuracy and efficiency of the proposed approach.

8.       The results are sound; anyway, it would be great if the author can clarify the benefits of the presented approach when compared with some alternative model/software.

9.       The “Conclusions” section must be expanded.

10.   Kindly improve the English of the article.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 5 Report

1. Compare existing work with proposed work

2. Few questions are written in paragraphs and need to change accordingly

3. Add more results 

4. Enhance your introduction with recently published articles 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 4 Report

Accepted