Application of Wireless Sensor Network Based on Hierarchical Edge Computing Structure in Rapid Response System

Round 1

Reviewer 1 Report

This article presents a proposal for an Application of Wireless Sensor Network Based on Hierarchical Edge Computing, using local data processing, thus avoiding the over-allocation of network services and reducing communications latency with the cloud. This approach has as main focus the work on application layers that avoid the overload of communications, detecting errors early, preventing them from being sent to the cloud if they have errors or are incomplete, in a typical case study of a logistics operator.
The article is well written and the language is clear, however the number of support references is clearly insufficient, since there are several works on the theme presented here, and that should support the proposed solution. For example:
-> lines 29-46 needs others references, like this White Paper: "Internet of Things: Wireless Sensor Networks", available at https://www.ipwea.org/HigherLogic/System/DownloadDocumentFile.ashx?DocumentFileKey=e0619c58-f639 -080a-86c2-055ae9c8af4d;
-> lines 46-56 needs others references, reference 2 is not enough and it is not properly adjusted with the subject of the paragraph (it refers to λ-COAP ARCHITECTURE), it is more adequate to the first paragraph of the introduction (to replace the reference 1);
-> Line 111 MySQL and PostgreSQL needs references;
-> Line 117 Must include the MQTT (Message Queuing Telemetry
Transport) definition;

Other questions and suggestions include:
-> Figure 2. second EN-2 box, do you want to refer to EN-3? Algorithm for Process on EN-2 activate Process on EN-3.
-> Figures 3.c and 3.d: explain better, of values ​​close to normal values, to be classified as outliers.
Line 275: Malformed reference 4: see original document https://doi.org/10.1145/2342509.2342513
-> reference 8 and 10 missing access date.
-> reference 9 validate the reference with the 2020 publication "Lecture Notes in Electrical Engineering, vol 590. Springer"

Author Response

Please see the attachment. Thanks.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper presents a rapid response system architecture which includes the concept of hierarchical edge computing in Wireless Sensor Networks. Most of the paper is well written and minor English language review is necessary. 

However, after carefully reviewing this paper, some major technical and structural weaknesses have been raised as follows: 

1. In abstract, authors refer “minute response time” but the minute threshold was never stablished across the paper. Additionally, this type of “conclusion” is not in line with the results presented in the paper (mainly on table 3).

2. The authors use the words “rapid response” very often but never quantify them or detail them. What does “rapid response” really mean? 

3. The paper do not refer any kind of related work/state of the art. Does this mean that this work is the first one on this subject? If so, it must be clearly written by the authors, otherwise the related work is an important section in order to understand the state of the art but also to show how innovative is the work presented in the paper. 

4. In section 2.2., the figure 2 does not shows what the 3rd paragraph details. More concretely, the message queue/broker is not shown. 

5. In implementation section, the table 2 does not give the reader the implementation “big picture”. A block diagram or even the Node-RED flow would be more suitable…

6. First sentence of the last paragraph of section 3.2 is confusing and must be clarified. Additionally, authors must provide more details about the simulation stage and a comparison between the simulation context and the real context. For instance, does simulation simulates over-the-air message collisions, dead nodes, hotspots, etc.? Is the simulation tailored for specific wireless sensor network topology, communication interfaces and protocols? What is the main reason for the difference in average response time? What is the contribution of historical data retrieval for the response time? Authors must quantify and detail all this. 

7. Line 246. In the scientific context, what is a “relatively short period”?

8. Table 2. “Dara”, typo?

9. According to the last sentences of section 3.3, authors do not have yet all the answers that matter. Hence, the work reported on the paper is not finished yet. 

Based on the above comments, it is clear that this paper presents major flaws and the reviewer cannot suggest it’s acceptance. 

Author Response

Please see the attachment. Thanks.

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper presents an architecture combining WSN and edge computing to support the rapid response systems. Despite some of merits introduced in the work, there still exits obvious issues required to be revised and upgraded substantially.

1. Since the applications of edge computing in computing and sensor systems have been investigated widely in the literature, the work lacks a a review of related works. It is recommended to add a section Related Works as a foundation for developing the proposed systems.
2. The experimental configuration is simple and is insufficient to show (prove) the advantages of proposed systems. The performance of architecture should be examined through a set of configuration settings including changing the rate of IoT data generation, scale of IoT systems (number of nodes varries), and so on.
3. In addition, the proposed system should be compared with the related works to highlight the possible advantages and outperformance.
4. The figures showing the experimental results should be improved with respect to quality and usage of legends to highlight.
5. The work must be reviewed intensively to correct a lot of grammar typos, such as: no space between text and citation number within the text (line 36, , "...research[1].."). 

Author Response

Please see the attachment. Thanks.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The review improved the first version of the paper, however another set of questions need to be addressed.

Please, consider the point #3 of the first review:

"The paper do not refer any kind of related work/state of the art. Does this mean
that this work is the first one on this subject? If so, it must be clearly written by the authors,
otherwise the related work is an important section in order to understand the state of the
art but also to show how innovative is the work presented in the paper."

"Response 3: The WSN-Edge-Cloud architecture refers to several related researches such
as reference 9. The edge in a hierarchical structure with data processing tasks in different
computing complexity arises for the first time."

So, why not state that on the paper (around line 79)?

Please, consider the Point #4 of the first review:

"In section 2.2., the figure 2 does not shows what the 3rd paragraph details. More
concretely, the message queue/broker is not shown."

"Response 4: Brokers refers to the implementation of edge node in different grades. A
specification is added to line 118 – line 120."

Now the authors have added two sentences (lines 118-120) saying that "... communication between edge nodes shown
in Figure 2 is achieved by the brokers".

Some questions:
"... by the brokers"
Which brokers? You have not introduced "the brokers" before and figure 2 does not show them either.

"... communication between edge nodes ..."
Do you mean all edge nodes or just EN-2 type of edge nodes?

If brokers are so important components for your solution why aren't they included in figure 2?

Please, consider the Point #5 of the first review:

"In implementation section, the table 2 does not give the reader the
implementation “big picture”. A block diagram or even the Node-RED flow would be
more suitable..."

"Response: As a simulation environment, components list in Table 2 are implemented on
various distributed hardware/PCs. Instead of the hardware snapshot, more detailed
statistics analysis is presented to highlight the system performances."

How can authors assume that "providing more detailed statistics analysis" can solve the lack of information about the simulation environment?

In order to provide a good understanding of statistics to the future readers of your paper, you have to provide information ("test bed") about your simulation environment.
Why don't you provide an high level architecture of your simulation environment?

Please, consider the Point #6 of the first review:

"First sentence of the last paragraph of section 3.2 is confusing and must be
clarified. Additionally, authors must provide more details about the simulation stage and
a comparison between the simulation context and the real context. For instance, does
simulation simulates over-the-air message collisions, dead nodes, hotspots, etc.? Is the
simulation tailored for specific wireless sensor network topology, communication
interfaces and protocols? What is the main reason for the difference in average response
time? What is the contribution of historical data retrieval for the response time? Authors
must quantify and detail all this.

"Response: The experiments are enhanced with considering the impact of network scaleand sampling rate. The network scale is extended to 5000 nodes and 10000 nodes. One more sampling rate is considered during the experiment. Figure 4 and 5 are added to
give a clear vision of the statistics outcomes. The supporting data are attached in appendix A."

The authors' answer did not addressed all the questions raised on the first review. For instance:

"... does simulation simulates over-the-air message collisions, dead nodes, hotspots, etc.? Is the simulation tailored for specific wireless sensor network topology, communication interfaces or protocols?"

Again, in the reviewer opinion, authors must present the architecture of the simulated environment (as stated in point #5).

Author Response

Please see the attachment. Thanks.

Author Response File: Author Response.pdf

Reviewer 3 Report

Dear Editors,

The manuscript is significantly improved after revision. However, It can be accepted after an intensive review to correct minor errors of text editing, grammars.

Author Response

Some typos and grammatical issues are corrected. Please kindly find the change log in the revised paper. Thanks.

Round 3

Reviewer 2 Report

In my opinion, the article is now more clear for the (future) readers.

However, an important detail is still missing: as simulation does not take into account message collisions, dead nodes, hotspots, etc. (it assumes perfect communication conditions?) this must be stated somewhere in the document. For instance it could be stated as future work or stated in the simulation text section.

Author Response

By considering the reviewer’s suggestions carefully, a paragraph to state the potential future work that relates to the practical WSN communication issues is added to line 281-284.