A near Real-Time Monitoring System Using Public WI-FI Data to Evaluate COVID-19 Social Distance Measures

Round 1

Reviewer 1 Report

Please explain some of the differences between SmartWiFi and WiFi.

Please check again for alphanumeric errors, such as line 541-542

Author Response

Answers to Review 1

Please explain some of the differences between SmartWiFi and WiFi.

Thank you very much for your time reviewing the paper. The name SmartWifi is used in the paper with two meanings:

• SmartWifi is a commercial name chosen by the city council of Palma to identify the project to provide Wi-Fi connectivity in all the most popular tourist areas/streets of the city.
• SmartWifi is a name used in the paper to identify the Wi-Fi network deployed in Palma with improved capabilities or functions: discovery packets monitoring, automatic data access, user geolocation, between others described in the paper.

For the authors a Wi-Fi network provide the same service from the point of view of the final users, but without collecting data useful for Cities management.

Following your comment, and to avoid misunderstandings, we have included this information at the beginning of the Section 5.

Please check again for alphanumeric errors, such as line 541-542

Thank you again for your comments, the alphanumeric errors have been checked and corrected.

Reviewer 2 Report

1. The paper studies an important topic especially with the  COVID-19 pandemic.

I feel the paper lacks theoretical  details and explanation on several aspects of the paper. For example, the authors need to clarify more on how they can predict the distances between the people within a certain area (i.e. it is clear how to count the number of people connected to a certain access point, but it is not clear how to determine the distance between each person, thus knowing the social distance status.

2. The other problem I can see if the fact that the author assume that people are connecting to the mall or area WiFi AP. However, in the demonstrated city, this maybe possible due to the availability of free WiFi. However, this is not the case in many cities and countries. In addition, the world is now moving towards 5G cells as oppose to WiFi, so I am not sure about the viability of the proposed system,  especially in the near future once 5G becomes more dominant.

3. Please provide a comparison table to compare your proposed method with the literature.

4. The presented algorithms need to e presented in a more professional way, using more advanced templates and formatting. 

Author Response

Dear editor and reviewer,

Thanks for the opportunity to revise our manuscript; and for the time devoted in the revision in this challenging and increasingly busy times.

We have revised carefully all the reviewers’ comments and they have all been implemented in the revised manuscript.

Find bellow a detail answer to each of the comments.

1. The paper studies an important topic especially with the COVID-19 pandemic.I feel the paper lacks theoretical details and explanation on several aspects of the paper. For example, the authors need to clarify more on how they can predict the distances between the people within a certain area (i.e. it is clear how to count the number of people connected to a certain access point, but it is not clear how to determine the distance between each person, thus knowing the social distance status.

 Thank you very much for your comments about lack of the paper. We are improved the new version with some additional explanation.

The proposed method does not require to determine the distance between each person, just the density of people in a calibrated area of study. Therefore, the number of people detected by the access points are correlated with the geographical position estimated by the SmartWifi infrastructure. The Wi-Fi networks with device geolocation capacities associate each device with an estimated location data. Using this estimated location data it is possible to estimate the density of devices in an physical area and then obtain the level of service using the equation 2 as it is explained in section 4.3.

The number of devices detected in a zone is the density of devices.

We are included these details in the new version of paper in section 5.3.

2. The other problem I can see if the fact that the author assume that people are connecting to the mall or area WiFi AP. However, in the demonstrated city, this maybe possible due to the availability of free WiFi. However, this is not the case in many cities and countries. In addition, the world is now moving towards 5G cells as oppose to WiFi, so I am not sure about the viability of the proposed system, especially in the near future once 5G becomes more dominant.

The authors agree in part with the concern expressed by the reviewer on this point. The 5G technology is based on a licensed band and this makes it difficult for cities to deploy agile and inexpensive monitoring technologies such as the one proposed based on public Wi-Fi networks.

On the other hand, even if the device is connected to 5G technology, if it has an open Wi-Fi transmitter, it can be detected by the Wi-Fi network even if it is not connected to it. Thus, the two technologies can coexist in cities and as long as mobile devices have Wi-Fi transmitters, the above Wi-Fi based methodologies can be used even if the near future is dominated by technologies based on 5G or 6G.

3. Please provide a comparison table to compare your proposed method with the literature.

Thanks for the suggestion but we do not understand the need for this table when the technologies available in the literature for citizen mobility monitoring have such different characteristics.

In this sense, the introduction and the literature review have been modified in the new version of paper to provide a better understanding of the existing alternatives for monitoring pedestrian mobility in urban areas.

4. The presented algorithms need to be presented in a more professional way, using more advanced templates and formatting.

Thanks for the suggestion but the authors think that the use of pseudo-code for the presentation of complex algorithms allows a better understanding of the relationship between variables and sorting conditions.

In any case we are willing to adjust the format to the editorial indications of the journal if so considered.

Reviewer 3 Report

The authors focused on modeling and monitoring real-time mobility and congestion. The proposed methodology combines a detailed geographic analysis and high-frequency indicators generated from network data. But there are still some contents, which need be revised in order to meet the requirements of publish. A number of concerns listed as follows:

(1)   The abstract should be narrow down on the problem and highlight the need of the proposed work with experimental results

(2)   Please highlight your contributions in introduction.

(3)   The literature review is poor in this paper. You must review all significant similar works that have been done. For example, 10.3390/agriculture12060793; 10.1109/JSTARS.2021.3059451; 10.1007/s10489-022-03719-6; 10.1016/j.isatra.2021.07.017 and so on.

(4)   All abbreviations need to be written in full for the first time, such as In Line 69, GPS, Line 73, SSID, Line 123,  GSM, Line 156, MAC, ……

(5)   Please provide a flow of the proposed urban monitoring methodologies in Section 4.

(6)   Compared with the existing methods, the innovation of the proposed method needs more detailed description.

(7)   How about the computation complexity of the proposed method?

(8)   At Line 617, add the sections of the “Institutional Review Board Statement”, “Informed Consent Statement”, “Data Availability Statement”.

(9)   Result and discussion should be rewritten to summarize the significance of the work.

(10) The paper is in need of revision in terms of eliminating grammatical errors, and improving clarity and readability.

Author Response

The authors focused on modeling and monitoring real-time mobility and congestion. The proposed methodology combines a detailed geographic analysis and high-frequency indicators generated from network data. But there are still some contents, which need be revised in order to meet the requirements of publish. A number of concerns listed as follows:

(1)   The abstract should be narrow down on the problem and highlight the need of the proposed work with experimental results

Thank you very much for your comment about the abstract. The text has been reviewed to highlight the applications for pedestrian mobility patterns estimation in real time.

(2)   Please highlight your contributions in introduction.

Thank you very much for your comment. The introduction section has been modified including a list of main contribution of the paper.

(3)   The literature review is poor in this paper. You must review all significant similar works that have been done. For example, 10.3390/agriculture12060793; 10.1109/JSTARS.2021.3059451; 10.1007/s10489-022-03719-6; 10.1016/j.isatra.2021.07.017 and so on.

Thank you for the similar works suggestions. The paper is focused on pedestrian mobility, so we have centered the literature review considering this main feature.

The reviewer suggests improving the literature review of the paper with a wider vision with some examples that have been evaluated and referenced in the introduction section.

Thank you again to the reviewer for the reference’s suggestion. Some new research ideas have appeared after the review of the suggested papers, so we will consider these points of view in new works.

(4)   All abbreviations need to be written in full for the first time, such as In Line 69, GPS, Line 73, SSID, Line 123,  GSM, Line 156, MAC, ……

Thank you to reviewer for this suggestion. The paper has been reviewed and all abbreviations have been written in full for the first time.

(5)   Please provide a flow of the proposed urban monitoring methodologies in Section 4.

The Section 4 has been modified to provide a global view of the urban monitoring methodologies.

(6)   Compared with the existing methods, the innovation of the proposed method needs more detailed description.

The paper has been modified in the introduction section to include more detailed description of the innovation of the proposed methods.

(7)   How about the computation complexity of the proposed method?

Section 3.1 describes the software and hardware requirements to implement the methodologies. The Figure 1 shows the infrastructure deployed in Palma based on a produced and consumer strategy. All software is based on commercial and open-source servers.

The implementation of the proposed algorithms has been done using Python, but they can be implemented in any language able to run database commands.

(8)   At Line 617, add the sections of the “Institutional Review Board Statement”, “Informed Consent Statement”, “Data Availability Statement”.

These sections are included in the new version of the paper.

(9)   Result and discussion should be rewritten to summarize the significance of the work.

Thanks for your comment. Following your suggestion we have incorporated a paragraph at the end of section 5 that details the significance, lessons and applicability of the proposed methodology.

(10) The paper is in need of revision in terms of eliminating grammatical errors, and improving clarity and readability.

The paper has been reviewed trying to eliminate grammatical errors. Thanks again to the reviewer for the suggestion.

Round 2

Reviewer 2 Report

I feel the authors did not address all my comment properly. For example, I asked for a table of comparison between the  literature work to show their contributions but they did not do it. I also mentioned the fact that counting the numbers only may not be enough to determine if the people have enough social distances between each other, since you may have  a concentration of people in part of the place, and then the overall density (according to the concept of counting the number will be low).

In addition, not everyone connect to the public WiFi, if they are already connected to the 4G network for example, they will not be counted and will not be considered.

Still I have doubts about the feasibility and novality about the proposed framework.  

I also asked to improve the algorithm presentations and formatting and this did not happen as well.

Author Response

Answers to Reviewer

I feel the authors did not address all my comment properly. For example, I asked for a table of comparison between the literature work to show their contributions, but they did not do it.

Dear reviewer, following your comment, we have included a comparison table among some previous literature centered on pedestrian flow monitoring. The table has been included in introduction section and numbered as 1.

Additionally, before the table we have included the detail explanation of the main contributions of this paper. In this sense, one of the main contributions is the application of WIFI-based technologies on a high volume of users in outdoor spaces, at city level. The paper not only focus on counting devices, but also estimate the density and the level of service useful for urban management decisions.

We hope this time, the added table and the new text in the paper will address your comment properly.

I also mentioned the fact that counting the numbers only may not be enough to determine if the people have enough social distances between each other, since you may have a concentration of people in part of the place, and then the overall density (according to the concept of counting the number will be low).

Effectively, the Wi-Fi network only detects the number of devices, some additional information from each device, and their estimated geolocation. We use these estimated geolocation data to create a subset of devices. Afterwards it is possible to determine the number of devices in a previous defined area. In our work, we have used “calibrated areas” where the walkable space and the people-mobile factor is estimated accurately as it is explained in section 4.3.

Therefore, this work estimates the number of devices present in the calibrated area. And then we measure the level of social distance based on the thresholds determined in table 3. The equation 2 is used to obtain the social distance level from device counting and considering the walkability details of each area.

The work uses only 4 calibrated areas (PA1, PA2, PB1, PB2) along city as a proof of concept, and concentrations outside these areas are not considered. Only, in a full deployment of the solution, all WIFI coverage zones should be divided in calibrated areas to evaluate the social distances. The proposed methodology requires including geographical and urban information to estimate the walkability of each area and the use of filtered WIFI data to derive the social distance level in each area.

Regarding your specific comment, of course, that is the average use of the calibrated areas, and even if there are very few people they might still not fulfil the social distance if they gather all together. However, the point is to detect episodes of high urban concentration, which can be detected by a real time method which monitors a huge space. In this sense, if our proposed method identify a Level of Social Distance beyond the specified thresholds, that indicate an episode of risky overcrowding.

In this sense, we have added a new paragraph in section 5.3 explaining that specific, unusual, possibility.

In addition, not everyone connects to the public WiFi, if they are already connected to the 4G network for example, they will not be counted and will not be considered.

Regarding this point, in section 3 we have emphasized, again, that even if the device is connected to 4G or 5G networks it will be detected if the device has the WIFI transmitter on-line. If the transmitter is running, the mobile will continuously send probe requests at different intervals as shown table 2.

Still I have doubts about the feasibility and novality about the proposed framework.

As we mention in the paper in section 2: As far as the authors are aware, there are few real deployments that propose feasible methodologies to transform automatic Wi-Fi-based device detection data in extended urban areas into useful and interpretable indicators that can be incorporated in urban planning and management systems or provide real-time warnings in smart city decision support systems to prevent insecurity or unhealthy congestion. Regarding the feasibility, well, we would like to remind that this is a real application that has already been implemented, and proven useful.

I also asked to improve the algorithm presentations and formatting, and this did not happen as well.

Following your comment, we have improved the formatting and presentation of the algorithms following the reference [2] as example and the guidelines for author of MDPI.

Reviewer 3 Report

I have appreciated the deep revision of the contents and the present form of this manuscript. All my previous concerns have been accurately addressed. I think that this paper can be accepted.

Author Response

Thank you very much for yours comments to improve the final version of the paper.

Round 3

Reviewer 2 Report

The authors addressed my comments.