Large-Scale Dataset for Radio Frequency-Based Device-Free Crowd Estimation
Round 1
Reviewer 1 Report
In my opinion, there is a lack of experiment in details in one place in the one section.
People in the experiment had some API on their mobile phone to be noticed in the crowd or in other way you will be have data?
What about question with sensitive data of people?
In title Authors wrote about estimation, but where are presented estimation method in paper?
The section named Conclusions would be really valuable at the end of this paper. In addition, in the paper is lack of appropriate order.
In related works review presented in the first section it should be mention paper also in similar approach about estimation crowd on based WiFi connection in the building, please see link to this paper: https://www.mdpi.com/1424-8220/19/19/4211
Moreover, there is some typos and misspelling.
Please, find more comments in attached paper.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
The article presents a large dataset (14M traces) consisting of RSSI (radio-frequency power at the receiving node) measurements at the 433 MHz and 868 MHz bands, which have been collected during 2 consecutive years (2017 and 2018) at the Tomorrowland music festival in Belgium.
The dataset is then used to estimate the size of the crowd present at the music festival by measuring the relative RSS values caused by RF attenuation. The results presented as part of the study show that there is a positive correlation between the RSS values and the real number of people present in each location, which is validated using other measurements (i.e., scan system at counters, payment information in bars, etc.).
The dataset is freely accessible, the data is well labeled/organized and some scripts are provided, easing the process of using the dataset to other researchers. In addition, the paper is well written, providing detailed explanations of the setup and the measuring conditions.
Overall, I believe that the dataset and the provided tools can be useful to other researchers willing to develop and validate their crowd estimation algorithms, which is a topic that is expected to become interesting in the following years.
Some minor suggestions to improve the paper:
1) Include the position of the controller in Figure 4
2) Include the 433 MHz version of Figure 5
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Author’s responses for comments are satisfactory.
The paper is ready for publishing in the current form.
