Open Source System for Monitoring Wireless Outdoor Networks in Mining

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper describes an open-source hardware platform for monitoring wireless outdoor networks in mining scenarios. The subject is interesting, and the authors propose an open-source and affordable hardware solution for a large-scale wireless sensor network, including monitoring. The system, as proposed, is inexpensive. At the same time, the components are commercially available (e.g., Raspberry Pi and external Wi-Fi modules), presenting a valuable contribution to the open hardware and industrial Internet of Things (IoT) domains. Moreover, the authors include a comparison with the commercial Ekahau Site Survey tool.
However, many issues need to be addressed to improve clarity, reproducibility, and the scientific validity of the work:

1. Introduction and References
   The introduction provides a comprehensive overview of wireless networks in mining, although it could be enhanced by clearly articulating the specific contributions and novelty of the described platform You should take into consideration all other existing tools or open-source monitoring solutions. It would be helpful to include benchmarking references or a quick comparison with similar systems (e.g., different 802.11 network monitoring solutions). You can add a comparison table.

2. Research Design and Methodology
   The hardware and software modules are well-described, but the basic technical choices (e.g., the choice of Kriging as the interpolation method) must be better justified. Additionally, an explanation regarding sensor calibration, potential measurement errors, and limits to the reliability of the data would strengthen the methodology.

3. Experimental Validation
   The validation experiments are well designed, but quantitative analysis could be improved by adding statistics such as standard deviations or error bars in Figures 7-10. A brief consideration of environmental factors (e.g., terrain or weather) during in-loco tests would also be useful.

4. Figures and Tables
   Some figures (e.g., Figures 1, 3 to 10) would benefit from higher resolution and better contrast to be more readable. Figure captions could be improved, providing clarity so that they are self-explanatory. Moreover some tables are wrong numbered (e.g., Table 2.1).

5. Discussion and Results Interpretation
   Though the comparison to the Ekahau Site Survey is clearly a strength of your study, I would suggest explaining the differences between "Satisfactory with restrictions" and "Unsatisfactory" more clearly. You should evaluate the practical implications of those differences and their impact on the required actions for network deployment remediation.

6. Conclusions
   The conclusions are good, but considering demonstrating the broader application of the platform beyond mining (e.g., application in ports, railroads, or other outdoor industrial contexts).Additionally, a brief statement regarding future enhanced capabilities would improve the overall impact, such as low-power features or more integrated modules.

7. English Language and Style
   The writing is clear, but could be polished to take care of minor language issues to limit awkward expressions, and therefore increase clarity and readability. Some of your sentences are too long and repetitive (e.g., multiple recaps on the advantages of using Raspberry Pi).

Comments on the Quality of English Language

The English is understandable, but minor language polishing would improve the readability and flow. There are some small, awkward phrases and some rather long sentences that could be rephrased for clarity. I strongly recommend a proper proofread or language check to ensure the entire manuscript has concise and easy-to-read phrasing.

Author Response

Reviewer #1

The paper describes an open-source hardware platform for monitoring wireless outdoor networks in mining scenarios. The subject is interesting, and the authors propose an open-source and affordable hardware solution for a large-scale wireless sensor network, including monitoring. The system, as proposed, is inexpensive. At the same time, the components are commercially available (e.g., Raspberry Pi and external Wi-Fi modules), presenting a valuable contribution to the open hardware and industrial Internet of Things (IoT) domains. Moreover, the authors include a comparison with the commercial Ekahau Site Survey tool.
However, many issues need to be addressed to improve clarity, reproducibility, and the scientific validity of the work:

1. Introduction and References
   The introduction provides a comprehensive overview of wireless networks in mining, although it could be enhanced by clearly articulating the specific contributions and novelty of the described platform. You should take into consideration all other existing tools or open-source monitoring solutions. It would be helpful to include benchmarking references or a quick comparison with similar systems (e.g., different 802.11 network monitoring solutions). You can add a comparison table.

Response: Thank you for your review and insightful comments. To address the comment, we have updated the Introduction Section, inserting related works to position our system against widely used open-source 802.11 monitoring tools, and inserted Table 1 comparing similar systems.

Section I update

“Several open-source tools exist for Wi-Fi monitoring, including MonFi [15], IAX [16], ORCA [17], and the Open Source Capture and Analysis Tool [18]. While these solutions demonstrate strong research contributions—ranging from high-rate programmable monitoring (MonFi), fine-grained CSI extraction (IAX), and software-defined control (ORCA), to frame-level security analysis and indoor occupancy estimation—they remain limited in scope for harsh industrial applications. In contrast, our system is designed for field-ready, outdoor, and industrial environments, combining performance metrics (RSSI, latency, packet loss, throughput), GPS-based geolocation, and Kriging-based interpolation to generate actionable REMs. Moreover, unlike many tools focused on research or indoor applications, our platform prioritizes autonomy, safety, and cost-effective scalability, validated against a commercial benchmark, the Ekahau Site Survey. Table 1 summarizes other open source tools with our work. ”

 

System	Primary Focus
MonFi	High-rate programmable Wi-Fi monitoring
IAX	CSI extraction for Intel NICs (802.11a/g/n/ac/ax)
ORCA Toolchain	Kernel–user space API, SDN-based control
Open Source Capture Tool	802.11 management frame security analysis
Our System	Outdoor industrial monitoring with geolocation, performance KPIs (RSSI, latency, throughput, packet loss), and REM generation via Kriging

2. Research Design and Methodology
   The hardware and software modules are well-described, but the basic technical choices (e.g., the choice of Kriging as the interpolation method) must be better justified. Additionally, an explanation regarding sensor calibration, potential measurement errors, and limits to the reliability of the data would strengthen the methodology.

Response: We appreciate the reviewer’s valuable feedback. In the revised manuscript, we have improved our Introduction section, explaining the choice of using Kriging as the interpolation method. 

Section I update

“[...] We adopt Kriging interpolation because it provides the best linear unbiased prediction by explicitly modeling spatial autocorrelation through the variogram, minimizing prediction error variance compared to deterministic methods such as inverse distance weighting [ 15, 16 ]. Unlike other alternatives, Kriging also quantifies prediction uncertainty, offering variances that help assess the reliability of interpolated values [16 ]. Its flexibility to handle irregular data distributions and proven superior performance in geosciences, environmental monitoring, and mining applications make it particularly well-suited for our study [17,18].”

3. Experimental Validation
   The validation experiments are well designed, but quantitative analysis could be improved by adding statistics such as standard deviations or error bars in Figures 7-10. A brief consideration of environmental factors (e.g., terrain or weather) during in-loco tests would also be useful.

Response: We thank the reviewer for recognizing that our experiments were well designed. In the revised manuscript, we have enhanced Figures 7–10 and added a qualitative analysis. The in-loco tests were conducted directly in the operational mining yard, with uneven terrain, ore piles, and large machinery that can obstruct line-of-sight propagation. The tests took place under regular weather conditions (no rain), minimizing atmospheric effects; thus, the dominant environmental influences were related to the mining terrain and equipment. We will clarify this in the manuscript.

Section 5.2. update

“The field campaign took place in the operational mining yard on a dry, rain-free day, which is typical weather for the S11D complex. Terrain irregularities, ore piles, and the presence of large machinery were the primary environmental factors affecting propagation and accessibility to points of interest.”

4. Figures and Tables
   Some figures (e.g., Figures 1, 3 to 10) would benefit from higher resolution and better contrast to be more readable. Figure captions could be improved, providing clarity so that they are self-explanatory. Moreover some tables are wrong numbered (e.g., Table 2.1).

Response: . In the revised manuscript, we have: (i) replaced Figures 1 and 3–10 with higher-resolution versions. (ii) revised figure captions to ensure they are self-contained and explanatory, and (iii) corrected table numbering.

5. Discussion and Results Interpretation
   Though the comparison to the Ekahau Site Survey is clearly a strength of your study, I would suggest explaining the differences between "Satisfactory with restrictions" and "Unsatisfactory" more clearly. You should evaluate the practical implications of those differences and their impact on the required actions for network deployment remediation.

Response: We thank the reviewer for highlighting this important clarification. In the revised manuscript, we have updated Section  5.2.

Section 5.2. update

“We classified the monitored points of interest into the following categories:

•  Not observed: when there was no measurement of the point;
• Does not meet (Unsatisfactory): when two or more parameters are outside the minimum requirements. In this situation, the network is unable to guarantee the safe or stable execution of control commands, and corrective action such as the addition of new access points or a redesign of the coverage area is required.
• Complies with restrictions (Satisfactory with restrictions): when exactly one parameter is outside the minimum requirements. In this case, the application may still function, but with degraded performance. This classification indicates the need for optimization measures such as antenna realignment, transmit power adjustment, or selective reinforcement of coverage.
•  Complies: when all parameters are within the minimum requirements.”

 

6. Conclusions
   The conclusions are good, but considering demonstrating the broader application of the platform beyond mining (e.g., application in ports, railroads, or other outdoor industrial contexts).Additionally, a brief statement regarding future enhanced capabilities would improve the overall impact, such as low-power features or more integrated modules.

Response:  We would like to thank the reviewer for his complimentary comments. In the revised manuscript we have: expanded the Conclusions to highlight the potential of our platform in other industrial domains, including ports, rail yards, and large-scale outdoor facilities, where wireless coverage is equally mission-critical and infrastructure presents similar constraints; and added a forward-looking statement on possible enhancements, including low-power operation for battery-constrained deployments, tighter integration with IoT modules, and improved modularity for rapid adaptation to new environments.

Section 6 update

“It is also important to emphasize that the platform is not restricted only to mining and processing areas but can be applied in other industrial domains such as ports, railroads, and large-scale outdoor facilities, where wireless coverage is equally mission-critical and infrastructure presents similar constraints.

Finally, future enhancements will focus on low-power operation for battery-constrained deployments, tighter integration with IoT and sensing modules, and improved modularity to allow rapid adaptation to diverse industrial scenarios. These developments will broaden the applicability and long-term impact of the platform.”

7. English Language and Style
   The writing is clear, but could be polished to take care of minor language issues to limit awkward expressions, and therefore increase clarity and readability. Some of your sentences are too long and repetitive (e.g., multiple recaps on the advantages of using Raspberry Pi).

Response: We thank the reviewer for this feedback. In the revised manuscript, we improved the language by enhancing clarity, eliminating awkward phrasing, breaking up long sentences into concise, direct statements, and reducing repetition.

Reviewer 2 Report

Comments and Suggestions for Authors

Comments to Authors

1. Explain in detail the reason of choosing Kriging over other interpolation techniques like (e.g., IDW, spline) for generating the RSSI map, what are it specific advantages.
2. What are precise hardware and software components that are used to make up the proposed sensing platform, and how are they integrated into mobile machines, if some app. can also be developed from the integration?
3. How can we finalize that the accuracy and calibration obtained from the RSSI measurements across different mobile machines and environmental conditions are correct?
4. How can we say that the proposed platform performs "similarly" as "Ekahau" system and what are the metrics or benchmarks that can be used in this comparison?
5. Is due to extreme terrain or signal interference any discrepancies are reported between the platform and Ekahau in specific scenarios?
6. How does the platform handle network issue and data collection things in areas which are exposed to the complete signal loss or have very low RSSI?
7. What are the main safety and operational concerns when installing this platform on mining machinery?
8. As a future perspective explain the possibilities when this platform be extended to support real-time network optimization or predictive maintenance issues based on real time network data?

Minor Comments

Increase the font size of the Figure 1, x-and y- labels of the Figure 7, 8, 9, 10,

Author Response

Reviewer #2

 

1. Explain in detail the reason of choosing Kriging over other interpolation techniques like (e.g., IDW, spline) for generating the RSSI map, what are it specific advantages.

Response: We would like to thank you for your revision and insightful feedback.. In the revised manuscript, we have improved the Introduction section to better explain why we chose Kriging as our interpolation method. 

Section I update

“[...] We adopt Kriging interpolation because it provides the best linear unbiased prediction by explicitly modeling spatial autocorrelation through the variogram, minimizing prediction error variance compared to deterministic methods such as inverse distance weighting [ 15, 16 ]. Unlike other alternatives, Kriging also quantifies prediction uncertainty, offering variances that help assess the reliability of interpolated values [16 ]. Its flexibility to handle irregular data distributions and proven superior performance in geosciences, environmental monitoring, and mining applications make it particularly well-suited for our study [17,18].”

 

1. What are precise hardware and software components that are used to make up the proposed sensing platform, and how are they integrated into mobile machines, if some app. can also be developed from the integration?

Response: We thank the reviewer for this question. The sensing platform is composed of cost-effective, widely available hardware:

• Raspberry Pi (single-board computer) – the central unit responsible for running the monitoring software, managing data collection, and coordinating with external peripherals.
• GPS antenna/module – provides precise geolocation for each network measurement, enabling the construction of geo-referenced Radio Environment Maps (REMs).
• Wireless network antenna (802.11b/g/n compatible) – performs active measurements of signal strength, latency, packet loss, and throughput.
• Electrical connection cables and power supply – allow stable integration with mobile machines in the mining yard.

On the software side, the platform leverages open-source Linux-based tools, Python scripts, and drivers that enable active probing of wireless links. The software automates data acquisition, stores the measurements, and applies Kriging interpolation to generate continuous REMs of the yard.

With regard to the integration into mobile machines, the compact hardware setup allows installation in stackers, reclaimers, and other mining vehicles without interfering with their operations. Power is supplied from the machine’s existing electrical system, while data collection is automated to minimize human intervention.

To address this issue, we have updated Sections 2 and 5.6 to elucidate it.

Section 2 update:

“The software developed in Python for the modules has four states of operation: initial, search, collection, and final. The Raspberry Pi serves as the central unit, coordinating the GPS receiver and Wi-Fi antenna via USB interfaces and Linux drivers. The system operates in unattended mode, automatically collecting data during machine operation without interfering with normal industrial processes.

The monitoring scripts automate data collection, geo-reference each measurement with GPS coordinates, and store results for interpolation. Given this modular architecture, the system can be extended with a mobile or web application to visualize radio environment maps in real time or to trigger alerts when thresholds are exceeded, supporting proactive network management.

Section 5.6 update:

“The prototype is secured to a handrail using nylon ties and connected to a power source on a nearby distribution board. Given that the prototype weighs approximately 300 grams, a more robust installation was deemed unnecessary. Because of its compact design and seamless integration with mobile equipment, the platform can continuously operate in the yard. Its outputs can also feed higher-level applications such as dashboards or maintenance alert systems, thereby improving the decision-making process for network deployment and remediation.”

1. How can we finalize that the accuracy and calibration obtained from the RSSI measurements across different mobile machines and environmental conditions are correct?

 

As shown in Section 5.1, we conducted controlled laboratory tests where both systems measured the same signal under identical conditions with varying transmission powers and antenna orientations. The error bar analysis (Fig. 8) demonstrates that all 16 sample sets show intersecting measurement ranges between our prototype and the Ekahau system.

The quantitative analysis we've added in Section 5.5.1 shows a moderate to strong positive correlation (r=0.58) between measurements in the mW domain, indicating consistent relative patterns across different regions despite using different hardware.

1. How can we say that the proposed platform performs "similarly" as "Ekahau" system and what are the metrics or benchmarks that can be used in this comparison?

We calculated a Pearson correlation coefficient of 0.58 in the mW domain between the measurements of both systems, indicating a moderate to strong positive correlation.

Both systems identified the same regions as having the strongest signals (Regions 1, 4, and 7) and weakest signals (Region 2), which is critical for network planning decisions.

As shown in Figure 9, both systems produced maps with seven identifiable regions showing similar signal strength patterns, which would lead network specialists to make similar network improvement decisions.

1. Is due to extreme terrain or signal interference any discrepancies are reported between the platform and Ekahau in specific scenarios?

The discrepancies observed between our platform and Ekahau are not primarily attributable to terrain or interference effects, as both systems were exposed to identical environmental conditions during our sequential testing at each measurement point. Instead, differences likely stem from:

1. Different Processing Algorithms: Our prototype uses documented Ordinary Kriging interpolation as described in Section 2, while Ekahau employs proprietary algorithms that are not publicly disclosed.
2. Hardware Differences: The systems use different radio receivers and antennas (as detailed in Section 2.1), which can result in different raw signal measurements even under identical conditions.
3. Sampling Methodology: As explained in Section 5.2, Ekahau uses interpolation to classify points it cannot physically access, while our system collects actual measurements at those locations.

Importantly, our field tests in the challenging terrain of the S11D complex (Section 5.2) demonstrated that our platform can collect data in areas where Ekahau cannot be safely deployed. This addresses one of the key limitations mentioned in the introduction—that commercial tools often cannot gather data from all points of interest due to harsh terrain and dangerous conditions, leading to interpolated maps that may not accurately represent the network's overall quality.

Our added quantitative analysis confirms that despite these processing differences, both systems produce sufficiently similar results for practical network planning purposes.

 

1. How does the platform handle network issue and data collection things in areas which are exposed to the complete signal loss or have very low RSSI?

Response: We would like to thank the reviewer for this question. We have performed the following update in Section 5:

“The system is unable to recover network parameters in the absence of a radio signal. Under conditions of high packet loss, specifically at 100% loss, the IPERF tool fails to report RTT and throughput values. Nonetheless, it remains possible to obtain the RSSI as this parameter does not depend on the successful transmission of data packets.”

1. What are the main safety and operational concerns when installing this platform on mining machinery?

Response: We thank the reviewer for the feedback on this matter. We have updated the manuscript to detail the main safety and operational concerns.

Section 5.6  update:

“When installing the platform on mining machinery, several safety and operational considerations must be taken into account. The device is lightweight, but it must be securely fixed to withstand vibrations. This prevents interference with machine operation. Electrical protection is ensured by a DC-DC regulator with a fuse. This setup prevents any connection to safety-critical circuits. The hardware must be housed in an enclosure due to harsh outdoor conditions. This enclosure shields against dust, rain, and mechanical shocks. Proper antenna positioning is essential. Good placement minimizes obstructions from metallic structures or ore piles that could degrade signal quality. The GPS accuracy margin is about 2.5 meters. This margin must be addressed, especially in areas with multipath effects or partial blockages. Automation features, such as auto-start configuration and remote operation, are also crucial. They reduce the need for human intervention in hazardous areas, increasing both safety and efficiency.”

1. As a future perspective explain the possibilities when this platform be extended to support real-time network optimization or predictive maintenance issues based on real time network data?

Response: We thank the reviewer for the feedback on this matter. We have updated the manuscript to detail the potential usage scenarios for the platform.

Section 6 update:

“The implementation of the platform at multiple locations within an operating area would ensure the generation of more comprehensive REMs. These REMs could be automatically produced to facilitate near real-time monitoring of the wireless network's health, enabling optimization and adjustments to maintain network integrity. Additionally, the system's remote operation feature allows for dynamic changes to monitoring coordinates, ensuring expanded and adaptive coverage.”

Minor Comments

Increase the font size of the Figure 1, x-and y- labels of the Figure 7, 8, 9, 10,

Response: We appreciate the suggestion regarding the quality of the figures. The figures have been updated to improve their interpretability.

 

 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Dear authors,

The 2nd version of your paper has improved compared to your 1st version. Nearly all of the 1st round comments have been addressed. The introduction has been better described relative to the relevant open source Wi-Fi monitoring tools (e.g., MonFi, IAX, ORCA, etc.), while "Table 1" provides comparative information on these. The observations regarding the "Kriging interpolation" are well described as expected, and the methodology section is more technically informative. As can be seen, the experimental section has improved since you have added standard deviations, headings for the error bars, and also a discussion of environmental conditions. Manuscripts' "figures" and "tables" have also been improved with respect to numbering and labelling. Also, the criteria for "Compliant", "Satisfactory with restrictions", and "Unsatisfactory" are better defined. The conclusion section includes new applications outside of mining (ports, railways, and other industrial outdoor situations) and discusses other possibilities, i.e., "Low power operation", "better integration" of IoT technology.

I believe that you need to make only some minor editorial specifications before the final acceptance of your paper:

First of all, please change all remaining placeholders for citations (e.g., "[15?]"), verifying citation number (line 68).

Revise Table 7 column headings if possible and give also the power conversion formula (dBm ↔ mW) (line 393).

Please clarify the description for RTT and iperf (i.e., ping measures RTT, iperf measures throughput to jitter) (line 292 you mention that "the IPERF tool fails to report 292
RTT and throughput values")

Please replace "X, Y, Z" coordinate terms (Figure 1) with "Latitude, Longitude, Altitude", and specify the reference system (i.e., WGS84).

Lastly, please provide your software dependencies and OS/kernel versions.

Author Response

Dear authors,

The 2nd version of your paper has improved compared to your 1st version. Nearly all of the 1st round comments have been addressed. The introduction has been better described relative to the relevant open source Wi-Fi monitoring tools (e.g., MonFi, IAX, ORCA, etc.), while "Table 1" provides comparative information on these. The observations regarding the "Kriging interpolation" are well described as expected, and the methodology section is more technically informative. As can be seen, the experimental section has improved since you have added standard deviations, headings for the error bars, and also a discussion of environmental conditions. Manuscripts' "figures" and "tables" have also been improved with respect to numbering and labelling. Also, the criteria for "Compliant", "Satisfactory with restrictions", and "Unsatisfactory" are better defined. The conclusion section includes new applications outside of mining (ports, railways, and other industrial outdoor situations) and discusses other possibilities, i.e., "Low power operation", "better integration" of IoT technology.

I believe that you need to make only some minor editorial specifications before the final acceptance of your paper:

1) First of all, please change all remaining placeholders for citations (e.g., "[15?]"), verifying citation number (line 68)

Response: Our team appreciates the reviewer for acknowledging that our paper can be accepted with minor revisions. We have updated the citations as the reviewer suggested.

2) Revise Table 7 column headings if possible and give also the power conversion formula (dBm ↔ mW) (line 393).

Response: We thank the reviewer for the suggestion. We have upgraded the table and inserted the conversion formula.

3) Please clarify the description for RTT and iperf (i.e., ping measures RTT, iperf measures throughput to jitter) (line 292 you mention that "the IPERF tool fails to report 292 RTT and throughput values")

Response: We appreciate the comment. We clarified that RTT is obtained with ping, while iperf reports throughput and jitter.

4) Please replace "X, Y, Z" coordinate terms (Figure 1) with "Latitude, Longitude, Altitude", and specify the reference system (i.e., WGS84).

Response: We appreciate the suggestion regarding this important detail. We updated the figure and included details of the reference system.

5) Lastly, please provide your software dependencies and OS/kernel versions.

Response: We thank the reviewer for the suggestion. We have added information about the operating system, kernel version, and software dependencies used in our experiments.