Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. Clarify OS1–OS8 at first mention

2. I understand that the primary scope of this manuscript is to evaluate the STM32L4 low-power configurations. However, the intended application is wPUM puff-topography monitoring. I therefore recommend briefly introducing the monitoring system in Section 2, using both text and a diagram, to help readers understand the experimental setup.

3. In Table 1, mA is a unit of current, not power. Please also explain how these values were obtained.

4. Sections 2.2 and 2.3 should be reorganized to reduce repetition.

5. Define the required monitoring duration. Without an application-specific target, the practical value of approximately 30 days of runtime is unclear. Explain whether this meets the intended study requirements, how actual puff monitoring would affect runtime, and whether a larger-capacity battery would be feasible within size and weight constraints.

6. Explain the STOP2 failures. Three failures among five devices materially affect the recommendation, yet the cause remains unresolved. Provide diagnostic evidence, distinguish wake-up failure from failure after waking, and clarify whether these trials were included in the analysis despite the stated exclusion criteria. Do not generalize an unresolved implementation failure into a limitation of STOP2 itself.

Comments on the Quality of English Language

The presentation is somewhat informal.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper evaluates four low‑power modes (Sleep, STOP0, STOP1, STOP2) on the STM32L433 microcontroller for a wearable sensor. The experiments are straightforward, and the main conclusion, that STOP1 gives the best trade‑off between runtime and wake‑up reliability, is supported by the data. However, the manuscript needs significant work before it can be published. Some statistical analysis is thin, and several technical points need deeper discussion. I outline the main issues below.

(1) The firmware uses a fixed 60‑s periodic wake‑up to simulate health‑data logging. This is fine as a benchmark, but the actual device is supposed to wake on puff events (pressure‑sensor interrupts). The manuscript should clearly state that the results apply to a periodic‑logging scenario, not the interrupt‑driven use case. A short discussion of how an event‑driven workload would change the power profile would help. Even a rough estimate based on typical puff frequency would be useful.

(2) Three out of five devices in STOP2 failed to wake after RTC alarms. This is a serious finding, but the paper just reports it as a reliability issue. The authors should try to explain why this happened. Possible causes: RTC clock behaviour in STOP2, backup domain power, or known errata in the STM32L433. If the exact cause isn’t clear, at least discuss the likely candidates based on the reference manual. This would make the paper much more useful to other engineers.

(3) Battery age was not controlled: eight batteries were ~6 months old, two were older, and they were “randomly assigned” to groups. While randomisation is good, the manuscript does not report the exact distribution of battery ages across the four groups. A table or supplementary note should list each device’s battery age and initial voltage/SOC. Additionally, the authors should discuss whether battery ageing could have interacted with the low‑power modes (e.g., higher internal resistance causing brown‑out at different voltages) and thus affected the runtime results.

(4) The Discussion section currently repeats the results without critically examining their impact on field deployment. Add a paragraph on how the 30‑day runtime of STOP1 translates to real‑world ambulatory studies (e.g., compared to the 7‑day typical deployment of previous generations). Also, explicitly list the limitations: (a) testing in a static office environment without temperature extremes, (b) no simulation of actual puff topography, (c) use of a single battery capacity (350 mAh), and (d) firmware that does not implement the final interrupt‑driven sensing. These limitations should be acknowledged as directions for future work.

(5) The text states mean runtimes of 3.27 days (Sleep), 27.78 days (STOP0), 30.15 days (STOP1), and 20.91 days (STOP2). However, the confidence intervals (e.g., ±20.50 days for STOP2) indicate extreme variability that is not visible in the curves of Figure 1. Verify the numerical accuracy of the reported means and CIs. If the STOP2 variability is indeed that high, explain it in relation to the three failed devices.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

All comments resolved. 

Author Response

Thank you for reviewing the revised manuscript and confirming that all previous comments have been satisfactorily addressed. We appreciate your constructive feedback throughout the review process.

Although no further revisions were required in response to your comments, we have made additional revisions at the editor's request to expand the reference list and reduce the self-citation rate. We have also incorporated additional sentences to provide context for the newly added citations. These changes strengthen the supporting literature without altering the experimental results or conclusions.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript can be accepted in present form.

Author Response

Thank you for your positive evaluation and recommendation for acceptance. We appreciate your time and constructive feedback, which helped improve our manuscript.

No further revisions were required in response to your comments. However, at the editor's request, we have expanded the reference list to include additional relevant publications from independent research groups. We also added selected sentences to clarify how these references relate to our study. These changes provide broader scientific context while preserving the experimental findings and conclusions.