Applied Research · Saskatchewan Polytechnic · 2026

Research Results

HIVIS Monitor — Applied Research Student Showcase 2026

Rani dos Santos Vilela · May 4, 2026

Research Questions

RQ1 — Feasibility

Can the system be built?

Is it feasible to develop a low-cost, self-hosted wearable IoT sensor system — battery powered and continuously connected to a real-time server — capable of monitoring indoor and outdoor air quality, temperature, and noise levels in trades workplace environments?

RQ2 — Reliability

Does it perform in the real world?

If feasible, does such a system perform reliably when deployed in real-world conditions?

Fleet Overview — Mar 25 – May 3, 2026

812,717

Valid Readings

742 hrs

Monitoring Hours

Devices Deployed

40 days

Longest Continuous Run

843.6 ppm

Fleet Mean CO₂

110.4

Fleet Mean IAQ

17 of 33

Days Avg CO₂ Above 800 ppm

Mar 25
– May 3

Study Period

RQ1 — Feasibility

Can the System Be Built?

✓ Yes — Feasibility Confirmed

System Built & Deployed

3 physical devices designed, assembled, and deployed. Custom 2-layer PCB fabricated via JLCPCB. Firmware deployed via OTA updates. WiFiManager provisioning required no firmware changes per device.

Data Transmission

812,717 valid readings captured. Fixed device: 62.6% delivery rate over 40 days. Mobile devices operated intermittently by design — low overall delivery rate reflects planned deployment, not system failure.

Battery Performance

Mobile devices sustained 4–5 hours per charge at workplace drain rates (~21.5%/hr). Firmware v2.0.2 introduced screen timeout: −1.64%/hr drain reduction confirmed on fixed device.

Note: One device excluded from all analysis — confirmed BSEC2 calibration failure producing frozen readings. Disclosed as a system limitation.

Workplace Air Quality — Mobile Devices

Trades Workplace Findings

Mobile devices carried during telecom service shifts in Regina, SK · Mon–Thu 16:30–20:30 / Sat 08:00–16:00 CST

Mobile-A 12,395 readings · Apr 11–25

Metric	Mean	Min	Max
IAQ (0–500)	114.6	0.1	500
CO₂ (ppm)	1,083.7	400	4,080
Temperature (°C)*	28.0	20.0	36.3
Humidity (%)	26.2	15.3	51.6
Sound (dB SPL)	82.2	50.0	111.8

* Enclosed device — ~+1.3°C upward bias from ESP32 heat.

CO₂ > 800 ppm: 77.5% CO₂ > 1,000 ppm: 50.0% CO₂ > 1,100 ppm: 34.0% IAQ > 100: 57.8%

Mobile-B 899 readings · Apr 15 (single session)

1,219.3 Mean CO₂ (ppm)

1,085–1,372 CO₂ Range (ppm)

1,372 ppm Peak CO₂ @ 17:15 CST

133.4 Mean IAQ

CO₂ > 1,000 ppm: 100% of readings

Limited dataset — one session only. Results are notable but not generalisable.

1,000 ppm CO₂ has been linked to a moderate decline in decision-making performance.
Satish et al. (2012). Is CO₂ an indoor pollutant? Direct effects of low-to-moderate CO₂ concentrations on human decision-making performance. Environmental Health Perspectives, 120(12), 1671–1677.

CO₂ by Day — Mobile-A

CO₂ by Day of Week

Average CO₂ concentration per working day

Mon

1,199 ppm worst

Wed

864 ppm

Sat

1,033 ppm

Friday excluded — devices left at home, not representative of workplace conditions.

Outdoor Detection

Automatic Outdoor Detection

The system identified outdoor conditions with no configuration required.

127

Readings below 500 ppm during workplace hours

440 ppm

Mean outdoor CO₂ (near-atmospheric)

Outdoor readings clustered at shift transitions:

08:30 Saturday
shift start

15:00 Saturday
shift end

16:30–17:00 Weekday
shift arrival

No GPS, no manual tagging — the sensor responded to real environmental changes.

Educational Deployment — Fixed Device

System Validation

Fixed device · Saskatchewan Polytechnic · 40 days continuous · 693,316 readings

Metric	Mean	Min	Max
IAQ (0–500)	114.1	0.3	500
CO₂ (ppm)	835.8	400	4,228
Temperature (°C)	25.5	—	30.5
Humidity (%)	17.4	10.4	76.0

This device did not include a microphone. Noise data is from mobile devices only.

Peak CO₂: 4,228 ppm on April 5 — IAQ reached sensor ceiling of 500.

Cost Context

Why Cost Matters

HIVIS Monitor

$73 CAD 18650 battery build

$90 CAD — LiPo 2100 mAh build
5-device fleet: ~$365–$450
No calibration required
No subscription
Self-hosted, open firmware

Consumer NDIR CO₂ Monitor

$150–$300 per unit

Fixed location only
Cloud subscription often required
No noise monitoring
No fleet management

Certified 4-Gas Industrial Monitor

$400–$600+ per unit

Annual calibration required
5-device fleet: $2,000–$3,000+
Required for safety-critical detection

The HIVIS Monitor is not a replacement for certified instruments. It is an auxiliary — giving workers and supervisors visibility into conditions that were previously invisible, at a cost that makes fleet deployment realistic.

Limitations

eCO₂ is estimated via MOX sensor (BME688), not measured by NDIR. Values are indicators, not certified measurements.
BME688 cannot detect specific gases such as H₂S or CO. Safety-critical detection requires dedicated electrochemical sensors.
One device excluded due to BSEC2 calibration failure (frozen readings).
Mobile-B has only one valid workplace session (899 readings). Results are notable but limited in scope.
Temperature readings from enclosed mobile devices carry ~+1.3°C upward bias due to ESP32 heat.
Battery life ~4–5 hours per charge. Mobile-B depleted to 0% twice during the study.
Real-time alerts require active Wi-Fi. No alerts during offline periods.

Research Paper

Full Report

Full applied research report — HIVIS Monitor, Applied Research Student Showcase 2026, Saskatchewan Polytechnic. Includes full methodology, raw data analysis, firmware development log, and complete findings.

Download Full Report (PDF)