Testing

In addition to tests performed in our video demo, below are graphs of various tests we did on our system to assess its accuracy and usefulness. Beneath those graphs are short explanations of the data. Additionally, there are some photos taken during each test. Further down the page, the algorithm that we use to generate the overall environmental score is explained.

The first graph contains the reference (thermostat) and the measured (from our monitor) temperature and humidity. As shown, the measured temperature starts slightly lower than the reference, but then stabilizes and remains quite close to the reference temperature. The measured humidity is slightly lower than the reference, but it is still within margin of error.

The second graph shows the gradual return of CO2 levels to normal after a brief burst of high CO2. As shown, the initial CO2 level is around 650 ppm, which is considered normal indoors. After a burst of high CO2, the levels rise to around 2500 ppm, but then gradually return to normal levels within a few minutes. This is a synthetic test, but it demonstrates the system's ability to measure CO2 accurately.

The third graph displays the reference (iPhone) and measured (from our monitor) sound intensity in dB. For lower decibel levels, the reference aligns closely with the measured sound intensity, but our monitor tends to overestimate the sound intensity at higher levels >60 dB. However, this may have been due to the exact position of the computer (noise source) relative to the phone and our monitor. When directly above both the phone and our monitor, we were able to measure 70 dB on both devices (shown in manual decibel test image).

Below are some images taken during the tests.

Score Algorithm

Our monitor uses a simple but effective algorithm to generate an overall comfort score based on AQI (PM2.5), CO2, noise, temperature, and humidity. This score is designed to provide a safety oriented measure of the environment (for example, a harmful CO2 level >2000 ppm would result in a score of 0).

There is a separate value that determines specifically whether the conditions are good for study (displayed on a different screen). This value (which is boolean/good or bad) is determined with similar thresholds as below (with below approx. the lowest threshold being good), except the noise level is required to be below 45 dB.

When creating the algorithm, we used multiple sources to help us determine the various thresholds for each measurement (sources are in references). There is not an overall AQI-like measurement that includes noise (and the other measurements), so we needed to start from scratch using our sources as a guideline.

The comfort score, ranging from 0 to 100, is calculated by starting at 100 and applying penalties based on several environmental measurements:

• PM2.5:
  • ≤15 µg/m³: No penalty.
  • >15 to ≤35 µg/m³: 5 points deduction.
  • >35 to ≤75 µg/m³: 15 points deduction.
  • >75 to ≤150 µg/m³: 70 points deduction.
  • >150 µg/m³: 100 points deduction.
• CO2:
  • ≤800 ppm: No penalty.
  • >800 to ≤1200 ppm: 15 points deduction.
  • >1200 to ≤1500 ppm: 30 points deduction.
  • >1500 to ≤2000 ppm: 60 points deduction.
  • >2000 ppm: 100 points deduction.
• Noise:
  • ≤35 dB: No penalty.
  • >35 to ≤45 dB: 5 points deduction.
  • >45 to ≤60 dB: 15 points deduction.
  • >60 to ≤80 dB: 30 points deduction.
  • >80 dB: 40 points deduction.
• Temperature:
  • 65°F to 80°F: No penalty.
  • 60°F to <65°F or >80°F to 85°F: 10 points deduction.
  • 55°F to <60°F or >85°F to 90°F: 20 points deduction.
  • 50°F to <55°F or >90°F to 95°F: 40 points deduction.
  • <50°F or >95°F: 65 points deduction.
• Humidity:
  • 25% to 75%: No penalty.
  • 15% to <25% or >75% to 85%: 10 points deduction.
  • <15% or >85%: 30 points deduction.

After these deductions, if the score is below 80, it is increased by 20% (multiplied by 1.2). The final score is then constrained between 0 and 100. Higher scores, achieved by maintaining optimal levels across these measurements, indicate a healthy and comfortable indoor environment.