Signal Processing of Exhaled CO2 as Tracer Gas in Residential Ventilation Assessment

Background: Indoor air contaminants generally have a greater impact on health than outdoor air contaminants, which increases the importance of a dependable, accessible, and minimally impactful method for measuring indoor air exchange rates. Objective: Evaluate the use of naturally generated CO2 as a tool to measure indoor ventilation. Methods: Indoor CO2 levels were measured over seven sample intervals in an airtight one-bedroom apartment with two residents. High frequency noise was removed from the measurements with Fourier, Kalman, LOESS, and rolling average filters. Root-mean squared errors (RMSE) between filtered and measured CO2 were calculated and compared for each sample interval and filter pair. A multivariable linear regression was used to assess differences between digital filters. Local minima and maxima were identified to calculate air exchange rates. The R statistical software was used for all data management and analysis. Results: The RMSE for all filter types had geometric standard deviations between one and two, indicating that all filters were stable across sample intervals. Results of the multivariable linear regression indicate that the RMSE of the Fourier filter were significantly lower than those of the Kalman filter with a P-value ofConclusions:The Fourier filter performed best based on visual analysis and RMSE comparisons. All filters except for the rolling average filter identified the majority of primary local minima/maxima effectively