Rapid detection of toxic agents in the indoor environment is essential to protecting building occupants from accidental or intentional releases. While there is much research dedicated to designing sensors to detect airborne toxic contaminants, little research has addressed how to incorporate such sensors into a monitoring system designed to protect building occupants. To design sensor systems, sensor designers must quantify design tradeoffs, such as response time and accuracy, to optimize the performance of an overall system. We illustrate the importance of a systems approach for properly evaluating such tradeoffs, using data from tracer gas experiments conducted in a three-floor unit at the Dugway Proving Grounds, Utah. We apply Bayesian statistics to assess the effects of various sensor characteristics, such as response time, threshold level and accuracy, on overall system performance. We evaluated the system performance by the time (and thus amount of data) needed to characterize the release (location, amount released, and duration). We demonstrate that a systems perspective is necessary to understand the potential benefits of selecting values of specific sensor characteristics to optimize sensor system performance.
BT - Atmospheric Environment DA - 06/2006 DO - 10.1016/j.atmosenv.2006.01.052 IS - 19 LA - eng N2 -Rapid detection of toxic agents in the indoor environment is essential to protecting building occupants from accidental or intentional releases. While there is much research dedicated to designing sensors to detect airborne toxic contaminants, little research has addressed how to incorporate such sensors into a monitoring system designed to protect building occupants. To design sensor systems, sensor designers must quantify design tradeoffs, such as response time and accuracy, to optimize the performance of an overall system. We illustrate the importance of a systems approach for properly evaluating such tradeoffs, using data from tracer gas experiments conducted in a three-floor unit at the Dugway Proving Grounds, Utah. We apply Bayesian statistics to assess the effects of various sensor characteristics, such as response time, threshold level and accuracy, on overall system performance. We evaluated the system performance by the time (and thus amount of data) needed to characterize the release (location, amount released, and duration). We demonstrate that a systems perspective is necessary to understand the potential benefits of selecting values of specific sensor characteristics to optimize sensor system performance.
PY - 2006 SP - 3490 EP - 3502 ST - Atmospheric Environment T2 - Atmospheric Environment TI - Systems Approach to Evaluating Sensor Characteristics for Real-Time Monitoring of High-Risk Indoor Contaminant Releases VL - 40 SN - 13522310 ER -