Microelectromechanical-system (MEMS) metaloxide gas sensors have reached a mature stage, which makes mass market applications in the automotive area possible. In contrast to the already established flap-control system, which controls the access of (combustion) gases from outside the vehicle to the car cabin, the system studied here detects odor events created within the car cabin. The events under study have been cigarette smoke, fast-food odor, manure, and bioeffluents (flatulence). As the reference cannot be a "simple analytical measurement," a human test panel for assessing the hedonic impression on a scale from 0 to 5 is used as reference. The technical system is a MEMS metal-oxide-sensor array consisting of three different sensors. The data-evaluation approach used here is combining the human-sensory data and the MEMS sensor data. The task is performed by the combination of two independent algorithms, where one is related to the normalized conductance and the other to signal variance. Using a combined approach has the advantage that "false" events are suppressed. After the algorithm was successfully transferred onto a microcontroller, real-life data were recorded and classified. Several practical examples are given in this paper. The overall gas-sensor system reaches good accordance with the human-sensory impression, which is represented by air quality levels. This enables the design of a demand-controlled ventilation system. Index Terms-Air quality, gas sensor, heating ventilation and air conditioning (HVAC), low power, microelectromechanical systems (MEMS), sensor array.

No abstract