Temperature optimization of MOX sensor arrays for odorant discrimination

We present a methodology based on Information theory tools to optimize the operating temperatures of metal-oxide (MOX) gas sensor arrays and maximize the ability of the system in odor discrimination tasks. We have demonstrated the feasibility of the method by optimizing the temperatures of a foursensor array for an effective discrimination of four odorants. We measured the resistance of four MOX sensors at different operating temperatures when exposed to different concentration levels of ethanol, acetone, 2-butanone and acetic acid. Based on the acquired data we built complete models to shape the responses of the sensors according to the gas exposure conditions and the operating temperature. We applied the Mutual Information (MI) theory to quantify the ability of the sensor array to determine the identity of the stimuli regardless of its concentration, and thereby, select the optimum operation temperature for each sensor.