Information coding in artificial olfaction multisensor arrays.

Abstract

High-density sensor arrays were prepared with microbead vapor sensors to explore and compare the information coded in sensor response profiles following odor stimulus. The coded information in the sensor-odor response profiles, which is used for odor discrimination purposes, was extracted from the microsensor arrays via two different approaches. In the first approach, the responses from individual microsensors were separated (decoded array) and independently processed. In the second approach, response profiles from all microsensors within the entire array, i.e., the sensor ensemble, were combined to create one response per odor stimulus (nondecoded array). Although the amount of response data is markedly reduced in the second approach, the system shows comparable odor discrimination rates for the two signal extraction methods. The ensemble approach streamlines system resources without decreasing system performance. These signal compression approaches may simulate or parallel information coding in the mammalian olfactory system.