Orthogonal decomposition of chemo-sensory cues

Most practices of gas and mixture identification follow the conventional signal processing protocol of mapping the sensor signal to a stand-alone feature upon each presentation of a gas sample. This single-feature representation scheme, however, may not convey enough qualitative information to successfully represent distinct chemical analytes. Following this argument, it is suggested that having additional qualitative information can significantly benefit these results. In this paper, we explore a multi-feature processing scheme for chemo-sensory response cues. The method consists of decomposing the sensor response signal to a set of orthogonal Bessel functions. Bessel functions give us the possibility of decomposing the sensor signal response by a finite number of series of orthogonal functions that are localized in the magnitude of the oscillations’ changes with the time. We believe, therefore, that expressing the sensor response signals by these coefficients is advantageous. We demonstrate the utility of this novel scheme in identifying and quantifying distinct analytes using metal-oxide gas sensors.