Spike encoding of artificial olfactory sensor signals

The investigations on artificial olfactory systems in the last 20 years were basically concerned with the development of chemical sensors with selectivity properties similar to that of natural receptor. As a consequence, among the various features of olfaction emerged in the last two decades only the presumed non-selectivity of receptors is currently considered. Therefore, as physiology progresses enlighted the mechanisms of elaboration of signals in the olfactory pathways, artificial nose concept did not evolve much since its first conjecture. In this paper, an attempt to introduce artificial noses more adherent to recent views on olfaction is presented. The typical methodology of translation of conventional analog sensor signals into biologically similar spikes is applied to chemical sensors and a demonstration of the preservation of information sufficient for odour classification is given through a practical example. (c) 2006 Elsevier B.V. All rights reserved.