Then selected statistical properties of the amplified noise are analyzed, and a corresponding pattern is generated as the stochastic fingerprint of the sensed agent.
Often the power density spectrum of the stochastic signal is used as output pattern however FES has been proven effective with more advanced methods, too, such as higher-order statistics.
During the 1990s, several authors (for example, Bruno Neri and coworkers, Peter Gottwald and Bela Szentpali) had proposed using the spectrum of measured noise to obtain information about ambient chemical conditions.
However, the first systematic proposal for a generic electronic nose utilizing chemical sensors in FES mode, and the related mathematical analysis with experimental demonstration, were carried out only in 1999 by Laszlo B. Kish, Robert Vajtai and C.G.
In 2003, Alexander Vidybida from Bogolyubov Institute for Theoretical Physics of the National Academy of Sciences of Ukraine has proven mathematically that adsorption–desorption fluctuations during odor primary reception can be used for improving selectivity.