By processing multi-dimensional information involved in nonlinear dynamical response, we propose an intellectual process to imitate the high ability of chemical sensing in living organisms. As an example of the application, we show an experimental result on gas sensing when a sinusoidal voltage is applied to a heater inserted into a SnO2 gas sensor. The nonlinearity in the resulting output resistance of the gas sensor is evaluated by using the algorithm of the fast Fourier transformation (FFT). It is demonstrated that one can distinguish small difference in the chemical structure in alkanes and alkenes, based on the information in the higher harmonics of the FFT.