Testing for Gaussianity in nonlinear system identification

The Hinich test, which makes use of the statistical properties of the bispectrum, is shown to be an appropriate test in the system identification of a second-order Volterra system. The test for Gaussianity of two important sources which are commonly used for Gaussian inputs in nonlinear system identification-commercial software routines for simulation experiments and noise generators for practical experiments-is shown. The deleterious effects on nonlinear system analysis of assuming a random Gaussian input, when in fact it is non-Gaussian, are demonstrated via an experimental investigation, utilizing higher-order spectra, of the nonlinear behavior of a loudspeaker