A comparison study of some impulse-response identification methods

Three methods for impulse-response identification are discussed and compared: the white noise input method, the correlation method with pseudorandom binary signal (PRBS) as input, and the fast Fourier transform (FFT) method. Digital simulation results are presented to indicate the advantages and shortcomings of each. The results show that for obtaining impulse-response estimates the PRBS method is the best of the methods tested. It is very effective in dealing with noise, and its high-precision version yields exceptionally accurate results. The white noise method is a theoretically good one, but in practice it is not very effective. However, it gives some useful information about the system. The FFT method is good but very sensitive to noise.<
>