SYSTEM-IDENTIFICATION USING NONSTATIONARY SIGNALS

The conventional method for identifying the transfer function of art u nknown linear system consists of a least squares fit of its input to i ts output, It is equivalent to identifying the frequency response of t he system by calculating the empirical cross-spectrum between the syst em's input and output, divided by the empirical auto-spectrum of the i nput process, However, if the additive noise at the system's output is correlated with the input process, e.g., in case of environmental noi se that affects both system's input and output, the method may suffer from a severe bias effect. In this paper we present a modification of the cross-spectral method that exploits nonstationary features in the data in order to circumvent bias effects caused by correlated stationa ry noise, The proposed method is particularly attractive to problems o f multichannel signal enhancement and noise cancellation, when the des ired signal is nonstationary in nature, e.g., a speech or an Image.