Crosscorrelation identification of linear time-varying processes using pseudorando sequences

The extension of the crosscorrelation technique, using pseudorandom sequences, to the identification of time-varying processes is considered. Two equivalent techniques are proposed which can track the weighting sequence of a time-varying process, provided that the rate of variation is sufficiently smooth. The methods are based on the assumption that the time variation of each term of the weighting sequence can be described over a period of at least im + 1 times the process settling time by the im terms of a Taylor-series expansion. The periodicity of the input is essential to the methods proposed. Digital-computer results are presented to show the effectiveness of the proposed methods.