System identification in the presence of drift using pseudorandom binary signals

This paper is concerned with the elimination of the effects of low frequency polynomial-type drift in the crosscorre-Iation identification of linear time-invariant system using pseudorandom binary test perturbations. Two methods are presented for satisfactory elimination of the drift effects. Of these, method-1 determines the drift error by means of simultaneous crosscorrelation of the system output with the input signal and its phase-inverse. Method-2 is an improvement on the drift-correction scheme by Davies et at. The two methods are computationally simple, and economical in the sense that no extra equipment is necessary.