ARMA-System Identification and Rational Approximation

Abstract

The problem of identifying a linear time-invariant (LTI) system from measurements of its output response to a known input excitation, such as a white noise source, is of fundamental importance in many areas in engineering. Such an identification naturally allows one to predict the system outputs, and as a result this problem has considerable impact in areas where forecasting is of extreme importance. Such applications exist in diverse fields such as ocean modeling, economics system theory, seismology, pattern recognition and signal processing, to name a few.