ADAPTIVE OVERFITTING LATTICE FILTER FOR PARAMETER-ESTIMATION OF TIME-VARYING SYSTEMS

This paper presents an approach to estimate AR and MA parameters of an ARMA model. A (p+q')-stage overfitting lattice structure filter is used to obtain estimates of the AR and MA parameters of an ARMA(p, q) (q' greater-than-or-equal-to q) model. At first. an overfitting lattice filter is used to estimate the AR parameters, and then with little added computation, the MA parameters can also be obtained easily. This paper also proposes a new adaptive overfitting lattice filter for parameter estimation of time-varying systems. A new method is proposed to obtain the PARCOR coefficients that can be used to estimate the ARMA parameters adaptively.