The extended instrumental variable(EIV) method for ARMA spectral estimation was presented for improving the convergence property of ELS and RML which have slow convergence in MA parameters. In this paper, analysis of numerical properties and lattice i...
The extended instrumental variable(EIV) method for ARMA spectral estimation was presented for improving the convergence property of ELS and RML which have slow convergence in MA parameters. In this paper, analysis of numerical properties and lattice implementation of EIV for high resolution spectral analysis has been researched as follows
(1) Improve performance of EIV using he prefilter,
(2) Prove the convergence of EIV using ordinary differential equations,
(3) Prove the unbiasness of EIV using the whiteness property of residual processes,
(4) Derive prewindowed type lattice algorithms which are computationally efficient (o(p) operations required) and profitable in VLSI circuit implementation.