Title :
A normalized lattice algorithm for AR spectral analysis and system identification filters with symmetric impulse response
Author :
Berberidis, Kostas ; Theodoridis, Sergios
Author_Institution :
Dept. of Comput. Eng., Patras Univ., Patra Rio, Greece
fDate :
3/1/1990 12:00:00 AM
Abstract :
A time-recursive normalized lattice algorithm is derived for the least-squares (LS) estimation of the unknown parameters of an FIR (finite impulse response) filter with symmetric impulse response. A byproduct is an algorithm for the LS estimation of the unit step backward predictor of the input time series, based on the simultaneous minimization of the forward and backward total squared error power, which has successfully been used in the maximum-entropy spectral analysis (MESA) method
Keywords :
digital filters; filtering and prediction theory; least squares approximations; parameter estimation; signal processing; spectral analysis; FIR filter; autoregressive spectral analysis; backward total squared error power; finite impulse response; forward total squared error power; input time series; least-squares estimation; maximum-entropy spectral analysis; parameter estimation; signal processing; simultaneous minimization; symmetric impulse response; system identification filters; time-recursive normalized lattice algorithm; unit step backward predictor; Entropy; Equations; Finite impulse response filter; Lattices; Linear systems; Nonlinear filters; Recursive estimation; Signal processing algorithms; Spectral analysis; System identification;
Journal_Title :
Acoustics, Speech and Signal Processing, IEEE Transactions on
