Title :
Efficient computation of multichannel Wiener filters with linear phase
Author :
Kalouptsidis, N.
Author_Institution :
Dept. of Phys., Athens Univ., Greece
fDate :
4/1/1988 12:00:00 AM
Abstract :
A family of fast-lattice and transversal algorithms for multichannel linear-phase (LP) Wiener filters is developed. Due to the strong symmetry properties imposed by the phase constraints, the proposed computational process involve only one internal filter that propagates according to a Durbin-like recursion, in contrast to unconstrained multichannel Wiener filtering (absence of phase constraints) where both the forward and backward predictors are carried, coupled in a pair of Levinson-like recursions. Normalized lattice realizations are derived using a suitable power-sum normalization and parameterized by internal reflection coefficients with singular values inside the unit circle
Keywords :
digital filters; filtering and prediction theory; Durbin-like recursion; computational process; digital filters; fast lattice algorithm; internal reflection coefficients; linear phase; multichannel Wiener filters; phase constraints; power-sum normalization; symmetry properties; transversal algorithm; Adaptive filters; Equations; Filtering algorithms; Finite impulse response filter; Frequency; Lattices; Nonlinear filters; Reflection; Transversal filters; Wiener filter;
Journal_Title :
Circuits and Systems, IEEE Transactions on