A new adaptive linear phase LS lattice structure with application to spectral line estimation

Least-squares estimation of the unknown symmetric impulse response of an FIR (finite-impulse-response) filter with linear phase results in a well-structured linear system of equations. Recently two novel algorithms for the efficient solution of this system, which respect the underlying symmetry of the linear phase filtering, were introduced. The symmetry of the problem is further exploited here to derive a novel, exact least-squares lattice algorithm for adaptive processing. The new lattice structure has considerably lower computational complexity and a more parsimonious parameterization, compared to previously derived lattice algorithms.<>