Title :
The Generalized Sliding-Window recursive least-squares lattice filter
Author :
Spinelli, Hugo M. ; Merched, R.
Author_Institution :
Dept. of Electron. & Comput. Eng., Univ. Fed. do Rio de Janeiro, Rio de Janeiro, Brazil
Abstract :
The Generalized Sliding-Window Fast Transversal Filter (GSWFTF) is an extension of the exponentially weighted fast RLS algorithm, which inherits its rapid convergence while featuring the desired robustness of affine projection algorithms. Motivated by the numerical difficulties typical from such adaptive transversal realizations, we develop the lattice version of the GSWRLS. Our simulations show significant improvement on stability compared to the GSWFTF and the classical, sliding-window RLS lattice recursions. The new order-recursive filter exhibits the best of both robust and fast converging algorithms, without showing divergence over millions of iterations.
Keywords :
lattice filters; least squares approximations; recursive filters; GSWFTF; GSWRLS; adaptive transversal realizations; affine projection algorithms; exponentially weighted fast RLS algorithm; fast converging algorithms; generalized sliding-window fast transversal filter; generalized sliding-window recursive least-squares lattice filter; order-recursive filter; robust converging algorithms; sliding-window RLS lattice recursions; Convergence; Estimation error; Frequency modulation; Lattices; Numerical stability; Signal processing algorithms; Vectors; Adaptive; RLS; fast;
Conference_Titel :
Acoustics, Speech and Signal Processing (ICASSP), 2013 IEEE International Conference on
Conference_Location :
Vancouver, BC
DOI :
10.1109/ICASSP.2013.6638762