Title :
A new 2D fast lattice RLS algorithm
Author :
Liu, X. ; Baylou, P. ; Najim, M.
Author_Institution :
Bordeaux Univ., Talence, France
Abstract :
A new two-dimensional fast lattice recursive least squares algorithm is proposed. This algorithm can update the filter coefficients in growing-order form with a computational complexity O((M +1)K1). By associating the previous 2-D data with the region of support, the causality is specified. After appropriately defining the partial order of 2-D data, the 1-D multichannel analogy, order recursion relations and the shift invariance property are derived. The geometrical approaches for the vector space and the orthogonal projection then can be used for solving this 2-D prediction problem. The performance of this new algorithm is examined in comparison with other fast algorithms
Keywords :
adaptive filters; computational complexity; filtering and prediction theory; least squares approximations; 2-D prediction problem; RLS algorithm; adaptive filter; causality; computational complexity; filter coefficients; order recursion relations; orthogonal projection; shift invariance property; two-dimensional fast lattice recursive least squares algorithm; vector space; Adaptive algorithm; Adaptive filters; Computational complexity; Equations; Filtering algorithms; Lattices; Least squares approximation; Least squares methods; Resonance light scattering; Transversal filters;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1992. ICASSP-92., 1992 IEEE International Conference on
Conference_Location :
San Francisco, CA
Print_ISBN :
0-7803-0532-9
DOI :
10.1109/ICASSP.1992.226234
