ADAPTIVE VOLTERRA FILTERING WITH COMPLETE LATTICE ORTHOGONALIZATION

This correspondence presents a new recursive least squares (RLS) adapt ive nonlinear filter, based on Volterra series expansion, The main app roach is to transform the nonlinear filtering problem into an equivale nt multichannel, but linear, filtering problem. Then, the multichannel input signal is completely orthogonalized using sequential processing multichannel lattice stages; With the complete orthogonalization of t he input signal, only scalar operations are required, instability prob lems due to matrix inversion are avoided and, good numerical propertie s are achieved, Avoidance of matrix inversion and vector operations re duce the complexity considerably, make the filter simple, highly modul ar and suitable for VLSI implementations. Several experiments demonstr ating the fast convergence properties of the filter are also included in the correspondence.