A recursive adaptive algorithm of multidimensional nonlinear filtering — bi-independent transform of signals

Abstract

New algorithms of multidimensional adaptive nonlinear filtration are suggested. The algorithms are based on generating a nonlinear estimate of the signal component and subtracting it from the realization of the input random process. The paper considers the rate of convergence of these algorithms and the possibility of their realization based on filters with finite and infinite memory, and with adjustment of weight coefficients by the direct calculation method.