sd.m

来自「A simple code for understanding lms nlms」· M 代码 · 共 23 行

function [w,J]=sd(mu,winit,N,R,p,Jmin);


% Input arguments:
% mu = step size, dim 1x1
% winit = start values for filter coefficients, dim Mx1
% N = no. of iterations, n=[0,...,N-1], dim 1x1
% R = autocorrelation matrix, dim MxM
% p = cross-correlation vector, dim Mx1
% Jmin = minimum mean square error, dim 1x1
%
% Output arguments:
% w = matrix containing the filter coefficients
% as a function of n along the rows, dim MxN
% J = mean square error as a function of n, dim 1XN
w=zeros(length(winit),N);
J=zeros(1,N);
w(:,1)=winit(:);
for n=1:N-1
    w(:,n+1)=w(:,n)+mu*(p-R*w(:,n));
    J(n)=Jmin+(w(:,n)-R\p)'*R*(w(:,n)-R\p);
end
J(N)=Jmin+(w(:,N)-R\p)'*R*(w(:,N)-R\p);