基于MATLAB的MMSE算法估计

function [mmse,mse,f_opt,delta] = MMSE_fse (h,hrec,snr,Lf) % ========================================================= % function [mmse,mse,f_opt,delta] = MMSE_fse (h,hrec,snr,Lf) % --------------------------------------------------------- % Find the MMSE for a specific estimated channel. % The two channels (h,hrec) must be of the same length. % % hrec : the reconstructed channel % snr : the SNR at the equalizer input % Lf : length of the equalizer to use % % mmse : minimum-MSE achieved over all delays % f_opt : the equalizer that achieves the mmse % delta : the optimal delay % % Author: H. Pozidis, September 23, 1998 % ========================================================= if (length(h)-length(hrec) ~= 0) error('The channels must be of the same length'); end if (snr >= 0) snr = 10^(snr/10); lambda = 1/snr; else lambda = 0; end h = h(:); hrec = hrec(:); H = convmtx(h,Lf); C = convmtx(hrec,Lf); P = size(C,1); He = H(1:2:P,:); Ce = C(1:2:P,:); P = size(Ce,1); A = inv(Ce'*Ce + lambda*eye(Lf)); f_eq = A*Ce'; h_del = eye(P); mse_mat = (h_del-He*f_eq)'*(h_del-He*f_eq) + f_eq'*f_eq*lambda; mse = diag(mse_mat); [mmse,delta] = min(mse); f_opt = f_eq(:,delta);