eigen.m

该程序对ofdm系统中mmsnr的信道缩短算法的误码率性率进行了仿真

%EIGEN Solves the eigenvalue problem in design of the unit-energy  constrained 
% minimum mean-squared error time domain equalizer.
% [B, W, D, MSE, R, Dv] = EIGEN(RXX, RYY, RXY, Dmin, Dmax, Nb, Nw, L) 
% returns the optimal target impulse response B, the time domain 
% equalizer W, and the delay D. 

%MSE is the resulting mean-squared error.  
% R is the input-output cross-correlation matrix obtained with the the optimum delay D
% and Dv is a vector containing the mean-squared error for delay values between Dmin and Dmax.

% RXX is the input autocorrelation matrix. 
% RYY is the output autocorrelation matrix.
% RXY is the input-output cross-correlation vector used to generate the
% input-output cross-correlation matrix depending on the current delay. 
% Dmin and Dmax define the search interval for the optimal delay.
% Nb is the number of taps in the target  impulse response.
% Nw the number of taps in the time domain equalizer. 

function [b, w, d, MSE, Rxyopt, Dv] = ...
	eigen(Rxx,Ryy,rxy,Dmin,Dmax,Nb,Nw,L) 

% initialize variables
Dv = ones(1,Dmax);
MSE = inf;


for delay = Dmin:Dmax; % for each delay to be searched 
  % calculate the input-output cross-correlation matrix
  Rxy = toeplitz(rxy(L+(0:Nb-1)+delay+1),rxy(L-(0:Nw-1)+delay+1));
  % calculate the MSE matrix
  Rdelta = Rxx - Rxy*inv(Ryy)*Rxy';
  % find the eigenvector corresponding to the minimum eigenvalue
  [mse bb] = mineig(Rdelta); 
% save the MSE 
  Dv(delay) = mse;
  
  if mse < MSE    % if the current MSE is lower then previous ones
	% save the current target, delay, MSE, and crosscorrelation matrix
	b = bb;
	d = delay;
	MSE  = mse;
	Rxyopt = Rxy;
  end
end
% use the optimum target impulse response to find the optimum TEQ
w = inv(Ryy)*Rxyopt'*b;