Reply by Ste_ee October 1, 20132013-10-01
no one can help me?
Please
Reply by Ste_ee September 30, 20132013-09-30
HI, i'm implementing all the steps of music algorithm, but i believe there are some problems. 

This is m script: 

N= 50; % number of antenna 
K= 6; %number of signal 
f0=10;   %frequency of signal 
fs=100; %sampling frequency 

d = 0.5; %distance/lambda ratio 

t = 0:1/fs:1; 

x1= sin(2*pi*3*f0*t); 
x2= 2*sin(2*pi*f0*t); 
x3 = cos(2*pi*f0*t); 

x = x1+x2+x3 + wgn(1,length(t),0); % signal + white noise 

[t_matrix Rx]= corrmtx(x,N-1,'mod');  %Rx is autocorrelation matrix NxN. 

[a_vector a_value] = eig(Rx); % find eigenvalues and eigenvector 

[a_val_sort index]= sort(diag(a_value),1,'descend'); 
a_vett_sort=a_vector (:,index); 
Qn = a_vett_sort(:,K+1:N);  %noise subspace 

angles = -90:1:90; 
steering = exp(-i*2*pi*d*(0:N-1)'*sind(angles)); %steering matrix 

for m= 1:length(angles) 
 spectrum(m)= 1/(steering(:,m)'*Qn*Qn'*steering(:,m));   
end 
grid on; 
plot(angles,abs(spectrum)); 

I'm not sure of value of K and its relation with signals x1,x2,x3. 
The spectrum is very strange, i don't understand why it's simmetric! 
I find always +/- angles. 

If is use SVD insted eig?? Rx is always square, and i believe i don't need SVD. 

Is there anyone who may help me? 
Thanks very much to all!