Problem in MUSIC Algorithm for DOA Estimation

Hi all.........

I want to find out angle of arrival using planer antenna array for radar
(Digital Monopulse) application. For concept development, first i make 2x2
planer array, and try to find out direction of arrival of sources, further
i need to make null at the direction of interference.

which algorithm is better for that???? or MUSIC algorithm is suitable for
that????

I try to use MUSIC algorithm for DOA application, and write codes in
MATLAB for 1D MUSIC.
these codes gives DOA, but have some problem; 
(i take linear array of 10 elements, f = 1GHz and sampling frequency Fs =
5GHz)
1) the resolution is not as good as written in literature.
2) if i take no of signal > 2 then it able to find out DOA but Estimation
of no. of signal is only 2 (it gives M - H (no of multiplicity of minimun
eigen values) is only 1 or 2 for all the cases).

I think some problem in signal(X(t)=AS + N), which i take as;
 Xn = Sin (wt + k d Sin(thetha))* exp (-j*(n-1)k d Sin(thetha))

Matlab codes for that is given below;
for k=1:samples         % no of time instants
    for i=1:M           % no of elements
        for ns=1:D      % no of incoming signals
            S1(i,k)=S1(i,k)+ sin(2*pi*f*k/Fs +
2*pi*sin(theta(ns))/dfactor)* exp(-j*(i-1)*2*pi*sin(theta(ns))/dfactor);
        end
    end
end

X = awgn(S1,20);     % Add white Gaussian noise to signal (M x sampless) 

can any one give ur valuable suggations.
THANKS.  

On 18 Sep, 14:55, "amit00" <getamitshu...@rediffmail.com> wrote:
> Hi all.........
>
> I want to find out angle of arrival using planer antenna array for radar
> (Digital Monopulse) application. For concept development, first i make 2x2
> planer array, and try to find out direction of arrival of sources, further
> i need to make null at the direction of interference.

This has been studied for decades already...

> which algorithm is better for that???? or MUSIC algorithm is suitable for
> that????

No. The Achilles hell of MUSIC is that the number of sensors in the
array is directly linked to the number of signals the algorithm
can handle. Which means that you need to know the number of
signals you will need to handle, and select the number of elements
in the array based on what you come up with. Do note that each source
counts as one signal, each multipath counts as one signal.

So one source + one multipath counts as two sources.

Rune

Hi Rune.........
Thanks ....

  then can u please tell me, which algorithm is better for that type of
application (Digital Monopluse radar as i mentioned in prev. mail).....

BTW ..can u help to resolve problem, coming in MUSIC algo.. as i write in
preveous mail..

       I take linear array of 10 elements, f = 1GHz and sampling frequency
Fs =5GHz.
1) the resolution is not as good as written in literature.
2) if i take no of signal > 2 then it able to find out DOA but Estimation
of no. of signal is only 2 (it gives M - H (no of multiplicity of minimun
eigen values) is only 1 or 2 for all the cases).
 
I think some problem in signal(X(t)=AS + N), which i take as;
 Xn = Sin (wt + k d Sin(thetha))* exp (-j*(n-1)k d Sin(thetha))

Matlab codes for that is given below;
for k=1:samples         % no of time instants
    for i=1:M           % no of elements
        for ns=1:D      % no of incoming signals
            S1(i,k)=S1(i,k)+ sin(2*pi*f*k/Fs +
2*pi*sin(theta(ns))/dfactor)* exp(-j*(i-1)*2*pi*sin(theta(ns))/dfactor);
        end
    end
end
X = awgn(S1,20);     % Add white Gaussian noise to signal (M x sampless) 




>No. The Achilles hell of MUSIC is that the number of sensors in the
>array is directly linked to the number of signals the algorithm
>can handle. Which means that you need to know the number of
>signals you will need to handle, and select the number of elements
>in the array based on what you come up with. Do note that each source
>counts as one signal, each multipath counts as one signal.
>
>So one source + one multipath counts as two sources.
>
>Rune
>

On 20 Sep, 14:21, "amit00" <getamitshu...@rediffmail.com> wrote:
> Hi Rune.........
> Thanks ....
>
> &#4294967295; then can u please tell me, which algorithm is better for that type of
> application (Digital Monopluse radar as i mentioned in prev. mail).....

I have no idea.

> BTW ..can u help to resolve problem, coming in MUSIC algo.. as i write in
> preveous mail..

The problem is that you use MUSIC with bad system parameters.
You need to read up on the MUSIC algorithm - and contemplate
what you read - and see how the array parameters interact with
the MUSIC algorithm.

Just remember that there is the very real possibility that you
migh conclude from such a review that MUSIC is not fit for
the purpose you want it for.

Rune