Plot functions in a for cycle

Hey guys, thanks in advance,

I have this code , that calculates ambiguity function for signals with a lot of values. On each iteration, I want to plot the graphic with the code below ( cross-correlation). Is there a better way, than for each iteration, interrupt for loop to see the plot?

for k = 1 : 500000 : length(Reference_signal) 
    for kk = 1: 500000 : length(Surveillance_signal) 
        
        Reference_signal_samples = Reference_signal(k : k + 499999);
        Surveillance_signal_samples= Surveillance_signal(kk : kk + 499999)
        [afmag3,delay3] = ambgfun(Reference_signal_samples,Surveillance_signal_samples,Fs,[250000 250000],'cut','Doppler');
        afmag3 = afmag3*1; % Select plot gain *1
        afmag3(afmag3>1 )= 1;
        
    end
    
end
% Plot cross-correlation
[pks3,index3] = max(afmag3);
xMax3 = delay3(index3);
yMax3 = pks3;
subplot(3,2,3)
plot(delay3,afmag3,'LineStyle','-'); 
hold on
%plot3(delay3(pks3,index3), afmag3(pks3,index3), pks3, '^r');
textString3 = sprintf('(%f, %f)', xMax3, yMax3);
text(xMax3, yMax3,textString3,"Color",'b','FontSize',10);
hold off
shading interp;
%xlim([-0.5e-5 16e-5]);
xlim auto;
grid on; 
colorbar;
xlabel('Delay \tau (s)');
ylabel('Ambiguity Function Magnitude');
title('Cross-correlation');

Answer 1

[Voss]

You can move your plotting code inside the loops if you want to see a plot in each iteration, yes.

I see that you're doing subplot(3,2,3), so I take that to mean that one of the loops iterates 3 times and the other iterates 2 times, and that you would like to see the plots arranged in a 3-by-2 subplot grid.

If that's the case, something like this would work (actually, this will work regardless of how many iterations each loop has, as long as the idea is to plot the result of each ambgfun call in its own plot):

pulse_size = 500000;

% random data:
Reference_signal = rand(1,2*pulse_size);
Surveillance_signal = rand(1,3*pulse_size);
Fs = 1e6;

% number of blocks of 500000 samples, for each signal:
nRef = numel(Reference_signal)/pulse_size;
nSurv = numel(Surveillance_signal)/pulse_size;

% index of "middle" axes vertically, for ylabel
% (you can use tiledlayout rather than subplot
% to have an actual common ylabel):
if mod(nSurv,2) == 1
    nSurvHalf = (nSurv-1)/2+1;
else
    nSurvHalf = nSurv/2;
end

% it's convenient to loop over natural numbers
for k = 1:nRef
    
    % Reference_signal_samples doesn't depend on kk, so get it once per k,
    % here, before the kk loop:
    Reference_signal_samples = Reference_signal((k-1)*pulse_size+(1:pulse_size));
    
    for kk = 1:nSurv
        
        Surveillance_signal_samples = Surveillance_signal((kk-1)*pulse_size+(1:pulse_size));
        
        [afmag3,delay3] = ambgfun(Reference_signal_samples,Surveillance_signal_samples,Fs,[1 1]*pulse_size/2,'cut','Doppler');
%         afmag3 = afmag3*1; % Select plot gain *1
        afmag3(afmag3 > 1) = 1;
        
        % Plot cross-correlation
        [pks3,index3] = max(afmag3);
        xMax3 = delay3(index3);
        yMax3 = pks3;
        
        subplot(nSurv,nRef,nRef*(kk-1)+k);
        
        plot(delay3,afmag3,'LineStyle','-'); 
        textString3 = sprintf('(%f, %f)', xMax3, yMax3);
        text(xMax3, yMax3,textString3,"Color",'b','FontSize',10);
        grid on;
        
        % xlabel the bottom-most plots only
        if kk == nSurv
            xlabel('Delay \tau (s)');
        end
        % ylabel the plot on the left which is halfway down
        if k == 1 && kk == nSurvHalf
            ylabel('Ambiguity Function Magnitude');
        end
        % title each plot
        title(sprintf('Ref. pulse: %d Surv. pulse: %d',k,kk));
        
    end
    
end

% common title for all plots
sgtitle('Cross-correlation');

Try it on your data and see if it produces something like what you have in mind.

[Miguel Albuquerque]

Hey man, thanks a lot for the help, I made this changes, since I dont need any pulse size:

But I keep getting this error in [afmag3,delay3]=...

%Error

Error using ambgfun>parseInput (line 283)
The pulse is longer than the available data in X.

Error in ambgfun (line 83)
[cut, cutvalue, xsub, xlen, ysub, ylen, fs, isCross] = parseInput(x,varargin{:});

% Code starts here:


nRef = numel(Reference_signal)/Fs;
nSurv = numel(Surveillance_signal)/Fs;

% index of "middle" axes vertically, for ylabel
% (you can use tiledlayout rather than subplot
% to have an actual common ylabel):
if mod(nSurv,2) == 1
    nSurvHalf = (nSurv-1)/2+1;
else
    nSurvHalf = nSurv/2;
end

% Time delay processing


for k = 1:nRef
    Reference_signal_samples = Reference_signal(k);
    
    for kk =1:nSurv
        
        
        Surveillance_signal_samples= Surveillance_signal(k)
        
        
        [afmag3,delay3] = ambgfun(Reference_signal_samples,Surveillance_signal_samples,Fs,[250000 250000],'cut','Doppler');
        afmag3(afmag3>1 )= 1;
        
        % Plot cross-correlation
        [pks3,index3] = max(afmag3);
        xMax3 = delay3(index3);
        yMax3 = pks3;
        
        subplot(nSurv,nRef,nRef*(kk-1)+k);
        
        
        plot(delay3,afmag3,'LineStyle','-'); 
        textString3 = sprintf('(%f, %f)', xMax3, yMax3);
        text(xMax3, yMax3,textString3,"Color",'b','FontSize',10);
        grid on; 
        
        % xlabel the bottom-most plots only
        if kk == nSurv
            xlabel('Delay \tau (s)');
        end
        % ylabel the plot on the left which is halfway down
        if k == 1 && kk == nSurvHalf
            ylabel('Ambiguity Function Magnitude');
        end
        % title each plot
        title(sprintf('Ref. pulse: %d Surv. pulse: %d',k,kk));
        
    end
    
end
% common title for all plots
sgtitle('Cross-correlation');

[Voss]

You want to run ambgfun on signals with only one sample?

% k is a scalar, so Reference_signal_samples is a scalar
Reference_signal_samples = Reference_signal(k);

% same
    Surveillance_signal_samples= Surveillance_signal(k)

The code in your question and the code in my answer both took blocks of 500000 samples at a time. The difference is that your code had k and kk going 1, 500001, 1000001, ... and mine had k going 1, 2, 3, ... but the sequence of signal blocks taken were the same in both codes.

I think you still need to use 500000 samples at a time.

[Miguel Albuquerque]

okay I understood it, thanks a lot

[Miguel Albuquerque]

My reference signal and surveillance signal, have more than 50 million values, is this code good to process it?

[Miguel Albuquerque]

And one more thing, when I run it, I dont see the plots, I dont know what is happening

[Voss]

If each signal is 50 million samples, then using blocks of 500000 samples at a time would give you 10000 plots, arranged in a 100-by-100 grid, so it would be difficult to see anything.