How can I solve findpeaks error in Octave

theta = -90:0.01:90;      
for i=1:length(theta)
    SS = zeros(Nr,1);
    SS = exp(-1j*2*pi*d*(0:Nr-1)'*sind(theta(i))/lambda);
    PP = SS'*(Vn*Vn')*SS;
    Pmusic(i) = 1/ PP;
end
Pmusic = real(10*log10(Pmusic)); %Spatial Spectrum function
[pks,locs] = findpeaks(Pmusic,theta,'SortStr','descend','Annotate','extents');
MUSIC_Estim = sort(locs(1:K))

I want to run MUSIC matlab code in Octave (because I don't have matlab)

in matlab free 30day edition, It does work well but not in Octave.

>> error: findpeaks: non-string for Parameter name or Switch

how can I solve this error? I downloaded signal library and changed but it didn't work.

In Octave, provided matlab example works well. why doesn't work above code?

t = 2*pi*linspace(0,1,1024)';
 y = sin(3.14*t) + 0.5*cos(6.09*t) + 0.1*sin(10.11*t+1/6) + 0.1*sin(15.3*t+1/3);

 data = abs(y + 0.1*randn(length(y),1)); # Positive values + noise
 [pks idx] = findpeaks(data,"MinPeakHeight",1);

 dt = t(2)-t(1);
 [pks2 idx2] = findpeaks(data,"MinPeakHeight",1,...
                              "MinPeakDistance",round(0.5/dt));

 subplot(1,2,1)
 plot(t,data,t(idx),data(idx),'or')
 subplot(1,2,2)
 plot(t,data,t(idx2),data(idx2),'or')
plot(t,data,t(idx2),data(idx2),'or')

please help. thanks.

I tried

>> pkg install "https://github.com/gnu-octave/pkg-control/releases/download/control-3.5.2/control-3.5.2.tar.gz"

>> pkg install "https://downloads.sourceforge.net/project/octave/Octave%20Forge%20Packages/Individual%20Package%20Releases/signal-1.4.3.tar.gz"

>> pkg load control

>> pkg load signal

in Octave command window but it didn't work.

This MUSIC algorithm matlab code is not written by me, but it's just a short calculator code, and I'll upload the entire code rather than write the code of the smaller version to see the value of MUSIC_Estim accurately.

DOA = [35 30];      %Direction of arrival (Degree)
T   = 200;         %Snapshots (or Samples)
K   = length(DOA); %The number of signal source(or traget)
Nr  = 9;          %Number of receiver's antennas
lambda = 1;      %Wavelength
d   = lambda/2;    %Receiver's antennas spacing
SNR = 10;           %Signal to Noise Ratio (dB)
A = zeros(Nr,K);   %Steering Matrix


for k=1:K
    A(:,k) = exp(-1j*2*pi*d*sind(DOA(k))*(0:Nr-1)'/lambda);
%Assignment matrix
end
Vj = diag(sqrt((10.^(SNR/10))/2));
s = Vj* ( randn(K,T) + 1j*randn(K,T) );
noise = sqrt(1/2)*(randn(Nr,T)+1j*randn(Nr,T));

X = A*s;
X = X+noise;      %Insert Additive White Gaussain Noise (AWGN)

% MUSIC (MUltiple SIgnal Classification)
Rx = cov(X');                     %Data covarivance matrix
[eigenVec,eigenVal] = eig(Rx);    %Find the eigenvalues and eigenvectors of Rx
Vn = eigenVec(:,1:Nr-K);          %Estimate noise subspace (Note that eigenvalues sorted ascendig on columns of "eigenVal")
theta = -90:0.01:90;       %Grid points of Peak Search
for i=1:length(theta)
    SS = zeros(Nr,1);
    SS = exp(-1j*2*pi*d*(0:Nr-1)'*sind(theta(i))/lambda);
    PP = SS'*(Vn*Vn')*SS;
    Pmusic(i) = 1/ PP;
end
Pmusic = real(10*log10(Pmusic)); %Spatial Spectrum function
[pks locs] = findpeaks(Pmusic, "DoubleSided");
locs = theta(locs);
MUSIC_Estim = sort(locs)

figure;
plot(theta,Pmusic,'-b',locs(1:K),pks(1:K),'r*'); hold on
text(locs(1:K)+2*sign(locs(1:K)),pks(1:K),num2str(locs(1:K)'))

xlabel('Angle \theta (degree)'); ylabel('Spatial Power Spectrum P(\theta) (dB)')
title('DOA estimation based on MUSIC algorithm ')
xlim([min(theta) max(theta)])
grid on

///
locs = theta(locs);
and
MUSIC_Estim = sort(locs)

value is not peak. so I tried

Pmusic = real(10*log10(Pmusic)); %Spatial Spectrum function
[pks,locs] = findpeaks(Pmusic, "DoubleSided","MinPeakHeight", 5);
[sorted_pks, sorted_idx] = sort(pks(1:K));  % Sort peaks in descending order
MUSIC_Estim = theta(locs(sorted_idx(1:K)))      % Select top K peak locations from sorted peaks

figure;
plot(theta,Pmusic,'-b',MUSIC_Estim,pks(1:K),'r*'); hold on
text(MUSIC_Estim(1:K)+2*sign(sorted_idx(1:K)),pks(1:K),num2str(MUSIC_Estim(1:K)'))

like this but it has another problem of sorting peak values and can't get more than 2 DOA.

Basically Octave findpeaks supports a single argument for the data: [pks, loc] = findpeaks(data) return the peak values in pks, and their indexes in loc.

Matlab findpeaks has an optional extra input argument that contains the abscissas of the data samples [pks, loc] = findpeaks(data,x). On output, loc contains the abscissas of the peaks instead of the indexes.

But it's not the only difference. The Octave version doesn't have the "SortStr" property. But you don't need it as you are sorting afterwards anyway. It doesn't have either the "Annotate" property, but you don't need it either (from Matlab doc: "This argument is ignored if you call findpeaks with output arguments.")

At the end, in your first exemple you have to replace the findpeaks call by these two lines:

[pks,locs] = findpeaks(Pmusic);
locs = theta(locs)   % replaces the indexes by the theta values

Note that you could have got all of this by just reading the documentations.