How do I convert a time domain signal to frequency domain?
Ältere Kommentare anzeigen
I am having some problems converting a time domain signal into frequency domain--I don't have constant sampling time.
In time domain, is this the way to plot the graph?
value = [42007935 111212895 184546560 238219725 238219725 184546560 111212895 42007935]; time = [0, ,5,1.25,2, 2,5,5, 4, 4.5, 6, 7, 8 ];
plot (time, value);
X and/or Y arguments ignored.
Could anyone help me plot a signal that does not have constant time?
2 Kommentare
Star Strider
am 12 Mai 2014
If you want to plot them, your time and value arrays have to have the same number of elements. They don’t.
Look for NUFFT (non-uniform FFT) literature -- there's quite a lot.
You surely have the sample times or at least delta-t's, don't you? W/o that you have nothing.
Don't believe TMW has an implementation packaged as yet, anyway.
ADDENDUM:
You might check the File Exchange for submissions and of course there's always a search including Matlab in the search.
Antworten (2)
Vilnis Liepins
am 13 Mai 2014
0 Stimmen
Hello,
To convert nonuniform time domain data to frequency domain you can use Extended DFT (program NEDFT.m) available on fileexchange http://www.mathworks.com/matlabcentral/fileexchange/11020-extended-dft
NEDFT call line:
F=nedft(value,time,fn);
where frequencies [fn] could be generated as
fn=[-ceil((N-1)/2):floor((N-1)/2)]/N/T;
As I see from your example [time] data the mean sampling frequency T=1. The length of DFT N usually is set equal to or greater than the length(value) that, as Star Strider commented, should have the same dimensions as [time] array.
After that you also be able to get back constant time version of your data [value1] by applying IFFT to the result of NEDFT, call line:
value1=real(ifft(fftshift(F)));
1 Kommentar
Yogesh Babu
am 16 Mai 2014
Vilnis Liepins
am 19 Mai 2014
Bearbeitet: Vilnis Liepins
am 19 Mai 2014
0 Stimmen
Actually they are specified in your post:
>> value = [42007935 111212895 184546560 238219725 238219725 184546560 111212895 42007935];
>> time = [0, ,5,1.25,2, 2,5,5, 4, 4.5, 6, 7, 8 ];
Note that you need to correct the 'time' vector: time(1)<time(2)<time(3)<... and length(value) must be equal to length(time):
- the signal value(1)=42007935 is taken at time(1),
- the signal value(2)=111212895 is taken at time(2), and so on.
8 Kommentare
Yogesh Babu
am 19 Mai 2014
Vilnis Liepins
am 19 Mai 2014
Hello again! Input parameters for nedft.m should be specified or calculated outside the function. Example code:
value = [42007935 111212895 184546560 238219725 238219725 184546560 111212895 42007935];
time = [0.5 1.25 2 2.55 4 4.5 6 7];
N=length(value); T=1;
fn=[-ceil((N-1)/2):floor((N-1)/2)]/N/T;
F=nedft(value,time,fn);
value1=round(real(ifft(fftshift(F))));
time1=0:N-1;
plot(time,value,'-g',time1,value1,'-b');
Yogesh Babu
am 19 Mai 2014
Vilnis Liepins
am 19 Mai 2014
Bearbeitet: Vilnis Liepins
am 19 Mai 2014
Nedft.m is just am example script saved as text file and it was created on old Matlab version. I don't have Matlab installed anymore, but I recommend to do something like that: (1) create a new *.m file, (2) copy all the text from nedft.m and paste it into this new matlab file, (3) then save it as nedft.m in your Matlab workspace.
Yogesh Babu
am 21 Mai 2014
Vilnis Liepins
am 22 Mai 2014
Hello, Try to execute example code posted by me on 19 May 2014 at 11:15. Copy/paste all lines in Matlab window. In result you should get two plots in one window: nonuniform 'value' and uniform 'value1' get from the nedft.m output. Is it works?
Yogesh Babu
am 28 Mai 2014
dpb
am 28 Mai 2014
It's about four or five comments past...you only need to use the
"Show N older comments"
button to see the previous that have been collapsed by the viewer by default.
Kategorien
Mehr zu Multirate Signal Processing finden Sie in Hilfe-Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!