Non uniform sample in frequency analysis! non uniform dft?

Hallo! I am working in project that I have to make a frequency analysis on a time series. However my sample is not uniformly sampled so the Fast Fourier Transformation(FFT) is not appropriate for my analysis. After a search I found that for my case( no uniform sample) there is the Non Uniform Discrete Fourier Transformation (nuDFT). Does anyone know where can I find such a code or can suggest me any other way to make it?

[evelyn]

NUFFT might be the function.

Answer 1

[Star Strider]

If you have the Signal Processing Toolbox, one approach (that I would use) is to use the resample function.

It can convert non-uniformly-sampled data to uniformly-sampled data, and also uses an anti-aliasing filter. It is the preferred function for resampling signals for subsequent signal processing.

[Panos Ale]

Hmm sounds interesting but I need some more clarification. My non-uniformly-sampled data have inter-trial intervals with uniform distribution Mean=1s Max=1.5s and Min=0.5s . Given that the resample command is resample(x,p,q), what should ‘p’ and ‘q’ be in order to resample my data at a frequency of 1Hz( 1s interval)? Thanks for your answer :)

[Star Strider]

My pleasure.

You need to read down a bit in the documentation. See the sections describing:

• y = resample(x,tx) resamples the values, x, of a signal sampled at the instants specified in vector tx. The function interpolates x linearly onto a vector of uniformly spaced instants with the same endpoints and number of samples as tx. NaNs are treated as missing data and are ignored.
• y = resample(x,tx,fs) uses a polyphase antialiasing filter to resample the signal at the uniform sample rate specified in fs.
• y = resample(x,tx,fs,p,q) interpolates the input signal to a uniform grid with a sample spacing of (p/q)/fs and then filters the result to upsample it by p and downsample it by q. For best results, ensure that fs × q/p is at least twice as large as the highest frequency component of x.

There are a few more options as well.

[Greg Dionne]

You can also use plomb to get a Lomb-Scargle periodogram.

[Vilnis Liepins]

One more option is to use simple matlab function nedft.m available on fileexchange http://se.mathworks.com/matlabcentral/fileexchange/11020-extended-dft . It allows you to calculate the DFT directly from non-uniformly-sampled data without using of resample function which may introduce interpolating errors.