Dear Colleague,

In which language are you programming?

If you''ll conduct your study in Matlab®, you can compute the DFT using a fast-fourier-transform algorithm ("fft" function) as follows:

Y = fft(X,n)

-->where "X" must be a vector ( you'll include the signal to spectrally estimate here) 

--> where "n" is the n-point DFT (256 in your case)

For more information about the FFT function, visit this page:

http://www.mathworks.com/help/matlab/ref/fft.html

Best regards.

Dear Karim, not sure I understand your question. Why wanting to apply a 256 points DFT? Your spectral resolution migth be be roughly poor in that case. Moreover, using 256 points FFT means, for instance, analysing only 1 second of a time domain signal sampled at 256 samples/second.I remember that MIT-BIH database contains records like 30 minutes of ECG sampled around 300 samples/second.Usually, a high spectral resolution is wanted for a DFT/FFT analysis and this is obtained by techniques like "zero padding" in time domain signal if the signal has limited duration or limited number of samples.