Definition of FFT Analysis

 

FFT(Fast Fourier Transform) is a mathematical algorithm that takes a time-domain data and maps it into its frequency-domain.

 

Discrete Fourier Transform

Suppose that we have N consecutive sampled values

img86.gif

so that the sampling interval is img87.gif. To make things simpler, let us also suppose that N is even. With N numbers of input, we can estimate the Fourier transform img88.gif at the discrete values.

img89.gif

The extreme values of n correspond exactly to the lower and upper limits of the Nyquist critical frequency range.

The Fourier transform is approximated by a discrete sum. The discrete Fourier transform is obtained as,

img90.gif

If the number of sampled values is an even power of two (for example, 256, 512, 1024, and so on), the discrete Fourier transform can be computed in img91.gif  operations with an algorithm called the fast Fourier transform, or FFT.

 

     Magnitude

FFT/magnitude determines the magnitude (abs) of the complex value returned from the FFT algorithm.

 

     Phase

FFT/Phase determines the phase angle of the complex value returned from the standard FFT algorithm.

 

     Power Spectral Density (PSD)

FFT/Power Spectral Density determines the periodogram estimation of the complex value returned from the standard FFT algorithm. If we take an N point sample of the function c(t) at equal intervals and use the FFT to compute its discrete Fourier transform

img93.gif

Then the periodogram estimate is defined at N/2+1 frequencies as

img94.gif

where, img95.gif  is defined only for the zero and positive frequencies

img96.gif

 

     Window Method

The time-domain data is assumed a periodic sample from a continuous, infinite series of data for FFT. Window methods are used to reduce discontinuities from mismatching start and end and ensure periodicity of the FFT conditions.

The functions of window methods are defined below. In the equation, img97.gif is the function of the window method and N is the number of sampled values.

    Square Window

 

    Bartlett Widow

 

    Welch Window

 

    Hanning Window

 

    Hamming Window

 

    Blackman