Biomedical Digital Signal

Processing BDSP-513
Lecture 4

Discrete Fourier Transform

Dr. Mahbubunnabi Tamal

Overview
I. Discrete Fourier Transform and their properties

a) The algorithm transforming the time domain signal samples

to the frequency domain components is known as the
discrete Fourier transform, or DFT.

b) The DFT is widely used in many other areas, including

spectral analysis, acoustics, imaging/video, audio,
instrumentation, and communications systems.

II. The role of DFT algorithms to compute signal

amplitude spectrum and power spectrum.
Discrete Fourier Transform Formulas

N: number of samples.
The sampling period ;.
the sampling rate.
; the fundamental frequency
in radians per second
With ; the fundamental
frequency.
The frequency resolution:
/N
Discrete Fourier Transform Formulas
› constitutes the DFT coefficients using one-period data
samples :

The index is the time index representing the sample number of

the digital sequence.

is the frequency index indicating each calculated DFT

coefficient and can be further mapped to the corresponding
signal frequency in terms of Hz.
Discrete Fourier Transform Formulas

for

This equation can be expanded as:

++

for

where the factor is defined as:

Discrete Fourier Transform Formulas

The inverse DFT is given by:

for

This equation can be expanded as:

for
Matlab FFT Functions

› We can use MATLAB functions fft() and ifft() to compute the

DFT coefficients and the inverse DFT with the following syntax:

The following examples serve to illustrate the application of

DFT and the inverse of DFT.
Example I
This example serve to illustrate the application of DFT :

Given a sequence for , where and

a) Evaluate its DFT

Solution: Since N=4; for

You can also verify the result using the Matlab function
Example II
› This example serve to illustrate the application of the
inverse of DFT.
II - Given a sequence for , where X

a) Evaluate its IDFT to determine the time domain sequence

Solution: Since N=4  for

Frequency bin k & its associated frequency

› The calculated coefficients represent the frequency components ranging

from (or radians/second) to (or radians/second), hence we can map the
frequency bin to its corresponding frequency as follows:

The fundamental frequency in radians per second.

where =

› The frequency resolution or the frequency step between two consecutive

DFT coefficients :

/N
Example III
In Example I, given a sequence for , where , we have computed four DFT
coefficients for as . If the sampling rate is 10 Hz.

a) Determine the sampling period, time index, and sampling time instant
for a digital sample in time domain.

b) Determine the frequency resolution, frequency bin number, and mapped

frequency for each of the DFT coefficients and in frequency domain.
 Solution Example III
a.

b.
Amplitude spectrum and power spectrum
› We achieve the digital sequence by sampling the analog signal and
truncating the sampled signal with a data window with a length
where T is the sampling period and the number of data points.

› The magnitude and phase of each DFT coefficient (we refer to them
as the amplitude spectrum and phase spectrum, respectively) can be
determined and plotted versus its frequency index. We define the
amplitude spectrum as:
Amplitude spectrum and power spectrum

› Correspondingly, the phase spectrum is given by:

› Besides the amplitude spectrum, the power spectrum is also

used. The DFT power spectrum is defined as:
Example IV
› Consider the sequence:

Assuming that fs = 100 Hz,

a. Compute the amplitude spectrum, phase spectrum, and
power spectrum.
Solution: Example IV
Example V
Faster DFT computation
› DFT definition:
2nk N 1 2nk
N 1 1 j
 X [k ]e
j
X [k ]   x[n]e N x[n]  N

n 0
N n 0

› Requires N2 complex multiplies and N(N-1) complex additions

› Take advantage of the symmetry and periodicity of the complex exponential (let WN=e-
j2p/N
)
WNk [ N  n ]  WN kn  (WNkn )*
– symmetry: WNkn  WNk [ n  N ]  WN[ k  N ]n
– periodicity:

› Note that two length N/2 DFTs take less computation than one length N DFT:

2(N/2)2<N2

› Algorithms that exploit computational savings are collectively called Fast Fourier
Fast Fourier transform
› The FFT is simply a highly optimized implementation of the DFT. They both

produce identical results.

› The FFT algorithm requires the time domain sequence to have a length of data

points equal to a power of 2; that is, samples, where is a positive integer.

› For example, the number of samples in can be N=2,4,8,16, etc.

› If the length of the available data is not equal to a power of 2 (required by the

FFT), we can pad the data sequence with zeros to create a new sequence with a

larger number of samples, . The modified data sequence for applying FFT,

therefore, is:
Example VI
Fast Fourier transform
It is very important to note that the signal spectra obtained via zero-padding
the data sequence in Equation in slide 20 does not add any new information
and does not contain more accurate signal spectral presentation.
Method of Decimation-in-Time
› In this method, we split the input sequence into the even
indexed and each with data points. Then Equation becomes:

N 1
X [k ]   x[n]WNnk
n 0

  x[ n
n even
]W N 
nk
 x[
n odd
n ]W nk
N

N
2
1 N
2
1
  x[2r ](WN2 ) rk  WNk  x[2r  1](WN2 ) rk
r 0 r 0
N
2
1 N
2
1
  x[2r ]WNrk/ 2  WNk  x[2r  1]WNrk/ 2
r 0 r 0
Method of Decimation-in-Time
› Result is the sum of two length DFTs

X [k ]  G
[k ] W  H
k
N [k ]
N/2 DFT N/2 DFT
of even samples of odd samples

› Then repeat decomposition of to DFTs, etc.

x[0,2,4,6] N/2
DFT
X[0…7]

x[1,3,5,7] N/2
DFT
“Butterfly”
› Cross feed of and in flow diagram is called a “butterfly”, due
to shape

WNr
or simplify:

WN( r  N 2 ) WNr -1
( WNr )
8-point DFT Diagram

25
Example IX
Example X