Discrete Fourier Transform

1
 Introduction
• As the name implies, the Discrete Fourier Transform is purely
discrete: discrete time data sets are converted into a discrete
frequency representation

• Mathematically, The DFT of discrete time sequence x(n) is denoted

by X(k).It is given by,
N 1  j 2kn
X ( k )   x ( n)e N

n 0

Here k=0,1,2,….N-1
Since this summation is taken for ‘N’ points, it is called ‘N’ point
DFT.

2
 We can obtain discrete sequence x(n) from its DFT. It is called
as
Inverse Discrete Fourier Transform (IDFT).It is given by,
1 N 1 j 2kn
x ( n )   X ( k )e N
N k 0

Here n=0,1,2,…N-1
•This is called as N-point IDFT.

•Now we will define the new term ‘W’ as

WN=e-j2π/N
•This is called as “twiddle factor”.
•Twiddle factor makes the computation of DFT a bit easy and fast.
•Using twiddle factor we can write equation of DFT and IDFT as
N 1
follows:
X (k )   x(n)WNkn
n 0
Here k=0,1,2,….N-1
3
 N 1
1
 X (k )W
 kn
And x ( n)  N
N k 0

Here n=0,1,2,…N-1

Numericals using defintion of DFT:

1.Obtain DFT of unit impulse δ(n).

Solution:
•Here x(n)= δ(n) ……….(1)
•According to the definition of DFT ,we have
N 1  j 2kn
X ( k )   x ( n)e N
……..(2)
n 0

•But δ(n)=1 only at n=0.Thus equation (2) becomes,

X(k)= δ(0)e0 = 1
DFT
δ(n) 1
•This is the standard DFT pair. 4
 2.Obtain DFT of delayed unit impulse δ(n-n0)
Solution: We know that δ(n-n0) indicates unit impulse delayed
by ‘n0’samples.
Here x(n)= δ(n-n0) ……..(1)
N 1  j 2kn
Now we have X (k )   x(n)e N ……..(2)

n 0

But δ(n-n0) =1 only at n=n0.Thus eq.(2) becomes

X (k )  1.e  j 2kn0 / N

 (n  n0 )  e  j 2kn0 / N
Similarly, DFT of
 (n  n0 )  e j 2kn / N
0

5
 3.Find the 4 point DFT of following window function
w(n)=u(n)-u(n-N).
Solution:
According to the definition of DFT,
N 1  j 2kn
X ( k )   x ( n)e N
……..(1)
n 0

•The given equation is x(n)=w(n)=1 for0  n  N 1 . Here

N=4,so we will get 4 point DFT.
3  j 2kn
 X (k )  1.e 4
……..(2)
n 0

•The range of ‘k’ is from ‘0’ to ‘N-1’.So in this case ‘k’ will vary
from 0 to 3.
3 3
For k=0, X (0)  1.e  1  1  1  1  1  4
0

n 0 n 0

6
 3  j 2n
For k=1, X (1)   e 4

n 0

 X (1)  e 0  e  j 2 4  e  j 4 4  e  j 6 4

 2 2   4 4   6 6 
 X (1)  1   cos  j sin 
  cos  j sin 
  cos  j sin 
 4 4   4 4   4 4 

 X (1)  1  j  1  j  0
3  j 2  2 n 3
For k=2, X (2)   e 4
  e  jn
n 0 n 0

 X (2)  e 0  e  j  e  j 2  e  j 3

 X (2)  1  cos   j sin    cos 2  j sin 2   cos 3  j sin 3 

 X (2)  1   1  0  1  0   1  0  1  1  1  1  0
7
 3  j 2 3 n 3  j 6n
For k=3, X (3)   e 4
 e 4

n 0 n 0

 j 6  j 9
 j 3
 X (3)  e  e
0 4
e e 2

 6 6   9 9 
 X (3)  1   cos  j sin   cos 3  j sin 3    cos  j sin 
 4 4   2 2 
 X (3)  1  0  j    1  0  0  j   0

 X (k )  4,0,0,0

8
 Cyclic Property of Twiddle factor
(DFT as linear transformation matrix)

•The twiddle factor is denoted by WN and is given by,

WN=e-j2π/N ……….(1)
•Now the discrete time sequence x(n) can be denoted by x N .Here
‘N’ stands for ‘N’ point DFT.
•Range of ‘n’ is from 0 to ‘N-1’.
•x N can be represented in the matrix form as follows:
 x ( 0) 
 x(1) 
 
 x ( 2) 
xN   
 . 
. 
 
 x( N  1) N 1

•This is a “N×1” matrix and ‘n’ varies from ‘0’ to ‘N-1’

9
Now the DFT of x(n) is denoted by X(k).In the matrix form
X(k) can be represented as follows:
 X ( 0) 
 X (1) 
 
 X ( 2) 
XN   
 . 
. 
 
 X ( N  1) N 1
•This is also a “N×1” matrix and ‘k’ varies from ‘0’ to ‘N-1’
•From the definition of DFT,
N 1
X (k )   x(n)WNkn
n 0

10
 kn
•We can also representW N in the matrix form. Remember that
‘k’ varies from 0 to N-1 and ‘n’ also varies from 0 to N-1.

n=0 n=1 n=2 . . . n=N-1

 0 0 0 0

k=0 W 0N W N
1
W N
2
. . . W N
N 1 
k=1 W N W N W N
. . . WN 
k=2 W 0 W
2
W
4
. . .
2 N 1
W
 N N N N 
W N  .  .
kn
. . . 

.  . . . . 
.  . . . . 
 0 N 1 2  N 1  N 12 
k=N-1 W N W N W N
. . . W N 

•Note that each value is obtained by taking multiplication of k

and n. kn 4
•For example, if k=2, n=2 then we get W N = W N

11
Thus, DFT can be represented in the matrix form as

X N
 WN  xN

•Similarly IDFT can be represented in the matrix form as

1 *
xN  N WN  X N

•Here WN* is the complex conjugate of WN .

Periodicity property of Twiddle factor WN

•WN possesses the periodicity property.
•After N-points, it repeats its value

12
 Problem-1:Calculate the four point DFT of four point sequence
x(n)=(0,1,2,3).

Solution:The four point DFT in matrix form is given by

X  W  x
4 4 4

1 1 1 1  0 
1  j  1 j   1 
X 4
    
1  1 1  1   2 
   
1 j  1  j  3

 0 1 2  3   6 
0  j  2  3 j   2 j  2 
X 4
  
 0 1 2  3    2 
  


 0  j  2  3 j    2 j  2 

X 4
{6, 2j-2, -2, -2j-2}
13
 Properties of DFT
DFT
1-Linearity: If x1(n) X1(k) and x2(n) DFT X2(k) then,
N N
DFT
a1x1(n) + a2x2(n) a1X1(k) + a2X2(k)
N
Here a1and a2 are constants.

2-Periodicity:
If x(n) DFT X(k) then
N
x(n+N) = x(n) for all n
X(k+N) = X(k) for all k

14
15