Chapter-1

DFT ,FFT AND Z-TRANSFORM

Dr.Satyasis Mishra
Associate Professor
M-0904433042
DFT(Discrete Fourier Transform)
• Before going to DFT ,we have to go through DTFT First
DTFT : (DISCRETE TIME FOURIER TRANSFORM)
• Discrete Time Fourier Transform (DTFT) is Fourier transform of
discrete time sequence represented by complex exponential sequence
𝑒−𝑗𝜔𝑛 where 𝜔 is the real frequency variable.
• It maps time domain sequence into a continuous and periodic function
of frequency variable.
DFT Pair
DFT

IDFT

 DFT computational complexity:

Each DFT coefficient requires
 N complex multiplications;
 N-1 complex additions
 All N DFT coefficients require
 N2 complex multiplications;
N(N-1) complex additions
Properties of DFT

11. Symmetry property

X ( N  K )  X  K 
TWIDDLE FACTOR
This property of twiddle factor is referred as Cyclic Property.
The same can be graphically represented in following Fig. 1

This periodicity if for N  8. The same can be shown in generalized mathematical formula.
8 Re sidue P / 8
W8  W8
P mod ulo 8

Where, W8  Exp   j2 / 8  e  j2  / 8

PROBLEMS ON DFT AND IDFT
Problem 1 : Compute DFT of Sequence
x n    1, 0, 0, 1, 

Solution : DFT equation is
N 1  j2 kn

Xk    xn e N

n 0
N-1 = 3, N = 4 therefore N point DFT therefore k = 0, 1,2, 3
3  j2 kn

Xk    xn e N

n 0
 j4 k
 Xk   x0e n  x1e  j2 k / N  x2e N
 x3e  j6 k / N
j 3
 j k
Xk   x0  x1e  x2e  jk  x3e
2 2

k0 X0  x0  x1e0  x2e0  x3e0

 1 0  0 1  2
 3
j 
k 1 X1  x0  x1e  j / 2  x2e  j  x3e 2

3
j 
 1  0  0  1 e  1  e  j3 / 2
2

3 3
 1  cos  j sin  1 j
2 2
k2 X2  x0  x1e  j  x2e  j2  x3e  j3
 1  0  0  1 e  j3  1  cos 3  j sin 3
 1 1  0
3  j9 
j
k 3 X3  x0  x1e  x2e  j3  x3e  j3  x3e
2 2

 1  0  0  1 e  j9 / 2  1  cos 9 / 2  j sin 9 / 2

X3  1 j
 Xk    2, 1  j, 0, 1  j 
PROBLEM-2
N 1
1 N1
Show that  x n    Xk  .
2 2

n 0 N k 0
Where Xk , 0  k  N  1, is DFT of sequence xn 0  n  N  1.
Solution:
N 1

 xn 
2
can be written as
n 0
N 1

 x n x 
n 
n 0
Use IDFT formula (synethisis equation)
N 1 j2 kn
x n    X  k e
 1 N
N n 0
 N 1 N 1

 xn    x n x  n 
2

n 0 n 0
N 1  1 N 1  j2 kn

  x n   X k e N 
n 0  N k 0 
1 N 1  N 1 j2 kn

  X k   x n e
 N

N k 0  n 0 
1 N1 
  X k Xk 
N k 0
1 N1
  Xk 
2

N k 0
NOTE:
The way we introduced W (twiddle factor) DFT equation, we can also introduce the same
IDFT equation.
N 1 j2 kn

xn   Xk e
1
IDFT   N
N k 0
n  0, 1,....N  1
1 N1  j2   kn 
xn    xk e   
N k 0  N
 j2 
as W e
1 N1
xn    Xk WN n  0, 1,.......N  1
kn

N k 0
DFT as Linear Transformation:
 We have

 x 0   X0 
 x 1   X1 
   
 x 2   X2 
   
xN  . X
 N   . 
.  . 
   
.  . 
.x N  1  XN  1
   
 WN0 WN0 WN0 ........... WN0 
 0 1 N 1

 WN WN WN2 ........... WN 
 0 
 WN WN2 WN4 ......... WN2 N 1 
WN  . . . .......... 
 
. . . .......... 
. . . ......... 
 
 WN0 WNN 1 WN2 N 1 .......... WN  N 1 N 1 
 
N  1N  1  Matrix
Thus we can represent N point DFT in matrix form.
XN  WN x N
Where, xN  Input sequence matrix
WN  Matrix of linear transformation
XN  DFT output points.
We observe that WN is symmetric matrix.
To get IDFT equation multiply both the sides by WN1
 X N WN1  WN x N WN1
So xN  X N WN1
Finally for IDFT,
1 
xN  WN X N
N
1 
Where, WN1  WN
N
Problem 3: Compute 4 point DFT of given sequence  1, 0, 0, 1.Use matrix for
DFT computation.
PROBLEM-4
Circular Convolution Using DFT and
IDFT method
To find circular convolution i.e. computing ym  by means of DFT and IDFT formula.
Procedure is as follows :
Step I :
Find DFT of x1 n i.e.X1 k 
Step II :
Find DFT of x 2 n i..e.X 2 k 
Step III :
Multiply both DFT i.e. Y k   X 1 k X 2 k 
Step IV :
Find IDFT of X 3 k . After this we will get elements ym.
PROPERTIES OF TWIDDLE FACTOR
Decimation in time FFT(Fast Fourier Transform)
BIT REVERSAL

FFT
Refer following table 1.
B2 B1 B0  B0 B2 B1  B0 B1 B2
0  0 0 0 0 0 0 0 0 0  0
1  0 0 1 1 0 0 1 0 0  4
2  0 1 0 0 0 1 0 1 0  2
3  0 1 1 1 0 1 1 1 0  6
4  1 0 0 0 1 0 0 0 1  1
5  1 0 1 1 1 0 1 0 1  5
6  1 1 0 0 1 1 0 1 1  3
7  1 1 1 1 1 1 1 1 1  7
___________________ ___________________
Natural order bit reversal

If you refer table 1, you find that instead of taking B 2 B1 B 0 if we write B0 , B1 , B 2 we get
reversed order. But actually in computer we have to change the positions of bits. For that we use
following flow chart.
Problems on DIT-FFT
Given xn  sequence as follows x n   1, 2, 2, 1 , Find corresponding 8-point DFT Xk 
using DITFFT.
Solution :
(1) We have to find DFT using DIT-FFT chart.
(2) For convenience we take N = 8,  N 1  7
 x n    1, 2, 2, 1, 0, 0, 0, 0 
(3) 
n0 ..........7
(4) Diagram. Refer fig. below
Fig. 5 shows input values of xn  .
Input to the stage 1 will be
A 1 E2
B0 F0
C2 G 1
D0 H0
 
As 80  e  j2 0 / 8  e  j0  e 0  1 ,
(5) Output of stage 1.
O11  A  W80 B  1  0  1 O12  A  W80 B  1  0  1
O13  C  W80 D  2  0  2 O14  C  W80 D  2  0  2
O15  E  W80 F  2  0  2 O16  E  W80 F  2  0  2
O17  G  W80 H  1  0  1 O18  G  W80 H  1  0  1
(6) Input to stage 2
A 1  O11  1 B1  O12  1
C1  O13  W80  2 1  2 D1
 O14  W82  2   j  2 j
E1  O15  2 F1  O16  2
G 1  O17  W80  11  1 H1  O18  W82  1  j   j
(7) Output of stage 2
O 21  A1  C1  1  2  3 O 22  B1  D1  1  2 j
O 23  A1  C1  1  2  1 O 24  B1  D1  1  2 j
 O 25  E1  G 1  2  1  3 O 26  F1  H1  2  j
O 27  E1  G 1  2  1  1 O 28  F1  H1  2  j
(8) Input to stage 3.
A 2  O 21  3 B2  O 22  1  2 j
C 2  O 23  1 D2  O 24  1  2 j
E 2  O 25  W80  3 1  3
F2  O 26  W81  2  j  0.707  j 0.707   1 / 2  j3 / 2
G 2  O 27  W82  1  j   j

H 2  O 28  W83  2  j   0.707  j 0.707  

1 3
j
2 2
(9) Output of stage 3 i.e.
X0  A2  E2  3  3  6

X1  B 2  F2  1  2 j 
1 j3

2 2
 1   3   2 1  2 2  3 
 1  
 j 2       j
 

 2  2  2   2 
2  2 1 2  2 2  3 2  2 43 2 
   j    j 
2 2  2 2    
2  2 
X2  C 2  G 2   1  j  1  j
  1 j3 
X3  D2  H2  1  2 j    
 2 2
2 2  43 2 
  j 

2  2 
X4  A2  E2  3  3  0
j3  2  2   4  3 2 
X5  B2  F2  1  2 j 
1
     j 
2 2  2   2 
 
X6  C 2  G 2   1   j  1  j
 1 j3  2  2  4  3 2 
X7  D2  H2  1  2 j       j 

 2 2 2  2 
DIF-FFT(Decimation in Frequency)
4-point DIT-FFT
4-point DIF-FFT
 CHAPTER-1
Z-Transform
Dr.Satyasis Mishra
Associate Professor
M-0904433042
Z-Transform
• Definition:

X (z )   x (n )z n
n  

• z is a complex variable: imaginary

z
r

z  re j
 r cos   jr sin  real

01-Oct, 98 EE421, Lecture 7 55

The z-Transform and DTFT
• The z-Transform is often time more convenient to use
• Definition:

Xz    xn z n

n  

• Compare to DTFT definition:

    xn e

j  j n
Xe
n  

• z is a complex variable that can be represented as z=r ej

• Substituting z=ej will reduce the z-transform to DTFT

56
The Z-Transform and unit circle
• The z-transform is a function of the complex z variable
• Convenient to describe on the complex z-plane
• If we plot z=ej for =0 to 2 we get the unit circle
Im  
X e j

Unit Circle

r=1
 0
Re 
2 0 2

57
Properties of the Z-Transform
Summary

• Definition of the z-Transform: X(z)   x[n]z
n  
n

• Properties of the z-Transform:

 Linearity: ax1[n]  bx2 [n]  aX1[ z]  bX 2 [ z]
 Time-shift: x[ n  n0 ]  z  n0 X [ z ]
dX [ z ]
 Multiplication by n: nx[n]   z
dz
PROBLEM-1

By applying the time-shifting property, determine the z-transform of the signal.

z 1
Xz  
1  3z 1
Solution: Given that
z 1
Xz   1
 z 1
X1 z 
1  3z
1
where X1 z  
1  3z 1
Here, from the time shifting property, we have K  1 and
xn   3 .un 
n

xn   3 . u n  1
n 1
Ans.
PROBLEM-2
Compute the convolution x n  of the signals
x 1 n   4,  2, 1
1 for 0  n  5
x 2 n   
0 for otherwis
Solution : The z-transforms for the given signals can be written as
X1 z   4  2z  z
1 2

X 2 z   1  z
1 2 3 4 5
z z z z
Therefore, we have
Xz   X1 z X 2 z   4  2z
1 2 3 4 5 6 7
 3z  3z  3z  3z z z
Taking inverse z-transform of above equation, we get
4, 2, 3, 3, 3, 3,  1, 1
x n     Ans.
 
STABILITY IN Z-TRANSFORM