UNIT-2

DISCRETE
FOURIER SERIES
AND TRANSFORM
Dr. D.MOHAN

Department of Electronics and Computer Engineering,

SNIST-HYD

1
 UNIT II
Syllabus

DISCRETE FOURIER SERIES AND TRANSFORM:

Discrete Fourier series representation of periodic sequences,

Discrete Fourier transforms: Properties of DFT, linear and
circular convolutions, Overlap add and Overlap save
methods, Computation of DFT. Relation between Z-transform
and DFT.

Applications: Analysis of DT signals-Periodic and Aperiodic.

2
 Fourier Representation of
Signals and LTI Systems.
1 Introduction.
2 Complex Sinusoids and Frequency Response of LTI Systems.
3 Fourier Representation of Four Classes of Signals.
4 Discrete Time Periodic Signals: Discrete Time Fourier Series.
5 Continuous-Time Periodic Signals: Fourier Series.
6 Discrete-Time Non Periodic Signals: Discrete-Time Fourier
Transform.
7 Continuous-Time Non Periodic Signals: Fourier Transform.
8 Properties of Fourier Representation.
9 Linearity and Symmetry Properties.
10 Convolution Properties.
3
 1 Introduction.
 Signals are represented as superposition's of complex sinusoids
which leads to a useful expression for the system output and
provide a characterization of signals and systems.
 Example in music, the orchestra is a superposition of sounds
generated by different equipment having different frequency
range such as string, base, violin and etc. The same example
applied the choir team.
 Study of signals and systems using sinusoidal representation is
termed as Fourier Analysis introduced by Joseph Fourier
(1768-1830).
 There are four distinct Fourier representations, each applicable
to different class of signals.

4
. Complex Sinusoidal and Frequency
Response of LTI System.

 The response of an LTI system to a sinusoidal input that lead to

the characterization of the system behavior that is called
frequency response of the system.
 An impulse response h[n] and the complex sinusoidal input
x[n]=ejn.

5
Cont’d…
Discrete-Time (DT)

Derivation: y n    h k x n  k 
k  

y n    h k e j  n  k 

k  

factor e jn out


y  n   e j n  h k e  jk

k  

 
 H e j e jn

 Where, the frequency response:

   h k e

H e j   jk

k  
6
Cont’d…

y t    h   e j  t  
d
Continuous-Time (CT) 
Derivation: 
e j t
 h    e  j
d


 H  j  e jt


Frequency Response: H  j    h   e  jt
d


 The output of a complex sinusoidal input to an LTI system is a

complex sinusoid of the same frequency as the input, multiplied
by the frequency response of the system. 7
Example 1: Frequency Response. 1  RC
t

The impulse response of a system is given as h t   RC e u  t 

Find the expression for the frequency response, and plot the
magnitude and phase response.  
1 

Solution: H  j     e RC
u    e  j
d
RC 
Step 1: Find frequency response, H(j). 1
 
  j 
1 


RC 0
 e  RC 
d
Substitute h(t) into equation below,


  1 
1 1   j  

H  j        RC 
 j t e
h  e d RC  1 
 j  
  RC  0

1 1
  0  1
RC  1 
 j  
 RC 
1
 RC
 1 
 j  
 RC  8
Cont’d…
Step 2: From frequency response, get the magnitude & phase
response.
The magnitude response is,
1
| H  j  | RC
2
 1 
2   
 RC 

9
Cont’d…

Figure 3.1: Frequency response of the RC circuit (a) Magnitude response.

(b) Phase response.

The phase response is arg{H(j)}= -arctan(RC)

10
 .
3. Fourier Representation of
Four Class of Signals.
 There are four distinct Fourier representations,
representations class of signals;
(i) Periodic Signals.
(ii) Non periodic Signals.
(iii) Discrete-Time Periodic Signals.
(iv) Continuous-Time Periodic Signals.

 Fourier Series (FS) applies to continuous-time periodic signals.

 Discrete Time Fourier Series (DTFS) applies to discrete-time
periodic signals.
 Non-periodic signals have Fourier Transform (FT)
representation.
 Discrete Time Fourier Transform (DTFT) applies to a signal that
is discrete in time and non-periodic.
11
Cont’d…

Time
Property Periodic Nonperiodic

Continuous (t) Fourier Series (FS) Fourier Transform (FT)

[Chapter 3.5] [Chapter 3.7]

Discrete (n) Discrete Time Fourier Discrete Time Fourier

Series (DTFS) Transform (DTFT)
[Chapter 3.4] [Chapter 3.6]

Table 3.1: Relation between Time Properties of a Signal an the

Appropriate Fourier Representation.

12
Cont’d…

(1)
(1)Discrete
Discrete
Time
Fourier
Fourier
TimePeriodic
Periodic Representatio
Signal
Signal Representatio
nn

(2)
(2)Continuous
Continuous
Time
TimePeriodic
Periodic
Signal
Signal

(3)
(3)Continuous
Continuous
Time
TimeNon
Non
Periodic
PeriodicSignal
Signal

(4)
(4)Discrete
Discrete
Time
TimeNon
Non
Periodic
PeriodicSignal
Signal13
3.1 Periodic Signals:
Fourier Series Representation
 If x[n] is a discrete-time signal with fundamental period N then
x[n] of DTFS is represents as,

ˆx n   A k e jko n
k

 where o= 2/N is the fundamental frequency of x[n].

 The frequency of the kth sinusoid in the superposition is ko
14
Cont’d…
 If x(t) is a continuous-time signal with fundamental period T,
x(t) of FS is represents as

xˆ  t    A k e jko t

 where = 2/T is the fundamental frequency of x(t). The

frequency of the kth sinusoid is k and each sinusoid has a
common period T.
 A sinusoid whose frequency is an integer multiple of a
fundamental frequency is said to be a harmonic of a sinusoid at
the fundamental frequency. For example ejkt is the kth harmonic
of ejt.
15
 The variable k indexes the frequency of the sinusoids, A[k] is the
3.2 Non-periodic Signals:
Fourier-Transform
Representation.
 The Fourier Transform representations employ complex
sinusoids having a continuum of frequencies.
 The signal is represented as a weighted integral of complex
sinusoids where the variable of integration is the sinusoid’s
frequency.
 Continuous-time sinusoids are used to represent continuous
signal in FT.

1
xˆ  t    X  j  e j t
d
2 

where X(j)/(2) is the “weight” or coefficient applied to a

sinusoid of frequency  in the FT representation
16
Cont’d…

 Discrete-time sinusoids are used to represent discrete time signal

in DTFT.
 It is unique only a 2 interval of frequency. The sinusoidal
frequencies are within 2 interval.

 X  e e
1
xˆ n  j jn
d
2 

17
3.4 Discrete Time Periodic Signals:
Discrete Time Fourier Series.
 The Discrete Time Fourier Series representation;
N 1
x n   X  k e jk o n
k 0

 where x[n] is a periodic signal with period N and fundamental

frequency =2/N./N X[k] is the DTFS coefficient of signal x[n].
N 1
1
X k   x  n e  jk o n

N n 0

 The relationship of the above equation,

x n   X  k 
DTFS ; o
18
Cont’d…

 Given N value of x[n] we can find X[k].

 Given N value of X[k] we can find x[n] vise-versa.
 The X[k] is the frequency-domain representation of x[n].
 DTFS is the only Fourier representation that can be numerically
evaluated using computer, e.g. used in numerical signal
analysis.

19
Example 3.1: Determining DTFS Coefficients.
Find the frequency-domain representation of the signal in Figure
3.2 below.

Figure 3.2: Time Domain Signal.

Solution:
Step 1: Determine N and .
The signal has period N=5, so =2/5.
Also the signal has odd symmetry, so we sum over n = -2 to n = 2
from equation

20
Cont’d…

Step 2: Solve for the frequency-domain, X[k].

From step 1, we found the fundamental frequency, N =5, and we
sum over n = -2 to n = 2 .

1 N 1
X  k    x ne  jko n
N n0
1 2
X  k    x ne  jk 2n / 5
5 n 2

  x  2 e jk 4 / 5  x  1e jk 2 / 5  x 0 e j 0  x1 e  jk 2 / 5  x 2e  jk 4 / 5 

1
5

21
Cont’d…
From the value of x{n} we get,

1  1 jk 2 / 5 1  jk 2 / 5 
X  k   1  e  e 
5 2 2 
1
 1  j sin  k 2 / 5
5
Step 3: Plot the magnitude and phase of DTFS.
From the equation, one period of the DTFS coefficient X[k], k=-2 to
k=2, in the rectangular and polar coordinate as

1 sin  4 / 5
X   2   j  0.232e  j 0.531
5 5
1 sin  2 / 5
X   1   j  0.276e  j 0.760
5 5 22
Cont’d…
1
X  0   0.2e j 0
5
1 sin  2 / 5
X 1   j  0.276e j 0.760
5 5
1 sin  4 / 5
X  2   j  0.232e j 0.531
5 5
The above figure shows the magnitude and phase of X[k] as a
function of frequency index k.

Figure 3.3: Magnitude and phase of the DTFS coefficients for the signal in Fig. 23
3.2.
Cont’d…
Just to Compare, for different range of n.
Calculate X[k] using n=0 to n=4 for the limit of the sum.

 
X  k    x  0 e j 0  1 e  j 2 / 5   2 e  jk 4 / 5   3e  jk 6 / 5   4e  jk 8 / 5 
1
5
1 1 1 
X  k   1  e  jk 2 / 5  e  jk 8 / 5 
5 2 2 
This expression is different from which we obtain from using n=-2 to
n=2. Note that,

e  jk 8 / 5  e  jk 2 e jk 2 / 5
 e jk 2 / 5
n=-2 to n=2 and n= 0 to n=4, yield equivalent expression for the
DTFS coefficient.
24
.
3.5 Continuous-Time Periodic
Signals: Fourier Series.
 Continues time periodic signals are represented by Fourier series
(FS). The fundamental period is T and fundamental frequency
o =2/T. 
x t    X ( k ) e jko t

k  

 X[k] is the FS coefficient of signal x(t).

T
1
X (k )   x t e  jkot dt
T 0

 Below is the relationship of the above equation,

x t   X (k )
FS ;o
25
Cont’d…
 The FS coefficient X(k) is known as frequency-domain
representation of x(t).

26
Example 3.2: Direct Calculation of FS Coefficients.
Determine the FS coefficients for the signal x(t) depicted in Figure
3.4.

Solution: Figure 3.4: Time Domain Signal.

Step 1: Determine T and 0.
The period of x(t) is T=2, so 0=2/2 = On the interval 0<=t<=2,
one period of x(t) is expressed as x(t)=e-2t, so it yields

2
Step 2: Solve for X[k]. 1  2t  jkt
1
T X (k )   e e dt
X (k )   x t e  jko dt 20
T 0
2
1  2 jk  t
X (k )   e dt
20 27
Cont’d…
Step 3: Plot the magnitude and phase spectrum.
2
1
X (k )  e  2 jk  t
2(2  jk ) 0

1
X (k ) 
4  2 jk

e  4 e  jk 2  1 
1  e 4
X (k ) 
4  jk 2

- From the above equation for example k=0,

X(k)= (1-e-4)/4 = 0.245.

- Since e-jk2=1 Figure 3.5 shows the magnitude spectrum|X(k)| and

the phase spectrum arg{X(k)}. 28
Cont’d…

Figure 3.5: Magnitude and Phase Spectra..

Spectra.
29
3.6 Discrete-Time Non Periodic Signals:
Discrete-Time Fourier Transform.
 The DTFT representation of time domain signal,

x n 
1
2  X e  
j
e jn
d


 X[k] is the DTFT of the signal x[n].

 

X e j   x ne  jn


 Below is the relationship of the above equation,

x n  X e
DTFT
 
j
30
Cont’d…
 The FT, X[j], is known as frequency-domain representation
of the signal x(t).


 X  e e
1
x n  j jn
d
2 

 The above equation is the inverse FT, where it map the

frequency domain representation X[j] back to time domain.

31
Cont’d… Table 3.2 : Fourier Representation
Time
Domain Periodic Non periodic
(t,n) (t,n)
C N
O Fourier Transform O
N N

1
   
T P


j t
I x t 
Fourier Transform2 X j  e dw E
N
Fourier Series 
R
U I


1
x t    X  k e jk0t x t    
O O

jt 
X j  e dw
  j   x t e  jt dt

U D
S k   2 X I
(t) T C
1
X k 
 
 
T 0
 jk 0t
x t e dt
X  j   x t e Discrete
(k,w)


 j  t
dt Time Fourier Transform P
Discrete Time Fourier Series
x t  has period  
X  e j e jnTransform
D 1 E
Discrete x nTime  Fourier
2 
 d
I N 1 R
x n   X 2k
e jk0 n 
x n   X e X  e xe ne d
S 1 j j jnjn I
C 0 O
R
k 0 t 2  j  D
1 N 1 X  e  has period 2
X  k    x ne
E  jk 0 n I

X  e    x ne  jn

T j C
E
N n 0 (k,)
x n and X  k  have period N


X  e j  has period 2
(t)

2
0 
n
Table 3.2: The Four Fourier Representations.
Discrete (k) Continuous () Freq.
32
Domain
3.9 Linearity and Symmetry
Properties.
 All four Fourier representations involve linear operations.
FT
z  t   ax t   by  t   Z  j   aX  j   bY  j 
FS ; o  Z  k   aX  k   bY  k 
z  t   ax t   by  t  

DTFT
 Z  e j   aX  e j   bY  e j 
     
z  n   ax n   by n  
     
DTFS;
z  n   ax n   by n  
o  Z  k   aX  k   bY  k 

 The linearity property is used to find Fourier representations of

signals that are constructed as sums of signals whose
representation are already known.
33
Example 3.5: Linearity in the Fourier Series.
Given z(t) which is the periodic signal. Use the linearity property to
determine the FS coefficients Z[k].
3 1
z  t   x t   y  t 
Solution: 2 2
From Example 3.31(Text) we have
1  k 
x t   X  k  
FS ; 2
Sin 
k  4 
1  k 
y  t   Y  k  
FS ; 2
Sin 
k  2 
The linearity property implies that
3  k  1  k 
z  t    Z  k  
FS ; 2
Sin   Sin  
2k  4  2k  2 
.
34
3.9.1 Symmetry Properties:
Real and Imaginary Signal.
Representation Real-Valued Imaginary-Valued
Time Signal Time Signal

FT X*(j) = X(-j) X*(j) = -X(-j)

FS X*[k] = X[-k] X*[k] = -X[-k]

DTFT X*(ej) = X(e-j) X*(ej) = -X(e-j)

DTFS X*[k] = X[-k] X*[k] = -X[-k]

Table 3.3: Symmetry Properties for Fourier Representation of

Real-and Imaginary–Valued Time Signals.
35
3.9.2 Symmetry Properties:
Even and Odd Signal.

X *
 j    x  t e  j ( t ) dt


 Change of variable = -t.


X *  j    x   e  j
d


 X  j 

 The only way that the condition X*(j) =X(j) holds is for the
imaginary part of X(j) to be zero.

36
Cont’d…

 If time signal is real and even, then the frequency-domain

representation of it also real.

 If time signal is real and odd, then the frequency-domain

representation is imaginary.

37
3.10 Convolution Property.
 The convolution of the signals in the time domain transforms
to multiplication of their respective Fourier representations in
the frequency domain.
 The convolution property is a consequences of complex
sinusoids being eigen functions of LTI system.
3.10.1 Convolution of Non periodic Signals.
3.10.2 Filtering.
3.10.3 Convolution of Periodic Signals.

38
0.1 Convolution of Non Periodic Signal
 Convolution of two non periodic continuous-time signals x(t)
and h(t) is defines as
y  t   h t  * x  t 




h x t    d .

 express x(t-) in term of FT:


1
x t      X  j   e j  t  
d
2  39
Cont’d…

Substitute into convolution integral yields,


 1 

y  t    h     X  j  e e d d
jt  j

  2  
1  
 

   h  e d  X  j  e d
 j jt

2  

1
y t    H  j   X  j   e jt
d
2 

40
Cont’d…
 Convolution of h(t) and x(t) in the time domain corresponds to
multiplication of Fourier transforms, H(j) and X(j) in the
frequency domain that is;

FT
y  t   h t  * x  t   Y  j   X  j  H  j .

 Discrete-time nonperiodic signals, if

x n  X e 
  DTFT
 j
and
h n   H  e j 
DTFT
then
y n   h n * x n    Y  e j   X  e j  H  e j .
DTFT

41
Example 3.6: Convolution in Frequency Domain.
Let x(t)= (1/(t))sin(t) be the input of the system with impulse
response h(t)= (1/(t))sin(2t). Find the output y(t)=x(t)*h(t)
Solution:
From Example 3.26 (text, Simon) we have,
1,   
x t   X  j   
FT

0,   
1,   2
h t   H  j   
FT

0,   2
Since
y  t   x  t  * h t  
FT
Y  j   X  j   H  j  
it follows that
1,   
Y  j    
0,    

We conclude that 1

y  t     sin  t .
.  t  42
DTFT and DFT

43
DTFT and DFT

44
DTFT and DFT

45
DTFT and DFT

46
DTFT and DFT

47
DTFT and DFT

48
DTFT and DFT

49
Periodicity of W kn and values of WN
N

50
DTFT and DFT

51
DTFT and DFT

52
DTFT and DFT

53
 DTFT and DFT
 The DTFT of an aperiodic discrete time signal is
defined as 
X [ w]   x[n]e
 jwn

n  
(1)

 The DFT of a signal is defined as

N 1
X(k )   x[n]e  jk 2 n / N

n0
(2)
 Inverse DFT is defined as
N 1
1
x[ n] 
N
 X [k ]e
k 0
j 2kn / N
(3)

What is difference between DTFT and DFT?

54
 DTFT and DFT
 The DFT is periodic with period N.
Proof: X[k ]   x[n]e  jk 2n / N
N 1

n0
N 1
X[k  N]   x[n]e  j( k  N ) 2 n / N
n0
N 1
  x[n ]e  jk 2 n / N e  j2 n
n0

Since e-j2n = 1
N 1
 X[k  N]   x[n]e  jk 2 n / N
n 0

 X[k ]
proved
55
Example 1:
Find the DFT of the following sequence [1 0 0 1]

N 1 3 3
X[k ]   x[n]e  jk 2 n / N   x[n]e  jk 2 n / 4   x[n]e  jkn / 2
n0 n0 n0
3
X[0]   x[n]  x[0]  x[1]  x[2]  x[3]  1  0  0  1  2
i 0
3
X[1]   x[n]e  jkn / 2  x[0]  0  0  x[3]e  j3  / 2
n0
 j3  / 2
 1  1.e  1  cos( )  j sin( )  1  j
3
2
3
2
3
X [2]   x[n]e  jn  x[0]  x[3]e  j 3  1  1.[cos( 3 )  j sin  3  ]  0
n 0

56
Circular Convolution

57
Example for Circular Convolution

58
Example for Circular Convolution

59
Example for Circular Convolution

60
Example for Circular Convolution

61
Example for Circular Convolution

62
Example for Circular Convolution

63
Example for Circular Convolution

64
Example for Circular Convolution

65
Example for Circular Convolution

66
Example for Circular Convolution

67
Example for Circular Convolution

68
Example for Circular Convolution

69
Example for Circular Convolution

70
Example for Circular Convolution

71
Example for Circular Convolution

72
Example for Circular Convolution

73
Example for Circular Convolution

74
Example for Circular Convolution

75
 DFT Sectional Convolution
Suppose, the input sequence x(n) of long duration is to be processed with a system having
finite duration impulse response by convolving the two sequences. Since, the linear
filtering performed via DFT involves operation on a fixed size data block, the input
sequence is divided into different fixed size data block before processing.

The successive blocks are then processed one at a time and the results are combined to
produce the net result.

As the convolution is performed by dividing the long input sequence into different fixed
size sections, it is called sectioned convolution.

A long input sequence is segmented to fixed size blocks, prior to FIR filter processing.

Two methods are used to evaluate the discrete convolution −

1.1.Overlap-save method

2.2.Overlap-add method
76
 Overlap save method
Overlap–save is the traditional name for an efficient way to evaluate the discrete
convolution between a very long signal x(n) and a finite impulse response FIR filter h(n).
Given below are the steps of Overlap save method −

Let the length of input data block = N = L+M-1. Therefore, DFT and IDFT length = N.
Each data block carries M-1 data points of previous block followed by L new data points
to form a data sequence of length N = L+M-1.

77
Overlap save method

78
Overlap add method

79
Overlap add method

80
Overlap add and Overlap Save method

81
Overlap add and Overlap Save method

82
Overlap add and Overlap Save method

83
Overlap add and Overlap Save method

84
Overlap add and Overlap Save method

85
Overlap add and Overlap Save method

86
Overlap add and Overlap Save method

87
Overlap add and Overlap Save method

88