Fourier Analysis

Download as pdf or txt
Download as pdf or txt
You are on page 1of 37

KING SAUD UNIVERSITY

ELECTRICAL ENGINEERING DEPARTMENT.

EE 301: Signals and Systems

Chapter 4:
Fourier Analysis for CT and DT Signals

Dr A. I. Sulyman
Assistant Professor
EE Department
King Saud University
Office: College of Eng. 2C 12/1 (Tel: x70586)
***********
********
Course Instructor: Dr A. I. Sulyman

Outline

1. Introduction: Response of LTI Systems to complex Exponentials

2. Fourier Series of Periodic Signals: CTFS

3. DT FS

4. CT FT

5. DT FT

6. Properties of Fourier representations

7. Sampling Theorem

2
Course Instructor: Dr A. I. Sulyman

1 - Introduction: Response of LTI Systems to complex Exponentials


LTI system

Input x(t)  est h (t) Output y (t)


s  j (complex variable)

 Output of CT LTI system is given by:

y (t )  h(t ) * x(t )


  h( ) x(t   ) d


 h( )e
s ( t  )
 d


 e st  h( )e  s d


 h( )e
 s
 y (t )  H ( s )e st , where H ( s )  d is the Laplace transform of h(t ) .


 Output of DT LTI system is given by:

LTI system

x(n)  zn h (n) y (n)


z  e j (complex variable)  z n  e jn

y ( n)  h( n) * x ( n)


  h( k ) x ( n  k )
k  


  h( k ) z
k  
nk


 zn  h( k ) z
k  
k

3
Course Instructor: Dr A. I. Sulyman


 y (n)  H ( z ) z n , where H ( z )   h(k ) z
k  
k
is the Z-transform of h(n) .

Summary:
LTI
CT: x(t )  e st  y (t )  H ( s )e st

LTI
DT: x ( n )  z n  e j n  y ( n)  H ( z ) z n

 Thus for both DT and CT cases, the response of an LTI system to a complex
exponential input is the same complex exponential with a scaling of its amplitude by a
constant factor { H (s ) , H (z ) }.

 Sum of complex exponentials:

Using the additivity and homogeneity properties of LTI systems the following follows
up from the basic results above:

LTI
CT: x(t )   a k e sk t  y (t )   a k H ( s k )e sk t
k k

LTI
DT: x(n)   a k z kn  y (n)   a k H ( z k ) z kn
k k

 The representation for x(t) and x(n) as sum of complex exponentials above are called
Fourier Series representations.

 We will consider four distinct Fourier representations:

 (i) Continuous-time Fourier Series (CTFS)

 (ii) Discrete-time Fourier Series (DTFS)

 (iii) Continuous-time Fourier Transform (CTFT)

 (iv) Discrete-time Fourier Transform (DTFT)

4
Course Instructor: Dr A. I. Sulyman

2 - Fourier Series of Periodic Signals


Continuous-time Fourier Series (CTFS):

 Consider a periodic CT signal x(t ) with fundamental period T0 and fundamental


frequency  0  2 / T0 .

 x(t ) can be decomposed as an exponential series called Fourier Series:


x(t )  a e
k  
k
jk0 t
(Synthesis) ---- (I)

where the coefficients a k are called the Fourier coefficients, and defined as:

T0 T0 / 2
1 1
 x(t )e 0 dt   x(t )e
 jk t  jk0 t
ak  dt (Analysis) ---- (II)
T0 0
T0 T0 / 2

 Proof:

 Recall  n (t )  e jn0t (CT complex exponential of frequency n 0 ).

 Starting from (I), try to obtain (II)

 Multiply both sides of (I) by  n (t ) * and integrate over one period, we have

  
T0 0T

0 0  k
 jn0 t
x (t ) e dt  a k e jk0t e  jn0t dt
  
 T0

 a e
k  
k
j ( k  n ) 0t
dt = RHS
0

T0 T0

e  cos[(k  n) t ]  j sin[(k  n) t ] dt


j ( k  n ) 0 t
 Consider the integral dt  0 0
0 0

 T0
  1. dt  T0 , if k  n
 0

T0
  cos[(k  n) 0 t ]  j sin[(k  n) 0 t ] dt  0, if k  n
 0

5
Course Instructor: Dr A. I. Sulyman

 T0

 Thus RHS =  a k  e j ( k n)0t dt  a   0    an  T0    a  0  anT0


k   0

T0

 x(t )e
 jn0 t
 Hence dt =RHS= a n T0 , (k = n)
0

T0
1
 x(t )e
 jk0 t
 i.e. a k  dt
T0 0

 Remarks:

 a k are called Fourier series coefficients

T0 T0
1 1
 x(t )e  x(t ) dt
 j 0
 For k  0 , a 0  dt  is referred to as the average or (DC)
T0 0
T0 0

value of x(t ) .

 For k  1 , a 1 : 1st harmonic components

 For k  2 , a  2 : 2nd harmonic components

 For k   M , a  M : Mth harmonic components

6
Course Instructor: Dr A. I. Sulyman

Example1:

Compute the Fourier Series of a train of periodic rectangular pulses x (t ) , with amplitude
duration 2T and period T0 as shown below.

x (t )
1

t
 T0 T 0 T T0

Solution:


x (t )  a e
k  
k
jk0t
, where

T0 T0 / 2 T
1 1 1 2T
a0 
T0  x(t ) dt 
0
T0  x(t ) dt 
T0 / 2
T0  1  dt  T
T 0

T0 T0 / 2 T
1 1 1
 x(t ) e 0 dt   x(t ) e 0 dt  e
 jk t  jk t  jk0t
ak  dt
T0 0
T0 T0 / 2
T0 T


1
jk0T0

e  jk0T  e jk0T 
1
 cos(k0T )  j sin(k0T )  cos(k0T )  j sin(k0T )
jk0T0
1

2
 2 j sin(k0T )
jk0 ( )
0
sin(k0T )
 , k  0.
k

2T 
sin k0T jk0t
Thus the FS expansion of x (t ) is given by: x (t )    e
T0 k  k
k 0

7
Course Instructor: Dr A. I. Sulyman

Fourier Series of Real Periodic Signals

 If x(t) is periodic, it can be represented as



x (t )  a e
k  
k
jk0t
; where a k is generally complex ----- (III)


 Then x * (t )  a e
k  
*  jk0t
k . Replacing k by –k in the sum doesn’t change the result.


 Therefore, x * (t )  a
k  
*
k e jk0t ----- (IV)

 If x (t ) is real, then x * (t )  x (t ) . Equating Eqs. (III) and (IV), we have

 a k  a * k or a k*  a k

 Therefore, for real signals an alternative form of the FS is obtained as



x (t )  a0   [a k e jk0t  a k e  jk0t ]
k 1


 a0   [a k e jk0t  a * e  jk0t ]
k 1

 Note that A+A*=2Re{A},  A  a  jb .


 Therefore x (t )  a0   2 Re{a k e jk0t } ----- (V)
k 1

 If a k is expressed in polar form: a k | a k | e jk , then Eq (V) gives one form of FS for
real, periodic signals as

x (t )  a0  2 Re{| a k | e jk0t e jk }
k 1


i.e., x (t )  a0  2 | a k | cos(k0 t   k )
k 1

 If a k is expressed in rectangular form: a k  bk  jck , where bk & ck are real, then Eq


(V) gives another form of FS representation for real, periodic signals as

8
Course Instructor: Dr A. I. Sulyman


x (t )  a0  2 Re[bk  jck ][cos(k0 t )  j sin(k0 t )]
k 1


i.e., x (t )  a0  2 [bk cos(k0 t )  ck sin(k0 t )]
k 1

T0 T0
1 1
 x (t )e  x(t )[cos(k t )  j sin(k t )] dt  b
 jk0t
 Since a k  dt  0 0 k  jck
T0 0
T0 0

 It follows that
T0
1
bk 
T0  x(t ) cos k t
0
0 dt

1 0
T

T0 0
ck  x (t ) sin k0 t dt

T0
1
a0 
T0  x(t ) dt
0

 Applications: allows Fourier analysis of real signals, e.g. voice signals, interms of real
coefficients.

 Dirichlet conditions: The following are sufficient conditions for Fourier series to
converge:

 (i) x (t ) must have at most a finite number of maxima & minima in one period.

 (ii) x (t ) must have at most a finite number of discontinuities in one period.

 (iii) x (t ) must be absolutely integrable.

 Remarks:

 For a periodic signal that varies continuously, the Fourier series converges and equals
the original signal at every value of t.

 For a periodic signal with finite number of discontinuities, the Fourier series equals
the original signal everywhere except at the isolated points of discontinuity, at which
the series converges to the average value of the signal on either side of the
discontinuity.

9
Course Instructor: Dr A. I. Sulyman

 Example 2:

Compute the FS representation of x (t ) | cos t | shown below.

x (t )
1

… …
t
 0 

2 2

Solution:

1 2
a0 
T0 T0
x (t ) dt ; T0   , 0 
T0
2



1 1 2 2
  2 cos(t ) dt  sin t  
 2
    
2

1
ak 
T0 T0
x (t ) e  jk0t dt ; 0  2


1

 j 2 kt
  cos(t ) e
2
dt

2


1 ( e jt  e  jt )  j 2 kt

 

2

2
2
e dt ,

Evaluating the integrals using e j  cos   j sin  , cos(  / 2 )  0 , sin(  / 2 )  1 , and cos(k )  ( 1) k ,
we obtain

1  2 cos(k ) 2 cos(k ) 
ak  
2  (1  2k ) (1  2k ) 

2( 1) k

 (1  4k 2 )

2 
2( 1) k
 Thus the FS expansion of x (t ) is: x (t ) 

 
k   (1  4 k 2
)
e j 2 kt
k 0

10
Course Instructor: Dr A. I. Sulyman

 Example 3:

Compute the FS representation of the signal x (t ) shown below

x (t )
1

t
 2 1 0 1 2

Solution:

2
T0  2 , 0  
2

1
a0 
T0 
T0
x (t ) dt

1  1  t2 1 t 2 2 1
1 2

2  0 1
  t dt  ( 2  t ) dt    ( )  ( 2 t  ) 
 2 2 0 2 1 2

1
ak 
T0 
T0
x (t ) e  jk0t dt

1 
1 2

 
 jk0t
 t e dt  (2  t ) e  jk0t dt 
2 0 1 

1 2 
 
2  ( k )
( 1) k  1  2 2 ( 1) k  1
1
  k
2

 Example 4:

Compute the FS representation of the signal x (t )  cos 4t  sin 6t .

Solution:

0  gcd(4,6)  2 , T0  

 Another method for sinusoids: express sin t , cos t interm of complex exponentials.

11
Course Instructor: Dr A. I. Sulyman

 Thus x (t )   e  e  j 4 t 
1 j 4t
2
1
2j
 e j 6t  e  j 6t 


2

1 j 2 0 t
e  e  j 2 0 t 
1

2j

e j 30t  e  j 30t 

 1
 2 j , k  3

 1
 Therefore a k   , k  2
 2

 0 , otherwise

3 - Discrete-time Fourier Series (DTFS)

 Consider the DT complex exponential:  ( n )  e j0n . It is periodic with fundamental


period N and fundamental frequency 0  2 / N .

 Then the set k (n )  e jk0n have frequencies that are multiples of 0 and are therefore
harmonically related.

 The signals in the set { k (n ) } are all periodic with period N, since
2 2 2 2
jk ( )( n  N ) jk ( )n jN ( )n jk n
{ x ( N  n )  x (n ) }. i.e. k ( n  N )  e N
e N
e N
e N
 k (n )

 Also, there are only N distinct periodic exponentials in the set { 0 ( n ) , 1 ( n ) ,  2 ( n ) ,
2 2 2 2
j ( k  N )( )n jk ( )n jN ( )n jk n
…}, because k  N (n )  e N
e N
e N
e N
 k (n )

 This implies that after generating the set { 0 ( n ) , 1 ( n ) ,…, N 1 ( n ) }, the next set
{ N ( n ) ,  N 1 ( n ) ,…, 2 N 1 ( n ) } is the same as first one.

 Applications: e.g. 128-DSL, 256-DSL are designed for fixed speed depending on the #
of distinct complex exponentials present in the system.

12
Course Instructor: Dr A. I. Sulyman

 Orthogonality of complex exponentials:

 Let k ( n ) and m ( n ) be two harmonically related complex exponentials. The inner


product is defined as:
2
j ( k m )
 k (n)m* (n)  e
n
N

n  N  n  N 

2 2
  {cos[(k  m) N
n  N 
n ]  j sin[( k  m )
N
n ]}

 N , k  m
 
 0 , k  m
(see also the DT complex exponentials)

 Therefore, k ( n ) and m ( n ) are orthogonal.

 Orthogonality of DT complex exponential is used in deriving the DT Fourier series


representation for DT periodic signal (the derivation is similar to the CT case).

 Let x (n ) be a DT periodic signal with period N.

 The DTFS representation for x (n ) is given by

2
jk ( )n
x(n)   ak e
k  N 
N
(Synthesis) ---- (I)

where the coefficients a k are the DT Fourier coefficients, and defined as:

2
1  jk ( )n 2
ak 
N
 x ( n )e
n  N 
N
; 0 
N
(Analysis) ---- (II)

 Remarks:

 The DT FS coefficients a k have the property that a k  N  a k

 The DTFS can be evaluated numerically and can be manipulated on a computer,


because the summation in (I) & (II) above are finite.

13
Course Instructor: Dr A. I. Sulyman

Example1:

Determine the DTFS coefficients for the periodic signal x ( n ) shown below:
x (n )
2
1
n
 5  1 0 1 5

Solution:
2
1  jk ( )n
 ak 
N
 x ( n )e
n  N 
N
; N  6.

2
1 1  jk ( )n
 
6 n 1
x ( n )e 6

 
1 jk ( )  jk ( )
 {2e 3  1  2e 3 }
6

1  1 2 
 {1  4 cos(k )}   cos(k )
6 3 6 3 3

Example2:

6
Determine the DTFS coefficients a k for the periodic signal x ( n )  sin n.
5

Solution:

 x ( n ) is periodic with period N  5 , because 0 NT  k  2  N  k  2 = 5 for k=3.


6 / 5

6 6
1 j5n j n 1 j 30n 1  j 30n 2
 Thus x ( n )  {e e 5 }  e  e }; 0 
5
2j 2j 2j

 1
 2 j k 3
 Therefore a k   , and a k  a k  N  a k 5 .
1
 k  3
 2 j

a k

k
 3  2 2 3 7 8

14
Course Instructor: Dr A. I. Sulyman

4 - The Fourier Transform

x (t )

 T1 T1

~
x (t )

 T0  T1 T1 T0

 Let x(t) be an aperiodic signal.

 Let ~
x (t ) be a periodic signal for which x(t ) is one period. Then


~
x (t )  
k  
ake jk  0 t
, (1)

T0
1 ~

jk  0 t
ak  2
T x (t )e  dt (2)
T0  0
2

T0
1

jk  0 t
 2
T x (t )e  dt ; since ~x (t )  x(t ) for | t |  T0 / 2 .
T0  0
2

1 

jk  0 t
 x (t )e  dt ; since x(t )  0 outside the interval | t |  T0 / 2 .
T0 

 Let

X ( jk 0 )   x(t )e  jk0t dt (3)


 Thus

1
ak  X ( jk 0 ) . (4)
T0
 Combining Eq. (4) and (1), we have

15
Course Instructor: Dr A. I. Sulyman


~ 1
x (t )  
k   T0
X ( jk  0 ) e jk  0 t


1

2

k  
X ( jk  0 ) e jk  0 t
0 (5)

 As T0 → ∞, ~
x (t ) → x(t ) , and the FS expression for ~
x (t ) converges to a continuous
function in frequency known as Fourier Transform (or spectrum) of x(t ) :

F .S .

~
x (t )

t
T0 
~
x (t )

t
T0 

~
x (t )  x (t )
T0  
t

 Thus taking the limit in Eq. (5) as T0 → ∞, we have

x ( t )  lim ~
x (t )
T0  


1
 lim
T0  2
 X ( j )e  
k  
j t
; where   (k  1) 0  k 0 , and   k 0 .

 After taking the limit we have

k


  d 

 Hence

1
 X ( j )e
jt
x(t )  d (Inverse FT)
2 

16
Course Instructor: Dr A. I. Sulyman

and substituting   k  0 in Eq. (3) we have


X ( j )   
x ( t ) e  j t dt (FT)

 x(t ) 
FT
X ( j ) are called Fourier transform pairs.

x(t ) X ( j)
F.T.

 Example1:

Relationship between FS and FT:

 Example 2: show that the FS of a periodic signal ~ x (t ) (with period T0 ) is the sampling
~
x (t ) 0  t  T0
of the spectrum of the aperiodic the signal x(t ) such that x(t )   .
0 otherwise

 Solution:
T0
2
1 ~
 x (t )e
 jk 0t
ak  dt
T0 T
 0
2

T0
1

jk  0 t
 2
T x (t )e  dt
T0  0
2

1 

jk  0 t
 x (t )e  dt
T0 

 Comparing this with the FT of x(t) above, we observe that

1
ak  X ( j ) sampled at   k0
T0

Illustrations:

17
Course Instructor: Dr A. I. Sulyman

X ( j )

ak


Remarks:

 FS: gives the frequency contents of signal at discrete frequencies, k  0 ; k=0,1,2, …

 FT: gives the frequency contents of signals at all frequencies, 

 Hence, F.T. is referred to as the spectrum of x (t )

 F.S. represents the sampling of the spectrum of x (t ) at k  0 , k=0,1, …

Convergence:

 Sufficient conditions for the existence of the Fourier Transform x ( ) for a signal
x (t ) is the same as those explained earlier for Fourier Series (Dirichlet Conditions,
see page 9).

 These conditions are only sufficient conditions, and not necessary conditions, which
means that some signals exist which violate at least one of these conditions and yet
have Fourier transforms.

 Example 1: compute the FT of x(t )   (t ) .


Solution: X ( j )    (t ) e  j t dt  1,  t


x (t ) X ( j )
F.T. 1

Thus, we establish another FT pair as:

 Example2: x (t )  e  at u(t )

Solution: Note: X ( j ) does not exist for a  0 {Fourier Integral does not converge}

 For  a  0 , we have 

18
Course Instructor: Dr A. I. Sulyman


X ( j )  e u(t )e  j t dt  
 at



1
    e at e  j t dt  , a  0 
0
a  j

Notes:
 X ( j ) is complex  X ( j ) | X ( j ) | e j ( ) ;  ( )  angle{ X ( )}
 we plot | X ( j ) | , known as magnitude spectrum, &  ( ) , known as phase
spectrum.

Sketch:
1 a  j a j
X ( j )   2  2  2 ; but | a  jb | a 2  b 2
a  j a  2
a  2
a  2
a2 1
| X ( j ) |  2  a2  2
(a   )
2 2 2
(a   2 )
1
| X( j) |
a

2
2a

a 0 a

  / 2 ,   
 imag ( X ( ))  1  
Also,  ( )  tan  1
   tan ( )     / 4 ,   a
 real ( X ( ))  a  0,
 0

 ( )
2

4
a 
a 0


4


2

 Example 3: x (t )  e  a | t | , a  0

e
 a| t |  j t
Solution: X ( j )  e dt

0 

 e e dt   e e dt
a t  j t  a t  j t

 0

19
Course Instructor: Dr A. I. Sulyman

1 1 2a
   2
a  j a  j a   2
sin c( x )
1

 cos x
sin x  x 0  1, x0
Note: sin c( x )    3 2 1 0 1 2 3 x
x 0 , x  1,2,3

 Example 4:

1 | t |  T1
Find the FT of x (t )  
0 | t |  T1
x (t )
1

t
 T1 0 T1

Then X ( j )   x (t )e
 j t
dt

T1

e
 j t
 dt
T1

2 sin T1 T1


  2T1 sinc ( )
 
2T1 X ( )


2   2
  0
T1 T1 T1 T1

 Example 5:
1 , |  |  B
Find the inverse FT for X ( j )  
0 , |  |  B
X ( )
1


B 0 B

 Solution:

20
Course Instructor: Dr A. I. Sulyman

B
1 sin Bt
e
j t
x (t )  d 
2 B
t

B Bt
 sin c( )
 
B x(t )

t
2   2
  0
B B B B

 Remarks: Ex 4 & 5 illustrates the duality of FT (more later).

Fourier Transform of Basic signals:

 (1) Complex exponential e j0t :

Consider the inverse FT of “ 2 (  0 ) ”.


1
Then FT 1 X ( j )  x (t )   2 (   )e
jt
d  e j0t
2
0


Thus x (t )  e j0t 
FT
 X ( j )  2 (  0 )

 (2) Constant function x (t )  1 , t . Set 0  0 in x (t ) above, we have:

x (t )  1 , t 
FT
 X ( j )  2 ( )

x (t ) X ( j )
1 F.T. 2

 (3) Periodic signals:

 Let x (t ) be periodic, with fundamental period T0 , and 0  2 / T0 ,

 Compute FT x (t )


 Because x (t ) be periodic, we have x (t )  a e
k  
k
jk0t

21
Course Instructor: Dr A. I. Sulyman

  
 Thus X ( j )  FT x (t )  FT   a k e jk0t 
k   

 Since FT is a linear operation, it is additive, hence

 a FT e  

X ( j )  k
jk 0t

k  


  2a  (  k )
k  
k 0


Thus we have the FT pair: x (t ) (periodic) 
FT
 X ( j )   2a  (  k )
k  
k 0

 (4) (sinusoid) x (t )  sin 0t :

1 
Then X ( j )  FT  ( e j0t  e  j0t )
2 j 

1
 2 (  0 )  2 (  0 )
2j

1
i.e, FT pair: x (t )  sin(0 t ) 
FT
 X ( j )   j 2 (  0 )  j 2 (  0 )
2


1
2

2 (  0 )e  j / 2  2 (  0 )e j / 2 
1 1
 Magnitude Spectrum: | X ( j ) | 2 (  0 )  2 (  0 )
2 2

 / 2 ,   0
 Phase spectrum:  ( )  
   / 2 ,   0

| X ( j) |
 ( )
2 1 2 
2
2

 0 0  0  0

2

 (5) (sinusoid) x (t )  cos 0t :

22
Course Instructor: Dr A. I. Sulyman

1 
Then X ( j )  FT  ( e j0t  e  j0t )
2 

1
 2 (  0 )  2 (  0 )
2

1
i.e, FT pair: x (t )  cos(0t ) 
FT
 X ( j )  2 (  0 )  2 (  0 )
2
| X ( j ) |
1  ( )
2 2 (no phase distortion)
2

 0 0  0 0

Example (using basic results above):



Compute the FT of x (t )    (t  kT )
k  
0

x(t )

 
t

 2T0  T0 0 T0 2T0

Solution:

Since x (t ) is periodic, x (t )  a e
k  
k
jk0t
,

T0 / 2
1
  (t )e
 jk0t
where a k  dt
T0 T0 / 2

1

T0

Thus using basic result # 3 above,



X ( j )   2a  (  k )
k  
k 0

23
Course Instructor: Dr A. I. Sulyman


2k
  2a  ( 
k  
k
T0
)

X ( j)
2
 T0 

2 0 2

T0 T0

Notes:

(i) FT of impulse train is another impulse train in frequency domain

(ii) As the spacing between the impulses in time gets larger, the spacing between
the impulses in frequency gets smaller.

24
Course Instructor: Dr A. I. Sulyman

5 - Discrete-time Fourier Transform (DTFT)

 Let x(n) be a general sequence of finite duration.

 From the aperiodic sequence x(n) , we can construct a periodic sequence ~


x (n) for
which x(n) is one period.

x (n )

 N1 0 N1

~
x (n)

n
N N
 N1 0 N1

2
 Then ~  ak e
jk n
x (n )  N
, where
k  N 

2
1  jk
 ~x (n)e N
n
ak 
N n  N 

2
1 N1
 jk
 x(n)e (since ~
n
 N
x (n) = x(n) in the interval  N 1  n  N 1 )
N n   N1

1 
2
  x(n) e  jk0n
; 0  , (since ~
x (n) =0 outside  N 1  n  N 1 )
N n   N


 Let X (e jk0 )   x ( n )e
n  
 jk0n

1
 Then a k  X ( e jk0 ) . Substituting this into ~
x (n) above,
N

1 2 1 0
 we get: ~
x (n ) 
N
 X (e
k  N 
jk0
)e jk0n , and using 0 
N
or 
N 2

25
Course Instructor: Dr A. I. Sulyman

1
 we have: ~
x (n ) 
2
 X (e
k  N 
jk0
)e jk0n  0

 The figure below shows the sketch of “ X ( e jk0 )e jk0n ”, and it can be observed the
signal is periodic with fundamental period 2 .

X (e j )e j

2 
 2 0

  , ~
 As N  x ( n) = x ( n )

 
 d


N 

 T  2

 Hence,

1
  X (e
j
x( n )  )e jn d (Inverse DTFT)
2 2

 Also, letting   k0 in the expression for X ( e j ) above, we have


X (e j )   x ( n )e
n  
 jn
(DTFT)

 Remarks:

 X ( e j ) is called the discrete-time Fourier transform of x(n) .

 X ( e j ) is periodic in  with fundamental period 2 .

 N a k corresponds to samples of the DTFT in one period.

 The infinite summation in the DTFT converges if x(n) is absolutely summable. i.e.,
 

 | x(n ) |  , which happens if the sequence has finite energy: n


n   
| x(n) |2   .

 Example 1: Find the DTFT of the sequence x ( n )  a n u( n ) , | a | 1

 Solution:

26
Course Instructor: Dr A. I. Sulyman


X ( e j )   a u ( n )e
n  
n  jn


1
  (ae  j ) n 
n 0 1  ae  j
j
angle { X ( e )}

| X (e j ) |

 2 0 2

 2 0 2 

 Example 2: Find the DTFT of the sequence: x ( n )  a |n| , | a | 1 .

 Solution:

X ( e j )  a
n  
|n|
u( n )e  jn

1 
 a
n  
 n  jn
e   a n e  jn
n 0

 
  ( ae j ) n  1   ( ae  j ) n
n 0 n 0

1 1 1  a2
  1  
1  ae j 1  ae  j 1  2a cos   a 2

 See textbook for more Examples

6 - Properties of Fourier Representations

 There are four Fourier representations, all of which share similar properties: CTFS,
DTFS, CTFT, DTFT

(1) Linearity:

27
Course Instructor: Dr A. I. Sulyman

 All the four representations are linear

Example: if x1 (t ) 
FT
 X 1 ( j )

x2 (t ) 
FT
 X 2 ( j )

Then z (t )  ax1 (t )  bx 2 (t ) 
FT
 Z ( j )  aX 1 ( j )  bX 2 ( j )

 Proof:

 z ( t )e
 j t
Z ( j )  dt


 ax (t )  bx (t )e
 j t
 1 2 dt


 
 a  x1 (t )e  j t dt  b  x2 (t )e  j t dt
 

 aX 1 ( )  bX 2 ( )

 Example1: Find the FT of x (t )   e 2 t u (t )  2e 3t u (t )

Solution:

Using linearity of FT: X ( j )  FT { e 2 t u(t )}  FT {2e 3t u(t )}

1 2
Thus, X ( j )  
2  j 3  j

 Example2: Find the FS coefficients for the signal z(t) shown

z (t)

2
1
… 2 …
t
0 1 1
 1
 2 8 4 1 2

Solution:

28
Course Instructor: Dr A. I. Sulyman

1 3
z (t ) can be written as z (t )  y (t )  x (t ) , where y(t) and x(t) are as shown below:
2 2

x (t)
1

t
1
 1 1
8

y (t)
1
t
1
 1 1
4

3 sin(k / 4) 1 sin(k / 2)
Thus a k  
2 k 2 k

1 1
 Example3: Find the DTFT of the signal: x (n )  4(  ) n u(n )  ( ) n u(n )
2 3

Solution:

4 4
Using linearity of DTFT, we have: X ( e j )  
1 1
1  (  )e  j 1  e  j
2 3

(2) Time Shift: “shift in time by  t0 ” 


FT
 “scaling in frequency by e  j t0 ”

Example: if x (t ) 
FT
 X ( j )

then x (t  t0 ) 
FT
 X ( j )e  j t0

 Proof:

 x ( t  t )e
 j t
FT {x (t  t0 )}  0 dt


 x (u )e
 j ( u  t0 )
 du ; u  t  t0



 (  x (u )e  ju du )e  j t0


 X ( j )e  j t0

29
Course Instructor: Dr A. I. Sulyman

 Same property hold for the other three Fourier representations:

E.g. CTFS: if x (t ) (periodic, with period T0  2 / 0 ) 


FS
ak

then x (t  t0 ) 
FS
a k e  jk0 t0

 1, | t |  T1 2
 Example 4: The FT of a signal x (t )   is X ( j )  sin(T ) .
0, otherwise 
1, 0  t  2T
Determine the FT of the signal y (t )   .
0, otherwise

Solution:

Note that: y (t )  x (t  T )

2
Thus using the shifting property: Y ( j )  X ( j )e  jT  sin(T )e  jT

(3) Frequency Shift: “scaling in time by e  j0 t ” 


FT
 “shift in frequency by  0 ”

If x (t ) 
FT
 X ( )

then e j0t x (t ) 
FT
 X (  0 )

 Proof:

1
 X (   )e
j t
FT { X (  0 )} 
1
d
2
0



1
 X ( )e
 j ( 0 ) t
 d ;     0
2 


 e j0 t  X ( )e
j t
d


 e j0 t x (t )

 Same property hold for other three Fourier representations.

E.g. CTFS: if x (t ) (periodic, with period T0  2 / 0 ) 


FS
ak

30
Course Instructor: Dr A. I. Sulyman

then e jk0 t x (t ) 


FS
a k k0

e j10t , | t |  
 Example 5: Find the FT of the signal z (t )   .
0, otherwise

Solution:

1, | t | 
Notice that: z (t )  e j10t x (t ) , where x (t )  
0, otherwise

Thus: Z ( j )  X (  10) ; x (t ) 
FT
 X ( )

2
 Z ( )  sin[(  10) ]
  10

(4) Time and Frequency Scaling: “Expansion in time”  “Compression in frequency & vice versa”
FT

If x (t ) 
FT
 X ( )

1
then x ( at ) 
FT
 X ( / a )
|a |

 Proof:

 x(at )e
 j t
FT {x (at )}  dt


1 
  x ( )e
 j (  / a )
d , a  0
 a 
 
  1 x ( )e  j ( / a ) d , a  0
 a 


1
 
| a | 
x ( )e  j ( / a ) d

1
 X ( / a )
|a |

31
Course Instructor: Dr A. I. Sulyman

 1, | t |  1 2
 Example 6: The FT of a signal x (t )   is X ( j )  sin( ) . Determine
0, otherwise 
 1, | t |  2
the FT of the signal y (t )   .
0, otherwise

Solution:

Note that: y (t )  x (t / 2)

1 2 2
Thus using the property of FT: Y ( j )  sin( / 0.5)  sin(2 )
0.5  / 0.5 

 Remarks: note also that reversing (reflecting) a signal in time  reverses also the FT

E.g. If x (t ) 
FT
 X ( )

then x ( t ) 
FT
 X (  ) (CT case)

and x (  n ) 
FT
 X ( e  j ) (DT case)

 See Lecture notes & the attached tables for other properties.

32
Course Instructor: Dr A. I. Sulyman

TABLE 4.1: PROPERTIES OF THE FOURIER TRANSFORM

33
Course Instructor: Dr A. I. Sulyman

TABLE 4.2: BASIC FOURIER TRANSFORM PAIRS

34
Course Instructor: Dr A. I. Sulyman

TABLE 5.1: PROPERTIES OF THE DISCRETE-TIME FOURIER TRANSFORM

35
Course Instructor: Dr A. I. Sulyman

TABLE 5.2: BASIC DISCRETE-TIME FOURIER TRANSFORM PAIRS

36
Course Instructor: Dr A. I. Sulyman

TABLE 5.3: SUMMARY OF FOURIER SERIES AND TRANSFORM EXPRESSIONS

37

You might also like