9/26/2022

Concept of filtering
Frequency domain

DIGITAL SIGNAL PROCESSING

SEM-1, 2022-23

ECE/EEE F434 Lecture 10

Dr. Sarang C. Dhongdi, Dept of EEE, BITS Goa

Filter design Filter concept

 Design a H.P. filter, where the input is sum of 2 cosine y (n)   h(k ) x(n  k )
functions of 0.1 rad/sample and 0.4 rad/sample. k

 Consider FIR filter with length 3, i.e. h(0), h(1), h(2) y ( n )  h ( 0 ) x ( n )  h (1) x ( n  1)  h ( 2) x ( n  2 )

 Assume the impulse response as symmetric

h(0)=h(2) = α0 and h(1) = α1 y ( n )   0 x ( n )   1 x ( n  1)   0 x ( n  2 )

Filter concept Filter concept


H ( e j )  2 0 cos    1
H ( e j )   h ( n )e
n  
 j n

 h ( 0 )  h (1) e  j  h ( 2) e  j 2  ( )    

  0 (1  e  j 2 )   1e  j  0   6.76195  ( )
 P ( )  
 1  13 .456335 
 2 0 cos    1 e  j

1
 9/26/2022

Discrete Time Fourier Transform (DTFT)

>> n = 0:1:6
n=
0 1
>> x1 = cos(0.1*n)
2 3 4 5 6
• Spectral representation for aperiodic DT
x1 =
1.0000 0.9950 0.9801 0.9553 0.9211 0.8776 0.8253
signals
>> x2 = cos(0.4*n)
x2 =
1.0000 0.9211
>> x = x1+x2
0.6967 0.3624 -0.0292 -0.4161 -0.7374
• DTFT can be derived from spectral
x=
2.0000 1.9161 1.6768 1.3177 0.8919 0.4614 0.0879
representation (i.e. DT Fourier series) for
periodic signals and letting the period
>> h = [-6.76195 13.456335 -6.76195];
>> y = conv(x,h) become infinitely long.
y=
-13.5239 13.9563 0.9211 0.6967 0.3624 -0.0292 -0.4161

Finding Fourier transformation… Finding Fourier transformation

 Impulse function  Shifted impulse

x[n]   [n] x[n]   [n  n0 ]


X (e ) j


  [n]e - jn
1
X ( e j )    [n  n ]e
n  
0
- jn
 e - jn0
n  
 Same amplitude (1), but linear
phase (-wn0)

Finding Fourier transformation Finding Fourier transformation

Exponentially decaying function 1

X ( e j ) 


x[n]  a u[n] 1  ae  j
n
a 1
1

X ( e j ) 
X ( e j )   a u[n]e
n  
n - j n
(1  a cos  )  ja sin 
 1

 , 0
X (e )   a e
j n - jn
j
X (e ) 
1   1 a
1
n 0
1  2a cos   a 2  ,  
 1 a
 

X (e j )   ae - j
n

n 0

2
 9/26/2022

Symmetry properties

For real sequence x[n]  a nu[ n], a 1


X * e  j   ? 
1
1  ae  j
 
 X e j

 
X R e j ? 
1  a cos 
1  a 2  2a cos 
 X R e  j  

 
X I e j ? 
 a sin 
1  a 2  2a cos 
  X I e  j  

Symmetry properties Symmetry properties

 
X e j 
1

 X e  j  Sequence x[n]
Any real x[n]
Discrete-Time Fourier Transform X(ej)
X(ej)=X*(e-j) (conjugate symmetric)
1  a 2  2a cos  Any real x[n] XR(ej)=XR(e-j) (real part is even)

  a sin  
   
Any real x[n] XI(ej)=-XI(e-j) (imaginary part is odd)
j  j
X e  tan 1    X e Any real x[n] |X(ej)|=|X(e-j)| (magnitude is even)
 1  a cos   Any real x[n] X(ej)=-X(e-j) (phase is odd)
xe[n] XR(ej)
xo[n] jXI(ej)

Symmetry properties F.T. – Important signal types

 What about the signals which are not absolutely summable nor
If F.T. of complex sequence x[n] is X (e j ) square summable?

 F.T. of
x[ n] X ( e  j )  Like 1, ejwon, u[n] etc.

 F.T. of x*[n] X * ( e  j )
 F.T. of
x [  n]
*
X * ( e j )

3
 9/26/2022

Finding Fourier transformation

x[n]  1  DTFT is periodic, hence the function has to show periodicity

   2   2k 


 Let us find out the “function” whose inverse F.T. X e j 

gives us value =1 k  

X e e d
1
xn  


2 
 j jn
1
 X (e )e d
j jn 
x[n] 
2 - 1   

    2   2k e
jn
 d
j
X (e )  2 ( ) 2 
     k  

     e jn d  e j 0 n  1
1 
 2 ( )e
j n
x[n]  d  1
2 -

Finding Fourier transformation Standard Fourier transforms

Sequence DTFT

   2   

X e j  0  2k 


k  
1  2  2k 
k  

xn  
1 
2 
 
X e j e jn d e jon


 2   o  2k 
k  

1    
  2   0  2k e jn d
2  k   1 
u[n]     2k 
 1  e  j k  
     0 e jn d  e j0 n


Fourier transform of periodic signals Fourier transform of periodic signals

2  2   2 
 Find F.T. of periodic signal cos o n let o  X (e j )                 

5  5   5 
e jon  2  
k  
o  2k  X (e j )

1 j 0 n 1  j 0 n
cos o n  e  e π
2 2
  -2π-w0 -2π -2π +w0 -w0 0 w0 2π-w0 2π 2π +w0
X ( e j )      
k  
o  2k       
k  
o  2k 

 2    2 
X ( e j )      
k    5
 2k       
 k    5
 2k 


4
 9/26/2022

Finding Fourier transformation

 Discrete time rectangular pulse

Discrete Fourier transform
From D.T.F.T.  D.F.T.

D.T.F.T.  D.F.T. Discrete Fourier transform

• Not practical for (real-time) computation on a digital computer.  D.F.T. is the powerful computational tool that allows to
evaluate F.T. on digital computer or specially designed
digital hardware.
• The DTFT is not numerically computable transform.

 D.F.T. is defined only for finite length sequences.

• Solution: limit the extent of the summation to N points and
evaluate the continuous function of frequency at N equally  D.T.F.T. is continuous and periodic, D.F.T. is obtained by
spaced points.
sampling one period of D.T.F.T. at a finite number of
frequency points.