UNIT – III FINITE IMPULSE FILTERS

Design of FIR filters - symmetric and Anti-symmetric FIR filters - design of linear phase FIR filters using
Fourier series method - FIR filter design using windows (Rectangular, Hamming and Hanning window),
Frequency sampling method. FIR filter structures - linear phase structure, direct form realizations

Introduction:
 A digital filter is just a filter that operates on the digital signals.
Types:
 FIR filter design
 IIR filter design
FIR filter:
The digital filter which designed using finite number of response co-efficient is called as finite impulse
response filters.
ho (n),h1 (n)....................h(N−1) (n)
Advantages:
1. FIR filters have exact linear phase.
2. FIR filters are always stable.
3. FIR filters can be realized in both recursive and non-recursive structure.
4. FIR filters with any arbitrary magnitude response can be tackled using FIR sequence.
Disadvantages:
1. For the same filter specification the order of the FIR filter design can be as high as 5 to 10 times that
of an IIR filter.
2. Large storage requirement needed.
3. Powerful computational facilities required for the implementation.
Linear Phase (LP) FIR Filters:
Derive the condition for Linear Phase (LP) FIR Filters. [Nov/Dec-2009]

The transfer function of a FIR causal filter is given by

N 1
H (z)   h(n)z  n
n 0

Where h(n) is the impulse response of the filter.

The Fourier transform of h(n) is
N 1
H (e j )   h(n)e  j n
n 0,
Which is periodic in frequency with period 2  .
H (e j )   | H (e j ) | e  j   
j
Where H (e ) is magnitude response and  (  ) is phase response.
We define the phase delay and group delay of a filter as
    d   
p  and ; g 
 d ---------------------------------------->(1)
For FIR filters with linear phase we can define
 ( )   ;      ------------------------------------------------>(2)
Where  is a constant phase delay in samples.

1
Substitute: equation 2 in 1, we have  p =  g=  , which means that  is independent of frequency. We can
write,
N 1

 h(n)e
n 0
 j n
  | H (e j ) | e  j   

Which gives us,

N 1

 h(n) cos n   | H (e j
) | cos  ( )
n 0 ------------------------------->(3)
N 1
 h(n) sin  n   | H (e j ) | sin  ( )
and n 0 ---------------------------------->(4)
By taking ratio of equation (3) to equation (4), we obtain
N 1

 h(n) sin  n sin 

n0
N 1
 ;[ ( )   ]
cos 
 h(n) cos  n
n 0 ------------------------------>(5)
After simplifying equation (5) we have
N 1

 h(n) sin(  n)  0

n 0 ------------------------------------------------->(6)
Equation (6) will be zero when
h(n)  h( N  1  n) ------------------------------------------------------->(7)
N 1

And 2 ------------------------------------------------------------------>(8)
Therefore, FIR filters will have constant phase and group delays when the impulse response is symmetrical
N 1

about 2
The impulse response satisfying equation (7) & (8) for odd and even values of N. When N=7 the centre of
1
2
symmetry of the sequence occurs at third sample and when N=6, the filter delay is 2 samples.
If only constant group delay is required, and not the phase delay we can write
 (  )=  -  
j j j    
Now we have H (e )   | H (e ) | e
Equation (9) can be expressed as
N 1

 h(n)e  j n
  | H (e j ) | e j    
n 0 -------------------------------------->(9)

which gives us
N 1

 h(n) cos  n   | H (e  ) | cos(   )

j

n 0 ---------------------------->(10)
N 1
 h(n)sin  n   | H (e j ) | sin(    )
and n 0 ------------------------------>(11)
By taking ratio of equation (11) to (10), we get

2
 N 1
 h(n) sin  n
sin(    )
n0
N 1

cos(    )
 h(n) cos  n
n 0

From which we obtain

N 1

 h(n)sin    (  n)   0
n0 ------------------------------------------>(12)


If 2 , Equation (12) becomes,
N 1

 h(n) cos(  n)  0

n 0 ------------------------------------------------->(13)
The equation 13 will be satisfied when h(n)  h( N  1  n)
N 1

And 2
g
Therefore, FIR filters have constant group delay, and not constant phase delay when the impulse response
N 1

is anti-symmetrical about 2 .
Example:
N −1 6−1 1
α= = =2 (centre of symmetry )
For N=6 2 2 2

( centre of symmetry )

0 1 2 2.5 3 4 5
For N=7
N −1 7−1
α= = =3( centre of symmetry )
7 2 2
For N=
centre of symmetry

0 1 2 3 4 n
Linear Phase FIR Filter:
An FIR filter has linear phase if its unit sample response satisfies the condition
h(n )=±h( M−1−n ); n=0,1,2 . . .. .. .. N −1
Case (i): Symmetric impulse response for “N is ODD”:
Determine the frequency response of FIR filter with symmetric impulse response and the order
of the filter is “N is Odd”.
3
The frequency response of impulse response can be written as,
M −3
2 jω ( M −1) M −1
M−1 −
jω
H ( e )= ∑ h(n )e
n=0
− jωn
+h
2
e ( ) 2
+
n=
∑
N +1
h(n )e− jωn
2 -------------------------->(1)
jω jω jω
Let n=M−1−n, where z=e |H (e )=H ( z )|z=e
M−3 M −3
2 jω ( M −1) 2

( M2−1 ) e
−
H ( e jω )= ∑ h(n )e− jωn+ h 2
+ ∑ h( M −1−n) e− jω ( M −1−n )
n=0 --------->(2) n=0

For a symmetrical impulse response, h(n )=h( M −1−n ), substituting this relation in above equation (2 )
M−3 M−1

jω
H ( e )=e
−
jω( M−1)
2
[ 2
∑ h(n )e
n=0

M−3
jω
M−1
( )+
2−n
2
∑ h(n) e
n=0
−
jω ( M−1 )
2−n
+h ( M−12 ) ]
jω
H ( e )=e
−
jω( M −1)
2

jω
[ 2
2
∑
n=0
h(n )cosω ( M−1
2 )−n + h (
M−1
2 ) ] ]
The polar form of H (e ) can be expressed as
jω )
jω jω j∠ H( e
H ( e )=|H ( e )|e
M −3
2
jω
∴ Magnitude of H (e ) is given as
|H ( e jω )|=h ( M2−1 )+2 ∑ h( n) cos ω ( M2−1 −n)
n=0

M−1
jω
Angle of H (e ) is given as
============================================================================
∠H(e =¿ −ω
2
jω)
{( )
, for |H(e jω )|>0 ¿ ¿¿¿

Case (ii) : Symmetric Impulse Response For –“N is EVEN”:

Determine the frequency response of FIR filter with symmetric impulse response and the
order of the filter N is Even. [Nov/Dec-2013]

The frequency response of impulse response can be written as,

M −2 M−2
2 2
H ( e jω) = ∑ h (n)e− jωn + ∑ h(n )e− jωn
n=0 n=0 ---------------------------------------------->(1)
jω jω jω
Let n=M−1−n, where z=e |H (e )=H ( z )|z=e
M−2 M−2
2 2
H ( e jω ) = ∑ h (n)e− jωn + ∑ h( M−1−n) e− jω ( M−1−n )
n=0 ------------------------------>(2)
n=0

For a symmetrical impulse response, h(n )=h( M −1−n ), substituting this relation in above equation (2 )
M−3 M−1
2 jω ( M−1) 2

( M−1
−

2 )
H ( e jω )= ∑ h(n )e− jωn+h e 2
+ ∑ h(n) e− jω ( M−1−n )
n=0 n=0 ---------------->(3)

4
 M−2 M−2

jω
H ( e )=e
−
jω( M−1)
2
[ 2
∑ h(n )e
n=0
M−2
jω
M−1
2( −n )+ 2
∑ h(n ) e
n=0
− jω ( M−12−n)
]
H ( e jω )=e
−
jω( M−1)
2

jω
[ 2
2
∑ h(n )cos ω( M−1
n=0 2
−n )

jω
] jω j∠ H( e
jω )
The polar form of H (e ) can be expressed as H ( e )=|H ( e )|e
M−2
2
∴ Magnitude |H (e jω )|=2 ∑
n=0
h(n )cos ω ( M−1
2
−n )

M−1
jω
Angle of H (e ) is given as
∠H(e jω)=¿ −ω
=============================================================================
{( ) 2
, for |H(e jω )|>0 ¿ ¿¿¿

Case (iii) : Antisymmetric for “N is ODD”:

Determine the frequency response of FIR filter with Antisymmetric impulse response and the order
of the filter N is Odd.

For this type of sequence

h ( M2−1 )=0
M−1
H (e )= ∑ h(n )e− jωn
jω

n=0
The frequency response of impulse response can be written as,
M−3 M−3
2 2
H ( e jω )= ∑ h(n )e− jωn+ ∑ h( M−1−n)e− jω( M−1−n )
n=0 n=0

for antisymmetric impulse response, h(n )=−h( M−1−n )

M−3 M−3
2 2
jω = ∑ h(n )e− jωn− ∑ h (n)e− jω ( M −1−n )
H (e ) n=0 n=0
M−3 M−3

=e
− jω
M −1
( )2
[ 2
∑ h(n )e
n=0
M−3
jω ( M−1
2
−n )− 2
∑ h(n)e
n=0
− jω ( M−12 −n)
]
=e
− jω
M −1
( )j
2
[ 2
2 ∑ h(n)sin ω
n=0
M−3
( M−1
2
−n ) ]
=e
− jω
M −1
( )e
2
j
π
2
[ 2
2
∑ h(n )sin ω ( M2−1 −n )
n=0
]
5
 M−3

H ( e jω )
=e
j
π
2
M−1
2
[
[ ( )] 2 ∑ h(n )sin ω M −1 −n
−ω 2

n=0
( 2 )
M −3
]
2

Magnitude of H ( e ) is given as,

jω |H (e jω )|=2 ∑
n=0
h( n ) sin ω ( M2−1 −n)
π M−1
jω
Angle of H (e ) is given as,
============================================================================
∠H(e jω )=¿ −ω
2 2 {( )
for |H(e jω )|>0 ¿ ¿¿¿

Case (iv) : Antisymmetric For –“N is EVEN”:

Determine the frequency response of FIR filter with Antisymmetric impulse response and
the order of the filter N is Even. [Nov/Dec-2013]

The frequency response of impulse response can be written as,

M−2 M−2
2 2
H ( e jω ) = ∑ h (n)e− jωn + ∑ h( M−1−n)e− jω( M−1−n)
n=0 n=0

for antisymmetric impulse response, h(n )=−h( M−1−n )

M−2 M−2
2 2
jω = ∑ h(n )e− jωn− ∑ h( n)e− jω ( M−1−n )
H (e ) n=0 n=0
M−2

=e
− jω ( M 2−1 ) j
[ 2
2
∑ h(n )sin ω( M−1
n=0 2
M−2
−n ) ]
H ( e jω )
=e
− jω
M −1
2( )e j
π
2
[ 2

M−2
2
∑ h(n )sin ω ( M2−1 −n )
n=0
]
H ( e jω )
=e
π
j −ω
2
M−1
2
[
[ ( )] 2 ∑ h(n)sin ω M −1 −n
M−2
2

n=0
( 2 ) ]
=e
− jω
M −1
2 ( )j
[ 2
2
∑ h(n )sin ω( M−1
n=0
M−2
2
−n ) ]
=e
j
π
2
M−1
2
[
[ ( )] 2 ∑ h(n)sin ω M −1 −n
−ω 2

n=0
( 2 ) ]
M −2
2

Magnitude of H ( e ) is given as,

jω |H (e jω )|=2 ∑
n=0
h(n )sin ω ( M−1
2
−n )

6
 π M−1
jω
Angle of H (e ) is given as,
∠H(e jω )=¿ −ω
2 2 {( )
for |H(e jω )|>0 ¿ ¿¿¿
=============================================================================
Structures of FIR Filters:
Explain with neat sketches the Structure of FIR filters. [Nov/Dec-2012]

The realization of FIR filter is given by

 Transversal structure.
 Linear phase realization
 Polyphase realization.
Transversal structure:
It contains two forms of realization such as,
 Direct form realization
 Cascade form realization.
Direct form realization:
The system function of an FIR filter can be written as
N 1
H (z)   h(n) z  n
n 0

 h(0)  h(1) z 1  h(2) z 2  ..........  h(N  1) z  (N 1) eq(1)

Y(z)=h(0)X(z)+h(1)z 1 X (z)  h(2)z 2 X (z)  .......  h(N  1) z  (N 1) X (z) eq(2)
This structure is known as direct form realization. It requires N multipliers, N-1 adders, and N-1 delay
elements.

z 1 z 1 z 1 z 1
X(z)

h(1) h(2) h(N-2) h(N-1)

h(0)

+ + + +
Y(z)
Cascade Realization:
Problem 1: Determine the direct form Realization of the following system function. (Nov/Dec-14)
−1 −2 −3 −4 −5
H ( z)=1+2 z +6 z +4 z +5 z +8 z
Solution:
−1 −2 −3 −4 −5
Given: The system function is H ( z)=1+2 z +6 z +4 z +5 z +8 z
Y ( z)
H( z )= =1+2 z −1 +6 z−2 +4 z −3 +5 z−4 +8 z −5
X ( z)
Y ( z)= X (z )+2 z−1 X ( z )+6 z −2 X ( z)+4 z −3 X ( z)+5 z−4 X (z )+8 z −5 X ( z)

7
 X(z)
1 1 1 1 Z-
z z z z 1

2 6 4 5
1

+ + + +

Y(z)

1 2 1
Problem 2: Obtain the cascade realization of system function H (z)  (1  2 z  z )(1  z  z 2 )
(May/June-12) (Nov/Dec-10)
Solution:
H (z)  H1 (z) H 2 (z)
Where H1 (z)  1  2 z 1  z 2 and H 2 (z)  1  z 1  z 2
Y1 (z)
H1 (z)   Y1 (z)  X 1 (z)  2 z 1 X 1 (z)  z 2 X (z) eq(1)
X 1 (z)
Y2 (z)
H 2 (z)   Y2 (z)  X 2 (z)  z 1 X 2 (z)  z 2 X (z) eq(2)
X 2 (z)
The equation (1) and equation (2) can be realized in direct form and can be cascaded as shown in figure.

H.W 1 : Obtain the direct form realization for the following system function.
1. H ( z)= 1+2z−1−3 z−2 −4 z−3 +5 z−4
1 3 1 1
2. ( 4 8 8 )(
H ( z)= 1− z−1 + z −2 1− z−1− z−2
2 )
H.W 2: Obtain the cascade form realization for the following system function.
5
1. H ( z)=1+ z−1 +2 z−2 +2 z−3
2
1
(
2. H ( z )=( 1+2z−1 ) 1+ z−1 +z−2
2 )
Obtain the linear phase realization of the system function. [Nov/Dec-10]
1 1 1 1 1 1
H  z   z  z 2  z 3  z 4  z 5  z 6
2 3 4 3 2

Solution:
By inspection we find system function H(z) is that of a linear phase FIR filter and,
h(n)=h(N-1-n)
Therefore, we can realize the system function as shown in Figure.

8
Lattice Structure:
The lattice structure formulas are,
α m( 0)=1
α m( m)=k m
α m( k )=α m−1 (k )+α m (m)α m−1 (m−1)
****************************************************************************************
1 1 1
K 1 = ;K 2= ; K 3=
Consider an FIR lattice filter with co-efficients 2 3 4 . Determine the FIR filter for the
direct form structure. [Nov/Dec-2013] [Nov/Dec-2015]

Solution:
1 1 1
K 1 = ;K 2= ; K 3=
Given: The FIR lattice filter with co-efficients are 2 3 4
1
α 3 (0 )=1 ; α 3 (3 )=K 3 =
4
1 1
α 2 (2)=K 2= ; α 1 (1 )=K 1 =
3 2
We know,
α m( k)=α m−1 (k)+α m (m)α m−1 (m−1)
For m=2 and K=1
α 2 (1)=α 1 (1)+α 2 (2)α 1 (1)
1 1 1 2
= + . =
2 3 2 3
For m=3 and K=1
α 3 (1)=α 2 (1)+α 3 (3 )α2 (2 )
2 1 1 3
= + . =
3 4 3 4
For m=3 and K=2
α 3 (2)=α 2 (2)+α 3 (3 )α 2 (1 )
1 1 2 1
= + . =
3 4 3 2
3 1 1
∴ The lattice filter coeficients are α 3 (0)=1;α 3 (1 )= ;α 3 (2)= ;α 3 (3)=
4 2 4
---------------------------------------------------------------------------------------------------
H.W: Realize the following system in lattice form. [May/June-07]
4 3 2
y (n)  2 x( n)  x( n  1)  x(n  2)  x(n  3)
5 2 3
Design of FIR Filter:
FIR Filter can be designed using three following techniques.

9
1. Fourier series method
2. Windowing technique
3. Frequency sampling method.
Filter design using windowing technique:
Explain the designing of FIR filters using windows. [April/May-2011]

The desired frequency response of any digital filter is periodic in frequency and can be expanded in a
Fourier series.
∞
jω
H d (e )= ∑ hd (n)e− jω n
n=−∞ ------------------------------------------------------>(1)
Where,
1 π
hd (n )= ∫ H (e jω )e jωn d ω
2π −π
------------------------------------------------>(2)
Gibb’s Phenomenon:
jω
One possible way of finding an FIR filter that approximates H (e ) would be truncate the infinite Fourier

series at
n=± ( M−1
2 ). Abrupt truncation of the series will lead to oscillation both pass band and stop
band. This phenomenon is known as Gibbs phenomenon.
Types of window:
 Rectangular window.
 Hanning window.
 Hamming window.

Rectangular window:
The rectangular window sequence is given by,

M−1 M−1
w R (n)=¿ 1 { ( ) ( ) for −
2
≤n≤
2
¿ ¿¿¿
Hanning window:
The hanning window sequence can be obtained by

2 πn M−1 M−1
{
w Hn (n)=¿ 0 .5+0.5cos
M−1
for − ( ) ( )
2
≤n≤
2
¿ ¿¿¿
Hamming window:
The hamming window can be obtained by

2πn M−1 M−1

{
w Hm(n)=¿ 0 .54+0.46cos
M−1
Filter coefficient (hd(n) ) for different types of Filters:
for − ( ) ( )
2
≤n≤
2
¿ ¿¿¿

Type of Filter hd(n)

LPF ωC
hd (n )= for n=α
π
sin ω C (n−α )
hd (n )= for n≠α
π (n−α )

10
 HPF ωC
hd (n )=1− for n=α
π
1
hd (n )= [ sin(n−α )π −sin(n−α )ωC ] for n≠α
π (n−α )
BPF ω −ω
hd (n )= C 2 C 1 for n=α
π
1
hd (n )= [ sin ωC 2 (n−α)−sin ωC 1( n−α )] for n≠α
π (n−α )
BSF ω C2 −ω C 1
hd (n )=1− [ π ] for n=α
1
hd (n )= [ sin ωC 1 (n−α )−sin ωC 2( n−α )+sin (n−α ) π ] for n≠α
π (n−α )

Design an ideal low pass filter with a frequency response

π π
Solution:
Given:
jω
{
H d (e )=¿ 1 for − ≤ω≤ ¿ ¿¿¿
2 2
π π
Find the values of h(n) for M=11 using hanning window. Find H(z). Plot the magnitude and frequency response. (May/June-14)

{
H d (e jω )=¿ 1 for − ≤ω≤ ¿ ¿¿¿
(Nov/Dec-14) (April/May 2011)(April/May-08) (Nov/Dec-09) (Nov/Dec-10)

2 2
Step 1: To find filter coefficient.
Hence
π
ω = .
2 C

1 π
hd (n ) = ∫ H (e jω )e jωn d ω
2π −π
π
1
= ∫ 2 π 1 . e jωn d ω
2π −2
sin (n−α)ω C
= ; ∵ α=0
π (n−α )
π
sin n
2
hd (n ) = ; −5≤n≤5
πn
π
sin (0 )
2
=
For n=0 ; hd (0) π (0)
πn
sin
1 2
=
2 πn
2
∴ lim sinθθ =1
θ−¿ 0
1
hd (0) =
2
π
sin (1)
2 1
hd (1 )=hd (−1)= = =0 . 3183
For n=1 ; π (1 ) π

11
 sin π
hd (2 )=h d (−2 )= =0
For n=2; 2π
3π
sin
2
hd (3 )=h d (−3 )= =−0 .106
For n=3; 3π
sin 2 π
hd ( 4 )=h d (−4 )= =0
For n=4; 4π
5π
sin
2
hd (5 )=h d (−5 )= =0 .06366
For n=5; 5π
Step 2: To find hanning window:

2 πn M−1 M−1
{
w Hn (n)=¿ 0 .5+0.5cos
M−1
2 πn
for − ( ) ( )
2
≤n≤
2
¿ ¿¿¿

=0 . 5+0 . 5cos for −5≤n≤( 5 )

10
π (0 )
w Hn (0 )=0 . 5+0 . 5 cos =1
For n=0; 5
π (1)
w Hn (1 )=w Hn (−1 )=0 .5+0 .5 cos =0 . 9045
For n=1 ; 5
π (2)
w Hn (2 )=w Hn (−2 )=0 . 5+0 .5 cos =0 . 6545
For n=2; ` 5
π (3 )
w Hn (3 )=w Hn (−3 )=0. 5+0 . 5 cos =0 .3454
For n=3 ; 5
π (4 )
w Hn ( 4 )=w Hn (−4 )=0 . 5+0 .5 cos =0 . 0954
For n=4 ; 5
π (5 )
w Hn (5 )=w Hn (−5 )=0 . 5+0 . 5 cos =0
For n=5 ; 5
Step 3: To find filter coefficients using hanning window are
h(n )=h d (n)∗w Hn (n) for −5≤n≤5
h(0 )=h d (0)∗w Hn (0)=(0 .5 )∗1=0.5
h(1)=hd (1 )∗w Hn (1 )=0. 3183∗0. 9045=0 .2879
h(2 )=hd (2 )∗wHn (2)=0∗0 . 6545=0
h(3 )=h d (3)∗w Hn (3)=−0 .106∗0.3454=0.0366

12
Step 4: The transfer function of the filter is given by
M−1
2
H ( z)=h( 0)+ ∑ [ h(n) ( z n +z −n ) ]
n=1
5
n −n
= 0 .5+ ∑ h(n ) ( z +z )
n=1
= 0 .5+h(1) ( z 1 +z−1 ) +h(2) ( z 2 +z−2 ) +h(3) ( z 3 +z−3 ) +h( 4 ) ( z 4 +z−4 ) +h(5 ) ( z5 +z−5 )
= 0 .5+0 .287 z 1 +0 . 287 z−1 +0 . 0366 z3 +0 . 0366 z −3 +0 .006 z 4 +0 . 006 z−4
Step 5: The transfer function of the realizable filter is
( M −1 )
−
2
H ' ( z)=z H ( z)
= z−5 [ 0 . 5+0 .287 z 1 +0. 287 z−1 +0 . 0366 z3 +0 . 0366 z−3 +0 . 006 z 4 +0 . 006 z−4 ]
H ' ( z)=0 . 5 z−5 +0 . 287 z −4 +0. 287 z−6 +0. 0366 z−2 +0 . 0366 z−8 +0 . 006 z−1 +0 . 006 z−9
The filter coefficients of causal filter are given by
h(0 )=h (3)=h(7 )=h(10 )== 0,h(1)=h(9 )=0 . 006;h(2 )=h(8 )=0. 0366;h (4 )=h (6)=0 . 287 ;h(5 )=0 . 5

Step 6: The frequency response is given b y

− 5
jω
H (e )=∑ a(n )cosωn
n=0
where ,

(2M−1 )=h(5)=0 .5
a(0)=h

M−1
a(n)=2h ( −n )
2
a(1)=2h(5−1)=2h( 4)=2(0. 006)=0 .12
a(2)=2h(5−2)=2h (3)=0
a(3)=2h (5−3 )=2h(2)=2* 0.036=0. 072
a(4 )=2h(5−4)=2h (1)=2* 0. 006=0.12
a(5)=2h(5−5 )=2h(0)=0
− 5
jω
H (e )= ∑ a (n)cosωn
n=0
= a(0 )+a(1)cos ω+a (2 )cos2ω+a(3 )cos3 ω+a (4 )cos 4ω+a(5)cos5 ω
=0 .5+0 .12 cosω+0.072 cos3 ω+0. 12 cos4 ω
Magnitude in dB is calculated by varying 0 to 10 and tabulated below.
ω(in degree ) 0 1 2 3 4 5
− 0.812 0.8115 0.810 0.8083 0.8054 0.8018
H (e jω )
jω -1.8 -1.814 -1.83 -1.85 -1.88 -1.91
|H (e )|dB
====================================================================================

π
Design an ideal high pass filter with a frequency response
Find the values of h(n) for N=11 using hanning window. (May/June-16)(April/May-08)
jω
4{
Hd (e )=¿ 1 for ≤|ω|≤π ¿ ¿¿¿
13
Solution:
Given :
π
π ωC =

{
H d (e jω )=¿ 1 for ≤|ω|≤π ¿ ¿¿¿
4
Hence 4

Step 1: To find filter coefficient.

1 π
hd (n ) = ∫ H (e jω )e jωn d ω
2π −π
π
1
¿ ∫ 2 π 1. e jωn d ω
2π −2
1
hd (n )
=
π (n−α )
[ sin(n−α )π −sin(n−α )ωC ] for n≠α
1 nπ
hd (n )
=
πn [
sin nπ −sin
4 ]
; −5≤n≤5

Step 2:
1 (0) π
For n=0
hd (0 )=
π (0)[sin(0 )π −sin
4 ]
1
hd (0 )=1− =0 . 75
U sin g , L'hospital rule 4
1
For n=1;
hd (1 )=hd (−1)= [
π (1 )
sin(1 )π−sin
(1 )π
4 ]
=−0 . 225

1
For n=2;
hd (2 )=h d (−2 )=
π (2 ) [
sin(2 )π −sin
(2)π
4 ]
=−0 .159

1
For n=3 ;
hd (3 )=h d (−3 )=
π (3) [
sin(3 )π−sin
(3 )π
4 ]
=−0 .075

1
For n=4 ;
hd ( 4 )=h d (−4 )=
π(4) [
sin (4 )π −sin
(4 )π
4
=0 ]
1
For n=5 ;
hd (5 )=h d (−5 )=
π (5) [
sin(5) π−sin
(5 )π
4 ]
=0 . 045

Step 2: Using Hanning window:

2 πn M−1 M−1
{
w Hn (n)=¿ 0 .5+0.5cos ,
M−1 2( ) ( )
≤n≤
2
¿ ¿¿¿
w Hn (0)=0.5+0. 5=1
π (1 )
w Hn (1 )=w Hn (−1 )=0 .5+0 .5 cos ( ) 5
=0 . 9045

π (2 )
(−2 )=0 . 5+0 .5 cos (
5 )
w Hn (2 )=w Hn =0 . 655

14
 w Hn (3 )=w Hn (−3 )=0. 5+0 . 5 cos ( π (3)5 )=0 .345
π (4 )
(−4 )=0 . 5+0 .5 cos (
5 )
w Hn ( 4 )=w Hn =0. 0945

π (5)
(−5 )=0 . 5+0 . 5 cos (
5 )
w Hn (5 )=w Hn =0

Step 3: The filter coefficients using hanning window are,

h(n )=h d (n)w Hn ( n) for −5≤n≤5
h(0)=h d (0)w Hn (0)=(0.75)(0)=0
h(1)=hd (1)w Hn (1)=(−0.225 )( 0.905 )=−0.204
h(2)=hd (2)w Hn (2)=(−0.159 )(0.655 )=−0 .104
h(3)=h d (3)w Hn (3)=(−0.075)(0.345)=−0.026
h( 4 )=hd ( 4 )w Hn (4 )=(0)(0. 8145 )=0
h(5 )=h d (5 )w Hn (5 )=(0 . 045 )( 0)=0
Step 4: The transfer function of the filter is given by
The transfer function of the realizable filter is

H'( z) =z
− ( M−1
2 )
H ( z)
= z−5 [ 0. 75−0 . 204z−1−0.204 z 1−0 . 104 z−2−0. 104 z 2 −0 .026 z−3 −0 .026 z3 ]
−5 −6 −4 −7 −3 −8 −2
= 0 . 75z −0 . 204z −0. 204 z −0. 104 z −0 . 104 z −0. 026 z −0 . 026 z H
M 1

 h( n) z 
2
n
 h(0)   zn
H( z) n 1
5
= h(0 )+ ∑ h(n ) [ z −n + z n ]
n=1

= 0 . 75+h(1 )[ z−1 +z 1 ]+h(2) [ z−2 +z 2 ] +h(3 ) [ z −3 +z 3 ] +h( 4 ) [ z−4 +z 4 ] +h(5 ) [ z−5 + z5 ]

= 0 . 75−0. 204z−1 −0 .204 z 1 −0 .104 z −2 −0 . 104 z 2−0. 026 z−3 −0 . 026 z 3
The causal filter coefficients are
h(0 )=h(1)=h (9)=h(10 )=0;
h(2)=h(8 )=−0. 026
h(3 )=h(7 )=−0 .104
h( 4)=h(6 )=−0. 204
h(5 )=0 . 75
M −1
2
jω
H (e )=∑ a(n)cos ωn
n=0

a(0 )=h ( M−1

2 )
=h(5 )=0 . 75

M−1
a(n )=2 h
[ 2 ]
−n =2 h(5−n)

a(1 )=2h (5−1 )=2h( 4 )=−0 . 408

15
 a(2 )=2h (5−2 )=2h(3 )=−0. 208
a(3 )=2h ( 5−3 )=2 h(2)=−0 .052
a( 4 )=2h (5−4 )=2h (1 )=0
a(5 )=2h ( 5−5 ) =2 h(0)=0
−
H (e jω )=0. 75−0 . 408 cosω−0 .208 cos 2 ω−0 . 052 cos 3 ω
ω(in degrees ) 0 1 2 3 4 5
−
H (e jω ) 0.082 0.0822 0.083 0.08433 0.08615 0.08848
jω
|H (e )|dB -21.72 -21.70 -21.61 -21.480 -21.29 -21.11

b) Using Hamming window:

The hamming window sequence is given by

2πn M−1 M −1
{ ( ) ( ) ( )
w Hm (n)=¿ 0 .54+0.46cos
M−1
; for −
2
≤ n≤
2
¿ ¿¿¿

πn
Hm {
w (n)=¿ 0 .54+0.46cos ( ); for −5≤n≤5 ¿ ¿¿¿
5
w Hm (0)=1
w Hm (1)=w Hm (−1)=0.912
w Hm (2)=w Hm (−2)=0.682
w Hm (3)=w Hm (−3)=0.398
w Hm (4 )=w Hm (−4)=0.1678
w Hm (5)=w Hm (−5)=0. 08
The coefficients using hamming window sequence are,
h(n)=h d (n)∗w Hm (n); −5≤n≤5
h(0 )=h d (0)∗w Hm (0)=1∗0. 75=0 .75
h(1)=hd (1)∗w Hm (1)=(−0.225 )∗(0.912 )=−0.2052
h(2 )=hd (2)∗wHm (2)=(−0.159 )∗(0.682 )=−0.1084
h(3)=h d (3)∗w Hm (3)=(−0.075 )∗( 0.398 )=−0.03
h( 4 )=hd ( 4 )∗w Hm (4 )=(0 )∗(0. 16787 )=0
h(5 )=h d (5 )∗w Hm(5)=(−0 .045 )∗(0.08 )=0. 0036
The transfer function of the filter is given by
M−1
2

=h ( 0)+ ∑ h( n ) [ z−n + zn ]
H ( z) n=1
5
= h(0 )+ ∑ h(n ) [ z −n + z n ]
n=1

=0 . 75+h(1 ) [ z−1 +z 1 ]+h(2) [ z−2 +z 2 ] +h(3 ) [ z −3 +z 3 ] +h( 4 ) [ z−4 +z 4 ] +h(5 ) [ z−5 + z5 ]

= 0 . 75−0. 2052z−1−0. 2052 z1 −0 . 1084 z−2−0. 1084 z 2 −0 .03 z−3 −0 . 03 z 3 +0 .0036 z −5 +0 .0036 z 5

16
 The transfer function of the realizable filter is

H'( z) =z
− ( M−1
2 )
H ( z)
= z−5 [ 0. 75−0 . 2052z−1 −0 .2052 z 1−0. 1084 z−2 −0 .1084 z 2 −0 . 03 z −3 −0 . 03 z3 +0 . 0036 z−5 +0 . 0036 z 5 ]
−5 −6 −4 −7 −3 −8 −2 −10
H ' ( z)= 0 .75z −0.2052z −0.2052 z −0.1084 z −0.1084 z −0.03 z −0. 03 z +0 .0036 z +0 .0036
The filter coefficients of causal filter are
h(0 )=h (10 )=0 . 0036;h(1)=h(9 )=0;h(2)=h(8 )=−0 . 03;h(3 )=h (7 )=−0 .1084;h(4 )=h(6 )=−0 . 2052;
h(5 )=0 . 75
M −1
2
H ( e jω )= ∑ a( n ) cos ωn
n=0

a(0 )=h ( M2−1 )=h (5)=0. 75

M−1
a(n )=2 h ( −n) =2 h(5−n )
2
a(1)=2 h(5−1 )=2 h( 4)=−0 . 4104
a(2 )=2 h(5−2)=2 h(3 )=−0. 2168
a(3 )=2 h(5−3)=2h (2)=−0. 06
a( 4)=2 h(5−4)=2 h(1)=0
a(5 )=2 h(5−5)=2h (0)=0 .0072
−
H (e jω )=0. 75−0 . 4104 cosω−0 . 2168 cos 2 ω−0. 06 cos 3 ω+0 . 0072cos 5 ω
ω(in degrees ) 0 30 60 90 120 150 180
−
0.07 0.28 0.7168 0.9668 1 1.003 1.0108
H (e jω )
jω -23.1 -11 -2.89 -0.29 0 0.028 0.093
|H (e )| dB
=====================================================================================

π π
H.W: 1. Design a filter with
{
H d (e jω )=¿ e−j3ω , − ≤ω≤ ¿ ¿¿¿
4 4 . Using a Hamming & Hanning
window with N=7

π π
2. Design a filter with
jω
H d (e )=¿ e{ −j5ω ,

.
− ≤ω≤ ¿ ¿¿¿
2 2
Using a Hamming & Hanning window
with N=11.
3. Design an FIR filter for the ideal frequency response using hamming window with N=7.

π
{
H d (e jω )=¿ e−j3ω , |ω|≤ ¿ ¿¿¿
8
For a FIR linear phase digital filter approximating the ideal frequency response

π
{
***************************************************************************************
H d (e jω )=¿ 1 , |ω|≤ ¿ ¿¿¿
6
Determine the coefficients of a 5 tap filter using rectangular window. 17
Solution:
Given:

π
{
H d (e jω )=¿ 1 , |ω|≤ ¿ ¿¿¿
Step 1: To find filter coefficient.
6 Hence
π
ωC = ; N =5.
6

1 π
hd (n ) = ∫ H (e jω )e jωn d ω
2π −π
π
1
= ∫ 2 π 1 . e jωn d ω
2π −2
sin (n−α)ω C
= ;
π (n−α ) ∵ α=0
π
sin n
6
hd (n ) = ;
πn −2≤n≤2
π
sin (0 )
6
=
For n=0: hd (0) π (0)
πn
sin
1 6
=
6 πn
6

hd (0 )
1
= =0 . 16
6
∴ lim sinθθ =1
θ−¿ 0

π
sin (1)
6 0.5
hd (1 )=hd (−1)= = =0 .159
for n=1 π (1 ) π
π (2 )
sin
6
hd (2 )=h d (−2 )= =0 .1379
For n=2; 2π
Step 2: Using Rectangular window:

M−1 M−1
w R (n )
=¿ 1 { ( ) ( )
for −
2
≤n≤
2
¿ ¿¿¿
= 1 for −2≤n≤2
w R (0 )=w R (1 )=w R (2 )=1
Step 3: To find filter coefficients using rectangular window are
h(n)=h d (n)∗w Hn (n) for −2≤n≤2

18
 h(0)=h d (0)∗w Hn (0)=(0.16)∗1=0.16
h(1)=hd (1)∗w Hn (1)=0.59∗1=0.59
h(2 )=hd (2)∗wHn (2)=0 .137∗1=0. 137
Step 4: The transfer function of the filter is given by
M−1
2
=h (0)+ ∑ [ h(n ) ( z n + z−n ) ]
H ( z) n=1
2
= 0 . 5+ ∑ h(n ) ( z n +z−n )
n=1

= 0 .16+h(1 )( z 1 +z −1 ) +h(2 )( z 2 +z −2 )
= 0 . 16+0 . 59 z 1 +0. 59 z−1 +0 .137 z −2 +0 .137 z 2
Step 5: The transfer function of the realizable filter is
( M −1 )
−
H ' ( z) =z 2
H ( z)
= z−2 [ 0 .16+0. 59 z 1 +0 .59 z−1 +0 .137 z−2 +0 . 137 z 2 ]
H ' ( z) =0 . 16 z−2 +0 .59 z−1 +0. 59 z−3 +0 . 137 z −4 +0 . 137
The filter coefficients of causal filter are given by
h(0 )=h (4 )=0 . 137 ,h(1)=h(3 )=0 . 59;h(2)=0. 1 6;h(5 )=0
M −1
2
jω = ∑ a( n )cos ωn
H (e ) n=0

a(0 )=h ( M2−1 )=h (2)=0. 32

M−1
a(n )=2 h ( −n)=2 h(2−n )
2
a(1)=2h(2−1)=2h(1)=1.18
a(2 )=2 h(2−2)=2 h(0 )=0 . 274
H (e jω )=0. 32+1 .18 cosω+0 .274 cos2 ω
0 30 60 90 120 150 180
ω(in degrees )
− 1.774 1.47 0.773 0.046 -0.407 -0.564 -0.586
H (e jω )
=====================================================================
Design an ideal band pass filter with a frequency response

π 3π
4 {
H d (e jω )=¿ 1 , ≤|ω|≤ ¿ ¿¿¿
4
Find the values of h(n) for N=11 using rectangular window.

Solution:
Given:

π 3π
{
H d (e jω )=¿ 1 , ≤|ω|≤ ¿ ¿¿¿
4 4
π 3 π Step 1: Filter coefficients are,
ωC 1 = and ω C2 =
4 4

19
 1
hd (n )= [ sin ωC 2 (n−α)−sin ωC 1( n−α )] for n≠α Step 2: Using rectangular window
π (n−α )
1 3 πn πn
=
πn [ sin
4
−sin
4]
For n=0;
hd (0 )=0 . 5
3 π (1 ) π (1)
sin −sin
4 4
hd (1 )=hd (−1)= =0
π (1)
3 π (2) π (2 )
sin −sin
4 4
hd (2 )=h d (−2 )= =−0. 3183
π (2)
3 π (3 ) π (3)
sin −sin
4 4
hd (3 )=h d (−3 )= =0
π (3 )
3 π (4 ) π(4)
sin −sin
4 4
hd ( 4 )=h d (−4 )= =0
π (4 )
3 π (5) π (5)
sin −sin
4 4
hd (5 )=h d (−5 )= =0
π (5 )

M−1 M−1
{ ( )( )
wR (n)=¿ 1 for − ≤n≤
2 2
¿ ¿¿¿ ¿
¿
Step 3: Filter coefficients using rectangular window
h(n )=w R( n)∗hd (n ); −5≤n≤5
h(0 )=w R( 0)∗hd (0 )=1∗0. 5=0 .5
h(1)=w R (1)∗hd (1)=1∗0=0
h(2 )=w R (2)∗hd (2 )=1∗−0 .3183=−0 .3183
h(3 )=w R (3 )∗h d (3)=1∗0=0
h( 4 )=w R (4 )∗hd ( 4 )=1∗0=0
h(5 )=w R (5 )∗h d (5)=1∗0=0
Step 4: The transfer function of the filter is
M−1
2
H (z )=h(0 )+ ∑ h( n) [ z−n +z n ] Step 5: The transfer function of the realizable filter is
n=1
=0 .5−0 . 3183z 2 −0 .3183 z−2

20
 H' (z )=z−5 ( 0 . 5−0 . 3183z 2 −0 .3183 z−2 )
=0 . 5z−5 −0 . 3183z−3 −0 .3183 z−7
The filter coefficients of the causal filers are
h(0 )=h (10 )=h( 9)=h(2)=h(8 )=h( 4 )=h(6 )=0
h(3 )=h(7 )=−0 . 3183
h(5 )=0 . 5
M−1
2
H (e jω )= ∑ a(n )cosωn
n=0

a(0 )=h (2M−1 )=h (5)=0. 5

M−1
a(n )=2 h ( −n)=2 h(5−n )
2
a(1 )=2 h(5−1)=2 h( 4 )=0
a(2 )=2 h(5−2)=2 h(3 )=−0. 6366
a(3 )=2 h(5−3)=2h (2)=0
a( 4 )=2 h(5−4 )=2 h(1 )=0
a(5 )=2 h(5−5)=2h (0)=0
−
H (e jω )=0. 5−0 .6366 cos 2ω
ω(in degrees ) 0 30 60 90 120 150 180
−
-0.1366 0.1817 0.818 1.1366 0.818 0.1817 -0.1366
H (e jω )
jω -17.3 -14.8 -1.74 1.11 -1.74 -14.8 -17.3
|H (e )| dB
====================================================================
Design an ideal band Reject filter with a frequency response Find the values of h(n) for N=11 using rectangular window.

π 2π1
{
H d (e jω )=¿ 1 , for|ω|≤ andhd|ω|≤
3
(n )= ¿ ¿¿¿ [ sin ωC 1 (n−α )−sin ωC 2 ( n−α )+sin (n−α ) π ]
3π (n−α )
1 πn 2 πn
(May/June-14)(Nov/Dec-12)(Nov/Dec-2011)
=
πn [sin −sin
3 3 ]
+sin nπ
Solution:
For n=0;
hd (0 )=0 . 667 Given:

πn 2 πn
sin −sin +sin nπ Step 1: Filter
3 3
hd (1 )=hd (−1)= =0 coefficients
π (1) are,
π 2π πn 2 πn
ωC 1 = and ω C2 = sin −sin +sin nπ
3 3 3 3
hd (2 )=h d (−2 )= =0 .2757 Step 2:
π (2)
πn 2 πn Using
sin −sin +sin nπ rectangular
3 3
hd (3 )=h d (−3 )= =0 window
π (3 )
πn 2 πn
sin −sin +sin nπ
3 3
hd ( 4 )=h d (−4 )= =−0 .1378
π (4 )
πn 2 πn
sin −sin +sin nπ
3 3 21
hd (5 )=h d (−5 )= =0
π (5 )
Step 3: Filter coefficients using rectangular window
h(n )=w R ( n)∗hd (n ); −5≤n≤5
h(0 )=w R ( 0)∗hd (0 )=1∗0. 667=0 . 667
h(1)=w R (1)∗hd (1)=1∗0=0
h(2 )=w R (2)∗hd (2 )=1∗0. 2757=0 .2757
h(3 )=w R (3 )∗h d (3)=1∗0=0
h( 4 )=w R (4 )∗hd ( 4 )=1∗−0 .1378=−0 .1378
h(5 )=w R (5 )∗h d (5)=1∗0=0

Step 4: The transfer function of the filter is

M −1
2
H (z )=h(0 )+ ∑ h( n) [ z−n +z n ]
n=1
=0 .667 +0 .2757 z 2 + 0. 2757 z−2 −0 .1378 z−4 −0 . 1378 z 4

22
Step 5: The transfer function of the realizable filter is
H'(z )=z−5 ( 0.667+0.2757z 2+0 .2757 z−2 −0.1378 z−4−0.1378 z 4 )
=0 .667 z −5 +0.2757 z−3+0.2757 z−7−0 .1378 z−9−0.1378 z−1
The filter coefficients of the causal filers are
h(0)=h (10)=h(2)=h(8)=h(4)=h(6)=0
h(1)=h(9)=−0.1378
h(3)=h(7)=0.2757
h(5)=0 .667
M−1
2
H (e jω )= ∑ a(n )cos ωn
n=0

a(0 )=h (2M−1 )=h (5)=0. 667

M−1
a(n )=2 h ( −n)=2 h(5−n )
2
a(1 )=2 h(5−1 )=2 h( 4 )=0
a(2 )=2 h(5−2)=2 h(3 )=0 . 5514
a(3 )=2 h(5−3)=2h (2)=0
a( 4 )=2 h(5−4 )=2 h(1 )=−0 . 2756
a(5 )=2 h(5−5)=2h (0)=0
−
H (e jω )=0. 667+0. 5514 cos 2 ω−0 . 2756 cos4 ω
ω(in degrees ) 0 30 60 90 120 150 180
−
0.9428 1.08 0.526 16 0.529 1.08 0.9428
H (e jω )
jω -0.5 0.67 -5.53 -15.9 -5.53 0.67 -0.5
|H (e )|dB
============================================================================
Design a high pass filter using window, with a cut-off frequency of 1.2 radians/sec and N=9. [Nov/Dec-2016]
Solution:
Given
 c  1.2radians / sec
if T =1 sec
c  cT  1.2 radians
The impulse response of a high pass filter with a cut off frequency c is
 sin c n
hd (n)  n 0


1- c for n=0

c  1.2

23
 1.2
hd (0)  1   0.618

 sin1.2
hd (1)  hd (1)   0.2966

 sin 2.4
hd (2)  hd (2)   0.1075
2
 sin 3.6
hd (3)  hd (3)   0.0469
3
 sin 4.8
hd (4)  hd (4)   0.0719
4
Hamming window for 4  n  4 is

2 n
wH (n)  0.54  0.46 cos for -4  n  4
8
n
=0.54+0.46cos
4
wH (0)  1
wH (1)  wH (1)  0.865
wH (2)  wH (2)  0.54
wH (3)  wH (3)  0.215
wH (4)  wH (4)  0.08
h(n)  h d (n) w H (n)
h(0)  (0.618)(1)  0.618
h(1)  h(1)  (0.2966)(0.865)  0.256
h(2)  h(2)  0.058
h(3)  h(3)  0.01
h(4)  h(4)  0.0057
The casual filter coefficients are
h(0)  (0.618)(1)  0.618
h(1)  h(1)  (0.2966)(0.865)  0.256
h(2)  h(2)  0.058
h(3)  h(3)  0.01
h(4)  h(4)  0.0057
=====================================================================
H.W: Design an ideal differentiator with frequency response H (e j )  j;      using hamming window with
N=8. (April/May-15)
=============================================================================
Frequency Sampling Method:( April/May-15)
Discuss the design procedure of FIR filters using frequency sampling method. [May/June-2013]

Generally, FIR filter can be specified by giving impulse response coefficients h(n) (or) DFT coefficients
H(k).

24
 N−1 j 2π kn
1 N
h(n )= ∑ H ( k)e
N k=0 -------------------------------> (1)
N−1 j 2 π kn
−
N
H (k )= ∑ h (n)e
n=0 ----------------------------------> (2)
H (k )=DFT samples
H (k )=H ( z )| j 2 π kn
z=e N
--------------------------------------> (3)
N−1
H ( z)= ∑ h(n )z−n
and n=0 --------------------------------------> (4)
N−1 N −1 j2 π kn

Put (1) in (4)

jω jω
H ( z )= ∑
n=0
[ 1
N
∑ H (k )e
k=0
N
] z −n

If z=e H (e )=H (z )
2 πk
ω k= ( ) N where ω k−−¿ sampling frequency .
General steps to design FIR filter using frequency sampling method [type-I design]:
Step 1: Draw the filter graph, as in FIR design using window function.
Step 2: Draw the unit circle and mark the points, if k-0,1…N-1.
o
360
θk = ∗k
N
0
360
θ1 = ∗1
if k=1; N
2πk
ω by
Step 3: To find H(k), replace N is the given equation.

jω −jαω
Hd(e )=¿ {e , 0≤ω≤ωC ¿ ¿¿¿ ¿
¿
Step 4: Find h(n)
If 'N' =odd
N −1

If 'N' =Even:
1
{ 2
h(n)= H (0 )+ ∑ 2 Re H (k )e N
N k=1
(
j2 π kn
) } ∴ H (0 )=1

{ }
−1 j2 π kn
2
1
h(n)= H (0)+ ∑ 2 Re H (k )e N
N
( )
k=1
Step 5: Find H(z):

25
 N−1
H ( z )= ∑ h(n )z−n
n=0

Design LPF which has the following specifications, N=7 using frequency sampling Technique. [Nov/Dec-2016][Nov/Dec-
15]

π
{
H d(e jω )=¿ e−j 3ω ,0≤ω≤ ¿ ¿¿¿
2
Solution:

π
Given:
jω
H d(e )=¿ e { −j3ω
,0≤ω≤ ¿ ¿¿¿
2
Step 1: From a unit circle, mark points from
o
360
θk = ∗k
N
θ1 =51. 420
θ2 =102. 80
θ3 =154 . 260
θ 4=205 . 680
θ5 =257 .1 0
θ6 =308 .52 0

Step 2:
N−1
H(k)=¿ {e
(−j
N ) πk
¿
for k=0,1,6 ¿¿¿
N=7.
6πk
H(k)=¿ e { −j
7 ¿
for k=0,1,6 ¿¿¿

26
 N−1

1
{
h(n )= H (0)+ ∑ 2 Re H (k )e N
N k =1
2
(
6 πk j 2 π kn
j2 π kn
) } ∴ H (0)=1

3
1
{ −j
= 1+ ∑ 2 Re e 7 . e 7
7 k=1
(
jπk( 2n−6)
)}
3
1
{
= 1+2 ∑ Re .e
7 k =1
7
}
3
πk (2 n−6 )
h(n )=
1
7 {
1+2 ∑ cos
k=1 7 ( )}
π (2 n−6 )
h(n )=
1
7 { (
1+2 cos
7 ( ))} for k=1 .

h(0 )=−0. 114;h(1 )=0 . 079;h(2)=0 . 321;h (3 )=0. 4286;h( 4 )=h(2 );h(5)=h(1);h(6 )=h (0)
=======================================================================

Solution:
Step 1:

N −1 N−1
(H(k)=¿{1.e ) ,k=0,12,3¿{0.4e ( ) ,k=4¿¿ ¿
−j πk
N
−j πk
N
Find coefficient of LP FIR with N=15 and it has symmetric unit sample response. It satisfies the following condition.

2πk
¿ H ()
15
=¿ {1 , k=0,1,2,3 ¿ {0.4 ,k =4¿¿¿¿
Step 2:
N−1

1
{
h(n )= H (0)+ ∑ 2 Re H (k)e
N k =1
2
(
j2 π kn
N ) } ∴ H (0)=1

j 14 πk j14 πk j2 π kn
h(n )=
1
15 { −
1+ ∑ 2 Re e 15
k =1
7
(( +0 . 4 e
−
15 )e )15
}
j 2πk ( n−7) j 2 πk

{ [∑ }
3
=
1
15
1+2 Re
k=1
15
e
] + 2( 0 . 4 ) ∑ e
k=4
15
(n−7)

1 2 π (n−7 ) 4 π (n−7 ) 6 π ( n−7 ) 8 π (n−7 )

=
15 { [
1+2 cos
15
+cos
15
+ cos
15
+0 .8 cos
15 ] ( )}
27
 h(0 )=−0.0141 ;h(1)=−0 .0195 ;h (2)=0 .04;h (3 )=0.0122
h( 4 )=−0 .0913;h(5)=−0 . 01809;h (6)=0 .313 ;h(7 )=0. 52.
========================================================================================
Determine the coefficients {h (n)} of a linear phase FIR filter of length M =15 has a symmetric unit sample response and a frequency

 2 k  1 for k  0,1, 2,3

Hr  
response that satisfies the condition  15  0 for k  4,5, 6, 7 (May/June-13) (April/May-11)(Nov/Dec-09)

Solution:

H (k)  1 for 0  k  3 and 12  k  14

=0 for 4  k  11
N 1
 (k)  ( ) k
N
=-14/15  k 0  k  7
and
14 k
 (k)=14 - for 8  k  14
15
H (k)  e  j14 k /15 for k=0,1,2,3
=0 for 4  k  11
=e  j14 (k 15)/15 for 12  k  14
 N 1 
1 2 
h(n)   H (0)  2  Re( H (k) e j 2 nk /15 ) 
n k 1 
 
1 7 
= 1  2  Re(e j14 k /15e j 2 nk /15 
5 k 1 
1 3 2 k (7  n) 
= 1  2  cos 
15  k 1 15 
1 2 (7  n) 4 (7  n) 6 (7  n) 
= 1  2cos  2cos  2cos
15  15 15 15 

`
h(0)  h(14)  0.05; h(1)  h(3)  0.041; h(4)  h(10)  0.1078
h(2)  h(12)  0.0666; h(3)  h(11)  0.0365; h(5)  h(9)  0.034

28
 h(6)  h(8)  0.3188; h(7)  0.466 .
Using frequency sampling method, design BPF with the following specifications. [May/June-2016]
Sampling frequency F=8000Hz
Cut off frequencies fc1=1000Hz
Cut off frequencies fc2=3000Hz Determine the filter coefficients for N=7.

Solution:
2 πf c 2 π ( 1000 ) π
ωc =2 πf c T = 1
= =
1 1 F 8000 4
2 πf c 2 π ( 3000 ) 3π
ωc =2 πf c T = 2
= =
2 2 F 8000 4
H ( K )=H d ( e jω )| 2π
ω= k
7 k=0,1,…6
0 for k=0,3
|H ( k )|
= { 1 for k=1,2
N−1 N −1
θ (k )
¿− ( ) N
π for 0≤k ≤
2
6
¿− πk for 0≤k≤3
7
0 for

|H ( k )| e
=
{for
− j 6 πk
7
k =0,3
k =1,2
The filter coefficients are given by
N−1

h(n)
1
[
= H (0 )+ ∑ 2. Re H (k). e
N k=1
2
[
j2 π kn
N
] ]
N−1

1
[
= 2 ∑ Re e
7 k=1
2

N−1
[
j6 πk
7
e
j 2 π kn
7 ] ]
2 2π
=
1
[
2 ∑ cos
7 k =1
4π
7
2
2 πk
( ( 3−n ) ) ]
7
cos ( 3−n ) +cos
7
=
[7
( 3−n ) ]
h(0)=h (6)=−0 .07928 .
h(1)=h(5)=−0. 321
h(2 )=h( 4 )=−0 . 11456
h(3)=0.57
*************************************************************************************
H.W:1. Design a linear phase FIR high pass filter using hamming window, with a cutoff frequency, ω c=0.8π
rad/sample and N=7.
2. Design a FIR low pass filter with cutoff frequency of 1KHz and sampling frequency of 4 kHz with 11 samples
using Fourier series method. Determine the frequency responses and verify the design by sketching the
magnitude responses

29
3. Using a rectangular window technique design a low pass filter with pass band gain of unity,  cutoff frequency of 1000Hz
and working at a sampling frequency of 5kHz.the length of the impulse response should be 7.
**********************************************************************************************

30