You are on page 1of 38

1

Introduction to DSP

Solutions of Examples for Practice


Example 1.4.7
Solution : The frequency of the sinusoid is,
W = 100 Hz
The sampling frequency is, fs = 240 Hz.
Here fs >2W

i.e.

240 > 200. Hence there is no aliasing.

The minimum sampling rate is given as,


fs 2W
1
1
We know that fs =
and W = . Then above equation becomes,
Ts
T
1
2
Here T is the period of x(t)

Ts
T
T
2
Thus sampling period should be less than or equal to half period of the signal.

Ts

Example 1.4.8
Solution : i) x( t ) = 25 e j 500 p t
w = 500 p

Here
\

2 p f = 500 p

\ Nyquist rate = 500 Hz

f = 250 Hz

ii) x( t ) = 1 + 0.1 sin (200 p t ) cos ( 2000 p t )


= cos (2000p t)+ 0.1 sin(200p t) cos (2000p t)
= cos (2000p t)+

0.1
[ sin(-1800p t) + sin( 2200 pt) ]
2

= cos (2000pt )- 0.005 sin(1800pt )+ 0.05 sin(2200pt )


Here
\

w max = 2200 p
f max = 1100 Hz

\ Nyquist rate = 2200 Hz


(1 - 1)
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

iii) x( t ) = 10 sinc 5t
= 10
Here

1-2

Introduction to DSP

sin 5p t
5pt

w = 5p

2p f = 5 p

f=

5
2

\ Nyquist rate = 5 Hz
iv) x(t) = 2 sinc (50 t) sin (5000 p t)
1
1
sin 50pt
= 2
[ cos 4950 p t - cos 5050 p t]
sin 5000 pt =
25 p t 2
50pt
Here

w max = 5050 p
2p f max = 5050 p

f max = 2525 Hz

\ Nyquist rate = 2 25 25 = 5050 Hz


Example 1.4.9
Solution : i) Here w 1 = 2000 p and w 2 = 4000 p
w1
w
2000 p
4000 p
= 1000 Hz and f 2 = 2 =
= 2000 Hz
=
2p
2p
2p
2p

f1 =

W = f max = f 2 = 2000 Hz

\ Nyquist rate = 2W = 2 2000 = 4000 samples/sec


Nyquist interval =

ii) Here

x (t) =

1
1
= 0.25 msec
=
2W
4000
sin 2 ( 4000 p t)
( p t) 2

(1- cos 8000 p t) / 2

w = 8000 p f = W =

( p t) 2
8000 p
w
= 4000 Hz
=
2p
2p

\ Nyquist rate = 2W = 2 4000 = 8000 samples/sec


Nyquist interval =

1
1
= 0.125 msec.
=
2W
8000

Example 1.4.10
Solution : i) g 1 ( t ) = sinc (200 t)
sin 200 p t
=
200 p t
Here
\

w max = 200 p
f max = 100 Hz
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1-3

\ Nyquist rate = 2 f max = 200 Hz


and Nyquist interval =

1
1
=
= 0.005 sec
200
2 f max

ii) g 2 ( t ) = sinc2 (200 t)


(1- cos 400 pt) / 2
sin 200 p t 2
=
=

200 p t
( 200 pt) 2
=
Here

1- cos 400 pt
2( 200 pt) 2

w max = 400 p

f max = 200

\ Nyquist rate = 2 f max = 2 200 = 400 Hz


And Nyquist interval =

1
1
= 2.5 msec
=
400
2 f max

iii) g 3 ( t ) = sinc (200 t) + sinc3 (200 t)


sin 200 p t sin 200pt 2
=
+
200p t
200 pt

sin 200 p t 1- cos 400pt


+
200p t
2 ( 200pt) 2

w 1 = 200 p

Here
\

and w 2 = 400 p

f 1 = 100 Hz and f 2 = 200 Hz

f max = 200 Hz

Nyquist rate = 2 f max = 2 200 = 400 Hz

and

Nyquist rate =

1
1
= 2.5 msec
=
400
2 f max

Example 1.5.6

Solution : i)

(-1) k d ( t - 2k)

k =-

The given signal can be expressed as,

- d(t - 2k) for odd values of k


(-1) k d(t- 2k) = d(t- 2k) for even values of k

k =-

The above signal is shown in the Fig. 1.1. Observe that the signal is
periodic with period T = 4

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Introduction to DSP

Digital Signal Processing Applications

1-4

d(t)
k=0

d(t + 4)
k = 2
6

Introduction to DSP

d(t 4)
k=2

2
5 4 3
1
d(t + 2)
k = 1

0 1

d(t 2)
k=1

d(t 6)
k=3

Fig. 1.1 Sketch of

(-1)k d (t - 2k )

k =-

ii) x( n) = (-1) n

= 1 ,
( n=0 )

-1 ,

( n=1)

1 ,

( n= 2 )

-1 ,

1 ,

( n= 3 )

( n= 4 )

-1 , .....
( n=5 )

This signal is periodic with period N = 2 samples


5

iii) x( t ) =

w ( t - 2k)

k =-5

5
x(t) = w (t 2k)
k = 5

11 10 9 8 7 6 5 4 3 2 1

2 3

(a)

x(t) = w (t 3k)
k =

13 12 11 10 9 8 7 6 5 4 3 2 1

2 3

9 10 11

k
(b)

9 10 11 12 13 k

Fig. 1.2 Sketches of example 1.5.6

From above Fig. 1.2 (a), observe that x(t) exists from k = 11 to 11 only. Hence it is
non periodic signal.

iv) x(t) =

w ( t - 3k)

k =-

Above signal is shown in the Fig. 1.2 (b). Observe that the signal is
periodic with period 3.
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1-5

Introduction to DSP

Example 1.5.7

Solution : i) x(n) = lm e jnp 4


np
np

, since e jq = cos q +j sin q


= lmcos + j sin

4
4
np
= sin
4
np
1
Compare this equation with x(n) = sin 2p f n, hence 2p f n =
f = cycles/sample. Since
4
8
k 1
f = = . There will be 8 samples in one period of DT sine wave.
N 8
np
Fig. 1.3 shows the waveform of x(n) = sin
and its even and odd parts are also shown.
4
Even and odd parts are given as,
1
Even part, x e (n) =
{x(n) + x(-n)} and
2
1
Odd part, x o (n) =
{x(n) - x(-n)}
2
Even part of x(n)
x(n)

Odd part of x(n)


x(n)
N=8

5
0 1

3 4

sin
x(n)

np
4
x(n)

xe(n) =

1
[x(n) + x(n)] =0
2

xo(n) =

1
[x(n) x(n)]
2
Odd part

Zero even part


n

Fig. 1.3 Even and odd parts of x (n )

t
0 t 1
ii) x(t) =
2- t 1 t 2
This is a triangular pulse as shown in the Fig. 1.4. It's odd and even parts are also shown
in the Fig. 1.4.
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1-6

Introduction to DSP

x(t)

x(t)

1
xe(t) =

0
1
[x(t) + x(t)]
2
Even part

0.5
2

1
xo(t) =

1
[x(t) x(t)]
2
Odd part

0.5
0

0.5

Fig. 1.4 Odd and even parts of traingular pulse

pt
iii) x(t) = cos 2
2

1+cos pt
1+cos 2q
, since cos 2 q =
2
2
1 1
= + cos( pt)
2 2

x(t) =

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

Here

2pft = pt f =

1-7

Introduction to DSP

1
or T = 2. The waveform x(t) and its odd and even parts are
2

shown in the Fig. 1.5.

1 cos p t
2
DC shift of
1
2

x(t)
1

T=2

x(t)
1

0
xe(t) =

1
[x(t) + x(t)
2
Even part

xo(t) =

1
[x(t) x(t)] = 0
2
[Zero odd part]
t

pt
Fig. 1.5 Odd and even parts of x(t) = cos2
2

Example 1.5.8
cos(pn)
Solution : i) x(n) =
0

for - 4 n 4
otherwise

This signal is given only for |n| 4. Hence it is not periodic. Let us calculate its energy
directly,
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

E =

x(n)

1-8

Introduction to DSP

cos 2 ( pn)

n=

n=4

Here cos (p n) = 1 for any value of n. Hence,


4

1 = 1 + 1 + 1 + 1 + 1 + 1 + 1 + 1 + 1 = 9 <

n =-4

Since an energy is finite and nonzero, this is an energy signal with E = 9 J.


cos ( pn) for n 0
ii)
x(n) =
otherwise
0
(-1)n
=
0

for n 0
otherwise

Let us calculate power for this function, since cos ( pn) is periodic function i.e.,
N
N
1
1
P = lim
|x(n)|2 = lim
|(-1) n|2

N 2 N +1
N 2 N +1
n =-N

lim

lim

1
2N +1
1
2N +1

n =-N

[(-1) 2 ]n by rearranging

n =-N
N

1 = lim

n =-N

1
2 N +1 = 1 W
2N +1

Since power is finite, this is power signal with P = 1 W.


Example 1.5.9
Solution : i) Power of the energy signal :
Let x(t) be an energy signal. The power is given by equation (1.5.13) as,
T/2
T/2

1
1
lim |x(t)|2 dt
P = lim
|x(t)|2 dt = lim

T T -T / 2
T T
T -T / 2

1
lim
|x(t)|2 dt Here limits are changed to -, as T

T T
-

1
lim E ,
T
T

since quantity inside brackets is E.

1
lim = 0
T T
Thus, power of the energy signal is zero over infinite time.
= 0E = 0,

since

ii) Energy of the power signal :


Energy of the signal is given by equation (1.5.11) as,
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1-9

Introduction to DSP

E =

|x(t)|2 dt

Let us change the limits of integration as -

T T
and take lim . This will not change
,
2 2
T

meaning of above equation.


i.e.,
T 2

E = lim
T -T 2

|x(t)|2

T 2

1
2

dt = lim T
|x(t)| dt
T

T
-T 2

by rearranging

T 2

1
= lim T lim
|x(t)|2 dt = lim T P since quantity inside brackets is P.

T
T
T
T -T 2

= By taking limits as T
Thus, energy of the power signal is infinite over an infinite time.
Example 1.5.10
pn
Solution : i) x(n) = 1+ cos
2
This signal is periodic, and it has a DC shift. It's frequency is,
pn
1 k
2p f n =
f= =
N= 4
2
4 N
Power is,

Here 1+ cos

P =

1
2N +1

| x(n)|2

n=- N

pn
pn
= 2 cos 2 , hence above equation becomes,
2
4
P =

1
9

2 cos 2

n =- 4

pn 2
4
= 9
4

cos 4

n =- 4

pn
4

3p
p
p
4 4
+ cos 4- + cos 4-
cos (- p)+ cos 49
4
2

p
p
3p
+1+ cos 4 + cos 4 + cos 4 + cos 4 (- p )
4
2
4

Since cos (- x) = cos (x),


P =

p
p
4
3p
+ 2 cos 4 + 2 cos 4
1+ 2 cos 4 (p )+ 2 cos 4

9
4
2
4
4
1
1 16
1+ 2+ + 0+ =
9
2
2 9
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1 - 10

Introduction to DSP

Example 1.7.3
Solution : i) x(3n - 1)

This waveform is obtained by precedence rule of time shifting and scaling. Fig. 1.6 (a)
shows the sketch of x(n).
Shifting : First x(n) is shifted/delayed by '1' sample to obtain x(n-1). This signal is
shown in the Fig. 1.6 (b).

Scaling : The time shifted signal x(n-1) of the Fig. 1.6 (a) is time scaled by 3 samples.
This means it is compressed by 3 samples. This signal gives x( 3n- 1). It is shown in the
Fig. 1.6 (c).
ii) y(1 - n)
Fig. 1.6 (d) shows y(n) and (e) shows y(-n). Then y(1- n) = y[-(n- 1)] is obtained by
delaying y(-n) by '1' sample. It is shown in the Fig. 1.6 (f).
3

x(n)
2

3 2 1

(d)

y (n)
(a)

1
2

3 2 1

1
2
x (n1)

3
2

2
1
2 1 0 1

y (n)
2

(b)

(e)
1

1
2

3 2 1
3
2
1

x (3n1)
3

y (1n)
2

(c)

1
2 1

(f)

0 1

2 1

1
1
2
3

Fig. 1.6 Sketches of x ( 3n - 1) , y (1- n )


TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1 - 11

Introduction to DSP

x (2n)
2

(a)

y (n4)

(b)

iii) x( 2n) +y( n - 4)


Fig. 1.7 (a) shows x( 2n). Note that x( 2n)
is obtained by time compression of x(n)
by the factor of '2'. Fig. 1.7 (b) shows
y(n- 4). The addition x( 2n) + y(n- 4) is
shown in the Fig. 1.7 (c).

1
1

3
4

1
2
3
3

x (2n) + y (n4)
2
1

1
2

(c)

2
3

2
3

Example 1.8.4

Fig. 1.7 Sketched of x ( 2n ) + y (n - 4)

Solution :
i) The system is memoryless, since y(n) depends upon x(n) i.e. present sample value
only.
ii) Since sine function is multivalued, it is not invertible. Hence system is noninvertible.
iii) System is causal, since y(n) depend on x(n) only.
sin 0
=1
0
Hence it is stable system.

iv) When x(n) 0, y(n) =

by L'Hospital's rule.

v) Response of system to delayed input by 'k' samples will be,


y (n, k) = T{x (n- k)} =

sin [x (n- k)]


x (n- k)

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1 - 12

Introduction to DSP

Let us delay the output by replacing 'n' by 'n k',


sin [x (n- k)]
y (n- k) =
x (n- k)
since y (n, k) = y (n- k) system is time invariant
vi) The output of the system to inputs x 1 (n) and x 2 (n) will be,
sin [ x1 (n)]
y1 (n) = T {x1 (n)} =
x1 (n)
y 2 (n)

= T

{x 2 (n)} =

sin [ x 2 (n)]

... (1)
... (2)

x 2 (n)

The linear combination of two outputs will be,


\
y3 (n) = a1 y1 (n)+ a 2 y 2 (n)
sin x1 (n)
sin x 2 (n)
= a1
By equations (1) and (2)
+ a2
x1 (n)
x 2 (n)
Now the output due to linear combination of two inputs will be,
sin [ a1 x1 (n)+ a 2 x 2 (n)]
y 3 (n) = T {a1 x1 (n)+ a 2 x 2 (n)} =
a1 x1 (n)+ a 2 x 2 (n)

... (3)

... (4)

Here from equations (3) and (4),


y3 (n) y 3 (n), the system is nonlinear.

The presence of 'sine' function makes the system nonlinear.


Example 1.8.5
Solution : y(n) = x( n)

Output is magnitude of present input, hence the system is static or memoryless.


Magnitude operation is not linear. Hence the system is nonlinear.
Delaying the input by ' k' samples, output will be,
y (n, k) = T {x (n- k)} = x (n- k)
And the delayed output will be,
y (n- k) = x (n- k)
Since y (n, k) = y (n- k), the system is shift invariant.
Output is magnitude of present input. Hence the system is causal.

As long as x(n) is bounded, its magnitude will also be bounded. Hence the system is
stable.
Example 1.8.6
Solution : i) Dynamic system, since output depend upon previous inputs.
ii) Causal system, since output depends upon present and past inputs.
iii) Stable system, since y(n) is bounded as long as x(n) is bounded.
iv) Time invariant, since y(n) is not direct function of time 'n'.
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1 - 13

Introduction to DSP

Example 1.8.7
Solution :
i)

Memory : It is dynamic (requires memory) since the term x(n+1) is present.

ii) Stability : Value of cosine function is between (1, 1). Hence y(n) is bounded as
long as x(n) is bounded. Hence the system is stable.
iii) Causality : The system is noncausal since x(n+1) requires future input.
iv) Linearity : Presence of 'cosine' function makes the system nonlinear.
v) Time invariance : System is time invariant, since there is no time index
manipulation.
Example 1.8.8
Solution : i) y( t ) = sin [ x( t +2)]
This system is not memoryless, noncausal, stable.
y1 (t) = sin [x1 (t + 2)]
y 2 (t) = sin [x 2 (t + 2)]
Linear combination of two outputs,
y 3 (t) = a 1 y 1 (t) + a 2 y 2 (t) = a 1 sin [x 1 (t + 2)]+ a 2 sin [x 2 (t + 2)]
Response to linear combination of inputs,
y3 (t) = f [a 1 x 1 (t) + a 2 x 2 (t)] = sin [a 1 x 1 (t + 2) + a 2 x 2 (t + 2)]

Here y 3 (t) y 3 (t) hence system is nonlinear.


Response to delayed input will be,
y(t , t 1 ) = f [x(t - t 1 )] = sin [x (t + 2- t 1 )]
Let us delay an output itself,
y(t - t 1 ) = sin [x (t + 2- t 1 )]
Here y(t , t 1 ) = y(t - t 1 ), hence this is time invariant system.
ii) y( n) = x[ 2- n]
This system is not memoryless, noncausal and stable.
y 1 (n) = x 1 [2- n]
y 2 (n) = x 2 [2- n]
Linear combination of the two outputs,
y 3 (n) = a 1 y 1 (n) + a 2 y 2 (n) = a 1 x 1 ( 2- n) + a 2 x 2 ( 2- n)
Response to linear combination of inputs,
y3 (n) = f [a 1 x 1 (n) + a 2 x 2 (n)] = a 1 x 1 ( 2- n) + a 2 x 2 ( 2- n)
Here y 3 (n) = y3 (n) hence the system is linear.
Response to delayed input by 'k' samples will be,
y(n, k) = f [x (n- k)] = x( 2- n- k)
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1 - 14

Introduction to DSP

Let us delay an output itself by 'k' samples,


y(n- k) = x [2- (n- k)] = x [2- n+ k]
Here y(n, k) y(n- k), hence this system is time variant.
Example 1.10.8
Solution :
x 1 (n) = 1 1 1
x 2 (n) = 1 1 1
_________________
111
1 1 1
1 1 1
_________________
x 1 (n) * x 2 (n) 1 2 3 2 1

Example 1.10.9
Solution : i) y ( n) = b n u ( n) * u ( n - 3)

y(n) =

h ( k) x (n- k) =

k =-

Here

u (k - 3)

bn-k u (n- k)

k =-

u ( k- 3) = 1 for k 3 and u (n- k) = 1 for n k, hence

u ( k - 3) u (n- k) = 1 for 3 k n.
n

y(n) =

bn-k =

k= 3

bn b- k = bn

k= 3

=
1-

1
b

1
b

1 k

k= 3

1 3 1 n+1
-
b b
n
= b
since
1
1

bn-3 -

N2

k= N1

ak =

a N 1 - a N 2 +1
1- a

bn-2 - 1
b- 1

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1 - 15

Introduction to DSP

d ( n- 4 p)

ii) y ( n) = u ( n) *

p=0

y(n) = u (n) *

d (n- 4 p)

p=0

u (n) * d (n- 4 p)

p=0

We know that x (n) * d (n- n0 ) = x (n- n0 ) from properties of impulse function


Therefore

u (n) * d (n- 4 p) = u (n- 4 p) . Then above equations will be,

y(n) =

u (n- 4 p)

p= 0

Example 1.10.10
Solution :

Convolution is given as,

x(n) * y(n) =

x (k) y (n- k)

k =-

Here

nxl = 2

nyl = -1

and

and
similarly
nxh = 1
nyh = 2
The index of summation will be,

(nxl +nyl )n (nxh +nyh )

( 2 1) n (1 + 2)

i.e. 3 n 3

x(n) * y(n) =

x ( k) y(n- k)

and - 3 n 3

k =-

The value of n is varied from 0 to 3 in Fig. 1.8 (c) (d) (e) and (f) respectively. The
corresponding output is shown in Fig. 1.8 (j), (k), (l) and (m). For n 4, there is no overlap
between x(k) and y (n k) Hence convolution becomes zero.
The index is varied from n = 1 to 3 in Fig. 1.8 (g), (h) and (i) respectively. The
corresponding output is shown in Fig. 1.8 (n) (o) and (p). For n 4, there is no overlap
between x(h) and y (n k). Hence convolution becomes zero.
Thus the convolution of two sequences becomes, x(n) * y(n) = {1, 2, 1, 0, 1, 2, 1}

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1 - 16

Introduction to DSP

x(k)
(a)
0

2 1

y(k)
(b)
1

1 0
y(k)

n=0
2 1

2 1

y(1k)

n=1
1

x(k) y(2k)

1
0

k
x(k) y(3k)

(f)

(m)

x(k) y(1k)

y(1k)
(n)

(g)
0

3 2

(h)

4 3

(o)
k

2 1

S x(k)y(2k)
= 0 + 0 + 1+ 1 + 0 + 0
=2

S x(k)y(3k)
= 0 + 0 + 0+ 1 + 0 + 0 + 0
=1

4 3 2 1
x(k) y(3k)

y(3k)

n = 3

S x(k)y(1k)
= 0 1 +1 + 1 + 0
=1

x(k) y(2k)

y(2k)

n = 2

S x(k)y(3k)
=0+0+01+0+0+0
k = 1

3
n = 1

S x(k)y(2k)
= 0 + 0 1 1 + 0 + 0
= 2

(l)
k

y(3k)
0

S x(k)y(1k)
= 0 1 1 + 1 + 0
= 1

(e)

n=3

(k)

y(2k)

n=2

x(k) y(1k)
(d)

S x(k)y(k)
= 1 1 + 1 + 1
=0

x(k) y(k)
(j)

(c)

(p)

(i)

5 4

3 2 1 0

5 4 3 2 1

Fig. 1.8 Convolution of x(n) and y(n) using graphical method

Example 1.12.5
Solution : i) Stability
Here
h(n) = ( 0.99)n u(n+ 3) = ( 0.99)n for n-3

Consider

|h(k)|

k = -

( 0.99) k

k= -3

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

1 - 17

Introduction to DSP

= (0.99) - 3 +(0.99) -2 +(0.99) -1 + ( 0.99) k


k= 0

= (0.99) - 3 +(0.99) -2 +(0.99) -1 +

ak
k= 0

1
1- a

= 103 <

since

1
, since
1- 0.99

|h(k)| is finite, the system is stable.

k = -

ii) Causality
This system is non causal since h(n) 0 for n < 0.
Example 1.12.6
Solution : i) Causality : Since u(-n) is present, h(n) 0 for n < 0. Hence this system is
noncausal.

k = -

k = -

1 k

1 k

|h(k)| = 2 u(-k) = 2

ii) Stability : Consider,

k= 0

1 -k

By adjusting the sign

k= 0

1 -k
Since = ( 2-1 ) -k = 2 k
2

2k

k= 0

Since impulse response is not absolutely summable, the system is unstable.


iii) Step response
Here x(n) = u(n). Hence calculate step response,
n

y(n) =

h(k)

k = -

1 k

2 u(-k)

k = -

Here u(-k) is equal to '1' from - to 0. Hence two cases will be possible in above
equation. If n > 0, upper limit of 'k' will be '0'. If n < 0, upper limit of 'k' will be 'n'
0

Case - I : For n > 0, y(n) =

1 k

k=

-k

By adjusting the sign

k= 0

2k

k= 0
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications


n

Case - II : For n < 0 y(n) =

1 - 18

Introduction to DSP

1 k

k = -

1 -k

By adjusting the sign

k = -n

2k

k = -n

Thus the step response is "infinity" since the system is unstable.

qqq

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

DTFT, DFT and FFT

Solutions of Examples for Practice


Example 2.1.7
Solution : The given difference equation is,
1
1
... (1)
x (n) + x (n- 1)
y (n) =
2
2
The linear convolution of unit sample response h (n) and input x (n) gives output y (n). i.e.,

y (n) =

h ( k) x (n- k)

k =-

This equation can be expanded as,


y (n) = .... + h (-1) x (n + 1) + h ( 0) x (n) + h (1) x (n- 1) + h ( 2) x (n- 2) + ....
On comparing above equation with equation (1) we find that,
1
1
and
h (1) =
2
2
And all the other terms are absent, hence they are considered zero. Consider,
h ( 0) =

H ( w) =

h ( k) e -jwk

k =-

This is Fourier transform of unit sample response and it is called transfer function. Putting
for h ( 0) and h (1),
H ( w) = h ( 0) e -jw 0 + h (1) e -jw 1
1 1 -jw
1
=
+ e
(1 + e -jw ) = 12 (1 + cos w - j sin w)
2 2
2
1
(1 + cos w) - j 12 sin w
=
2
1
w
\ Real part of H( w) H R ( w) = (1 + cos w) = cos 2
2
2
=

Imaginary part of H( w) H I ( w) =\

Magnitude of H( w) H( w) =

1
w
w
sin w = - sin cos
2
2
2
H R2 ( w) + H I2 ( w)
(2 - 1)
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

... (2)

Digital Signal Processing Applications

2-2

w
w
w 2
+- sin cos =
2
2
2

cos 4

w
w
w
cos 2 + sin 2 =
2
2
2
H ( w)
Phase of H ( w) = H( w) = tan-1 I
H R ( w)
=

-1

cos 4

w
w
w
+ sin 2 cos 2
2
2
2
w
w
= cos
2
2

... (3)

w
w

- sin 2 cos 2
w
-1 - tan w
=
= tan
2
2

2 w

cos

... (4)

cos 2

= tan

DTFT, DFT and FFT

cos 2

Table 2.1 shows the calculations of magnitude and phase of H ( w) for few values of w .

Magnitude
w
H( w) = cos
2

-p

p
2

2p
3

0.5

p
3

p
2

1
2

p
4

p
3

3
2

p
6

p
3

3
2

p
6

p
2

1
2

p
4

2p
3

0.5

p
3

p
2

Phase H( w) =-

w
2

Table 2.1 Calculation of H (w) and H (w)

Fig. 2.1 (a) shows the magnitude and Fig. 2.1 (b) shows the phase plot of transfer function
based on calculations in above table.
Here note that transfer function H ( w ) is the continuous function of ' w' in the range of
-p w p . Hence magnitude and phase plots in this figure are continuous.

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

2-3

DTFT, DFT and FFT

|H(w)|
1
(a) Magnitude plot
of the transfer
function

H(w)
p
2

p
(b) Phase plot
of the transfer
function

p
2

Fig. 2.1 Magnitude and phase plots of the transfer function H ( w) =

1
(1+ e- jw)
2

Comments on magnitude and phase plots of H ( w) :


1. The magnitude and phase plots are continuous function of ' w' for nonperiodic
sequences.
2. The magnitude plot is even symmetric around w= 0, where as phase plot has odd
symmetry. i.e.,

H ( w) = H (- w)

... (5)

and
H (-w) = - H ( w )
3. The magnitude and phase plots are periodic with period 2p . Readers can verify this
statement by actually calculating magnitude and phase transfer functions of the
preceding example.
Example 2.2.8
Solution : To obtain DFT
X4 = [W4 ] x 4

1
1
1 1 2
1
X( 0)
1 -j -1
X(1)
j 1 1- j

=
=

1 -1 1 -10 0
X( 2)

j -1 -j 0 1+ j
1
X( 3)
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

2-4

DTFT, DFT and FFT

To obtain IDFT
xN =
\

x4 =

1
[WN* ] XN
N
1
[W * ] X4
4 4
1 2
4 1

4 1
- j 1- j
= 1 =

4 0 0
-1 0


j 1+ j
0 0

1
1
1
x( 0)
1
x(1)
j -1
= 1

4 1 -1
x( 2)
1

1 - j -1
x( 3)

Thus the original sequence is obtained back.


Example 2.2.9
Solution : Here bandwidth, W = 4 kHz
Resolution = 50 Hz
i) To obtain minimum sampling rate
By Nyquist theorem,
Fs = 2W = 2 4 kHz = 8 kHz
ii) To obtain minimum number of DFT samples
Let the DFT samples be N. These samples are equally spread over the frequency range of
0 to 2p. In analog domain 2p corresponds to fs = 8000 Hz. In other words 'N' samples will
be equally spread over 0 to 8000 Hz. Since minimum spacing between the samples
(Resolution) should be 50 Hz,
50

8000
N

It is given that N = 2
Thus,

or
m

8000
= 160 samples.
50

point DFT is used. Hence, let us calculate value of m for N = 160.

m = log 2 N =

log 10 N
log 10 160
=
= 7. 322
log 10 2
log 10 2

Here 'm' must be taken nearest higher integer.


\
m 8
\

N = 2m = 2 8 = 256

samples.

Thus the DFT must have 256 samples.


iii) To obtain minimum duration
Since 8000 samples are taken in 1 sec. 256 samples will require,
Minimum duration =

256
= 32 msec.
8000

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

2-5

DTFT, DFT and FFT

Example 2.4.8
Solution : X(0) = 0.25
X(1) = 0.125 j 0.3018
X(2) = 0
X(3) = 0.125 j 0.0518
X(4) = 0
For real valued sequence, X(N k) = X * ( k)
\
\

X( 8- k) = X* ( k)
X(5) = X( 8- 3) = X *( 3) = 0.125 + j 0.0518
X(6) = X( 8- 2) = X *( 2) = 0
X(7) = X( 8- 1) = X *( 1) = 0.125 + j 0.3018

Example 2.5.3
Solution : Here, h(n) =

{2,

2, 1}

i.e.

M= 3

Input segments length is N = 8.


Since
N = M + L -1
8 = 3+ L - 1 L = 6
Let us form the segments of input sequence as follows :

x 1 (n) = 3, 0, - 2, 0, 2, 1,
0, 0
{
3

14442444
M -1 zeros
' L'samples of x( n)
Thesetwozeros

are appered

tomake ' L' samples

678

0 , 0 ,
x 2 (n) = 0, - 2, - 1, 0,
3
144444244444
' L'samples of x( n)

And

0, 0
123
M -1 zeros

h(n) = 2, 2, 1 , 0, 0, 0, 0, 0
424
3 1442443
1
L-1 zeros

' M'samples

Now let us calculate


y 1 (n) = x 1 (n) 8

h (n)

y 2 (n) = x 2 (n) 8

h (n)

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

2-6

DTFT, DFT and FFT

0
0
1
2
0 -2 0 2 6
y 1 (0)
3
( )
0
3
0
0
1
2
0 -22 6
y
1
1

y 1 (2)
-2 0
3
0
0
1
2
0 1 -1
( )

3
0
0
1
2 0 -4
y 1 3 = 0 -2 0
=
y 1 (4)
2
0 -2 0
3
0
0
1 0 2

2
0 -2 0
3
0
0 0 6
y 1 (5)
1
y (6)
0
1
2
0 -2 0
3
0 0 4
1

0
1
2
0 -2 0
3 0 1
0
y 1 (7 )
Similarly,
y 2 (n) = { 0, 2, 1, 0, 0, 0, 0, 0 } 8
=

{ 0,

{ 2,

2, 1, 0, 0, 0, 0, 0 }

4, 6, 4, 1, 0, 0, 0 }

Last M 1 samples of y 1 (n) are added to first M 1 sample of y 2 (n).


y1 (n)

y 2 (n)
y (n)

Actually y (n) should contain samples of x (n) + samples of h (n) 1 = 12. The last two zero
in above y (n) are obtained due to two zeros appended in x 2 (n). Hence last two zeros can
be discarded.
Thus the output will be,

y (n) = 6, 6, 1, 4, 2, 6, 4, 3, 6, 4, 1, 0

qqq

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

z-Transform

Solutions of Examples for Practice


Kept this unsolved example for student's practice.

Example 3.1.7
Example 3.1.8

1 n
Solution : i) y [n ]= u [n ]
2
1

We know that anu (n)


\

or

1 az 1

, z > a

1 n
z
u (n)
2
1
1 1
1 z
2

Y(z) =

1
1
, z >
1 1
2
1 z
2
1
, z >
2

ii) y [n ]= n x [n ]
z2

We have x (n)

z 2 16

, z <4
z

Differentiation in z-domain property states : n x (n) z


\

n x (n) z

dz z 2 16

d
X(z)
dz

du
dv

32 z
d u v dx u dx
, Here =
z .
dxv
2 16 2
v2

)
(z
z

or

Y(z) =

32 z 2

(z 2 16) 2
32 z 2

(z 2 16) 2

, z <4
(3 - 1)
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

3-2

z-Transform

Example 3.2.9
1 n
1 -n
Solution : x(n) = - u(n) + 5 u(-n - 1 )
5
2
Rearranging the given x(n),
-n
1 n
1
x(n) = - u(n) + 5(2-1 ) u( - n- 1) , Here = 2-1
2
5
= (- 0.2) n u(n) + 5(2) n u (-n- 1)
= (- 0.2) n u(n) - 5[-2n u(-n- 1)]
By using standard z-transform pairs,
X(z) =

1
1+ 0.2 z-1

for|z| > 0.2 and|z| < 2

1- 2 z-1

or ROC : 0.2 <|z| < 2


Example 3.3.9
Solution :

H(z) = (1- z-1 ) (1- z-6 )


= 1- z-6 - z-1 +z-7
= 1- z-1 - z-6 +z-7

We know that H(z)= h(n)z-n , comparing above equation with this H(z) we get,
n=0

h(n) =

{1,

-1, 0, 0, 0, 0, -1, 1}

Example 3.3.10
Solution : Step 1 : Converting X(z) to its positive powers,
X(z) =
\

Where

z 2 - z+ 0 . 5

z (z+1)
z 2 - z+ 0 . 5

X(z)
z+1
= 2
z
z - z+ 0 . 5
=

Step 2 :

z 2 (1+z-1 )

X(z)
=
z

z+1
1
1 1
1
z- - j z- + j
2
2
2
2

A1
A2
+
1
1
1
1
z- - j
z- + j
2
2
2
2

1
1 X(z)
A1 = z- - j
1
1
2
2 z

z= + j
2

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

3-3

z+1
1
3
= -j
2
2
1
1
z- + j
2
2 z= 1 + j 1
2

A2

and

1
1 X(z)
= z- + j
2 z
2

1
1
z= - j
2
2

z+1
1
3
= +j
2
2
1
1
z- - j
2
2 z= 1 - j 1
2

Key Point

z-Transform

Here note that A 1 = A 2* . This is always true. Hence value of only A1 is

sufficient to determine A 2 .
1
3
1
3
-j
+j
2
2 + 2
2
1
1
1
1
z- - j
z- + j
2
2
2
2

X(z)
=
z

Step 3 :

1
3
1
3
-j
+j
2
2
2
2
X(z) =
+
1
1
1
1
1- + j z-1 1- - j z-1
2
2
2
2

Step 4 : Here causal sequences are assumed. Taking inverse z-transform of above equation,
3 1
1 n
3 1
1 n
1
1
x(n) = - j + j u (n) + + j - j u (n)
2
2
2 2
2
2 2
2
Simplification of above equations : Convert all the term to their polar form. It can be
converted with the help of P-R function on calculator. i.e.,
x(n) = (1.58 71.56) (0.707 45)n u(n) + (1.58 71.56) (0.707 45)n u(n)
= 1 . 58 e - j 71.56 ( 0 . 707 e j 45 ) n u (n) +1 . 58 e j 71.56 ( 0 . 707 e - j 45 ) n u (n)
= [1 . 58 e - j 71.56 ( 0 . 707) n e j 45n +1 . 58 e j 71.56 ( 0 . 707) n e - j 45n ] u (n)
= 1 . 58 ( 0 . 707) n [e j ( 45n-71.56 ) + e -j ( 45 n -71.56 ) ] u (n)
= 1 . 58 ( 0 . 707) n 2 cos ( 45 n- 71 . 56) u (n) By Euler's identity
= 3 . 16 ( 0 . 707) n cos ( 45 n- 71 . 56 ) u (n)
Example 3.5.14
Solution : i) y( n) = y( n - 1) - 0. 5y( n - 2) + x( n) + x( n - 1)
\

Y(z) = z-1Y(z)- 0 . 5z-2 Y(z)+X(z)+z-1X(z)

H(z) =

Y(z)
1+z-1
=
X(z) 1- z-1 + 0 . 5z-2
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

z-Transform

z 2 +z

Im(z)

z 2 - z+ 0 . 5
z(z+1)

p1

(z- 0 . 5- j0 . 5)(z- 0 . 5+ j0 . 5)

Zeros : z 1 = 0,
Poles :

3-4

z 2 =-1

p 1 = 0 . 5+ j0 . 5,

z2

z1

p 2 = 0 . 5- j0 . 5

0.5

Re(z)

p2

Fig. 3.1 shows the pole zero plot.

Fig. 3.1 Pole zero plot of (i)

ii) y( n) = 0.7y( n - 1) - 0.1y( n - 2) + 2x( n) - x( n - 2)


Y(z) = 0.7z-1Y(z)- 0.1z-2 Y(z)+ 2X(z)- z-2 X(z)
\
\

H(z) =

=
Zeros : z 1 =

1
,
2

Poles : p 1 = 0 . 5,

Y(z)
2- z-2
=
X(z) 1- 0.7z-1 + 0.1z-2
2 z2 -1
z 2 - 0.7z+ 0.1

z 2 - 0.7z+ 0.1

1
1
z+
z

2
2

1
2

y(z) = z X(z)

{3 ,

2, 1, 0, 1, 2,

3}

X(z) = 3z 3 + 2 z 2 +z+ 0+z-1 + 2 z-2 + 3z-3

Y(z) = z{ 3z 3 + 2 z 2 +z+ 0+z-1 + 2 z-2 + 3z-3 }


= 3z 4 + 2 z 3 +z 2 + 0+1+ 2 z-1 + 3z-2
y(n) =

Re (z)

Fig. 3.2 Pole zero plot of (ii)

1
0.2 0.5 2

p2 = 0 . 2

Example 3.5.15
x(n) =

1
2

Fig. 3.2 shows the pole-zero plot.


Solution :

p 2 p 1 z1

z2

(z- 0 . 5) (z- 0.2)

z 2 =-

Im(z)

1
2 z 2 -
2

{3 ,

2, 1, 0, 1, 2,

3}
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

3-5

z-Transform

Example 3.5.16
Solution : First let us convert powers of z in H (z) to positive values. i.e.,
H (z) =

=
\

z2
z2

3 - 4 z-1
1 - 3.5 z-1 + 1.5 z-2

z ( 3z - 4)
z2

- 3.5 z + 1.5

H (z)
3z - 4
= 2
z
z - 3.5 z + 1.5

... (1)

The denominator polynomial is quadratic (i.e. second order). Hence it will have two roots.
They are given as,
-b b 2 - 4 ac
2a
Here b =- 3.5, c = 1.5 and a = 1.
p1 , p2 =

p1 , p2

3.5 (3.5) 2 - 4 (1) (1.5)


=
2 (1)
3.5 2.5
=
2

= 3, 0.5
Thus p 1 = 3 and p 2 = 0 . 5. Hence we can write equation (1) as,
H (z)
3z- 4
=
\
z
(z- 3) (z- 0.5)
A1
A2
+
z- 3 z- 0.5
A1 and A2 can be obtained as follows :
H (z)
A1 = (z- 3)
z z= 3
=

=
and

3z- 4
z- 0.5

z= 3

A2 = (z- 0.5)
=

=2

H (z)
z

z = 0.5

3z- 4
=1
z- 3 z = 0.5

Thus the partial fraction expansion is,


H (z)
2
1
=
+
z
z- 3 z- 0.5
The system has two poles and they are at p 1 = 3 + j 0 and p 2 = 0 . 5 + j 0. Thus both the
poles have real part only. We can further rearrange above equation as,
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

H (z) =
=

3-6

z-Transform

2z
z
+
z- 3 z- 0.5
2
1- 3 z

-1

... (2)

1 - 0.5 z-1

i) To determine ROC and h (n ) for stable system :


Since the system is stable, the ROC of H (z) must include the unit circle. The system has
one pole at p 1 = 3 and second pole at p 2 = 0 . 5. Hence the ROC must be
ROC : 0.5<||
z < 3 which includes||
z = 1 i.e. unit circle.
Thus the above specified ROC of 0.5 <||
z < 3 includes the unit circle of ||
z = 1 but does not
include the poles at p 1 = 3 and p 2 = 0.5. Since the ROC is the ring or annular region, the
unit sample response h (n) will be two sided. Since 0 . 5 <||
z or ||
z > 0 . 5, the second term of
equation (2) will correspond to causal part of h (n). Hence,

1
= ( 0.5) n u (n)
for ROC :||
IZT
z > 0.5
1
1 - 0.5 z

Similarly since ||
z < 3, the first term of equation (2) will correspond to anticausal part of
h (n). Hence,

1
= - ( 3) n u (-n- 1) for ROC :||
IZT
z<3
1

1
3
z

Hence the unit sample response can be obtained as inverse z-transform of H (z) of
equation (2). i.e.,
h (n) = IZT {H (z)}
= -2 ( 3) n u (-n- 1) + ( 0.5) n u (n)
This is required unit sample response for stable system.
ii) Causal system :
The poles of this system are at p 1 = 3 and p 2 = 0.5. For the causal system the ROC is the
exterior of the circle. The ROC should not contain any poles. Hence ROC will be ||
z > 3.
The unit sample response can be obtained by taking inverse z-transform of equation (2)
i.e.,

2
1
+ IZT

h (n) = IZT {H (z)} = IZT


1
1
1- 3 z

1- 0.5 z

Since ROC is ||
z > 3 and causal system we can obtain the inverse z-transform of above
equation as,
h (n) = 2 ( 3) n u (n) + ( 0.5) n u (n)
Here since ROC is ||
z > 3, it does not include unit circle of ||
z = 1. Hence this system is
unstable.

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

3-7

z-Transform

iii) Anticausal system :


The poles of the system are at p 1 = 3 and p 2 = 0 . 5. We know that ROC is the interior of the
circle for anticausal system. Hence the ROC will be ||
z < 0.5. Here note that ROC won't be
||
z < 3, since it is also the exterior of ||
z > 0.5. Therefore ROC of |z| < Magnitude of pole
nearest to origin is considered. The unit sample response h (n) can be obtained by taking
inverse z-transform of equation (2). i.e.,

2
1
+ IZT

h (n) = IZT {H (z)} = IZT


1
1

1- 3 z

1- 0.5 z

For ROC of||


z < 0.5 we obtain inverse z - transform of above equation as follows,
h (n) = -2 ( 3) n u (-n- 1) - ( 0. 5) n u (-n- 1)
Since the ROC of this system is for ||
z < 0 .5, it does not include unit circle at ||
z = 1. Hence
this system is unstable.
Example 3.5.17
Solution : The given difference equation is,
1
y (n)- y (n- 1) = 2 x (n) , y (- 1) = 0
2
Whenever initial conditions are given we must use unilateral z-transform.
Here since y (- 1) = 0, we can use z-transform also. Taking unilateral z-transform of above
equation,
1 -1
Y (z)z Y (z)+ y (- 1) = 2 X (z)
2

Putting y (- 1) = 0, above equation becomes same as z-transform,


Y (z)-

1 -1
z
Y (z) = 2X (z)
2
H (z) =

Y (z)
X (z)

2
1 -1
1- z
2

1
n
= (p k ) u (n), hence we get,
Here we use IZT
1

1- p k z
=

1 n
h (n) = 2 u (n)
2
This is the unit sample response of the given difference equation.

qqq

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Filter Structures

Solutions of Examples for Practice


Example 4.3.4

Kept this unsolved example for student's practice.

Example 4.3.5

Kept this unsolved example for student's practice.

Example 4.4.6
Solution :
i) y ( n) = 0. 5 (x(n) + x(n - 1 ) = 0. 5 x ( n) + 0. 5x ( n 1)
Fig. 4.1 (a) shows direct form - I and Fig. 4.1 (b) shows direct form - II structure. Here
b 0 = 0.5 and b 1 = 0.5
z 1

x(n)

x(n)

0.5

y(n)

z 1

0.5
+

b0 = 0.5

b1 = 0.5

y(n)

(b)

(a)

Fig. 4.1 Direct form - I and II implementation of example 4.4.6 (i)

ii) 3y ( n) 2y ( n 1) +y ( n 2) = 4x ( n) 3x ( n 1) + 2x ( n 2)
\
\

3y (n) = 2y (n 1) y(n 2)+ 4x (n) 3x (n 1)+ 2x(n 2)


y (n) =

2
Here a 1 = ,
3

2
1
4
2
y (n 1) y (n 2)+ x (n) x (n 1)+ x (n 2)
3
3
3
3

1
4
a2 = , b 0 = ,
3
3

2
b 1 = 1, b 2 = .
3

Fig. 4.2 shows the direct form-I and direct form-II

(4 - 1)
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

4-2

4
b0 =
3

x(n)
z 1

b1 = 1

Digital Filter Structures

y(n)

z 1

1
a2 =
3

z 1

2
b2 =
3

z 1

2
a1 =
3

(a) Direct form - I

x(n)

4
b0 =
3

2
a1 =
3

1
a2 =
3

z 1

y(n)

b1 = 1
+

z 1

2
b2 =
3

(b) Direct form - II

Fig. 4.2 Direct form-I and II implementation of example 4.4.6 (ii)

iii) y (n ) 5y (n 1) = 7x (n )
\
y (n) = 5y (n 1)+7 x (n)
Fig. 4.3 shows direct form I and II.

x(n)

y(n)

x(n)

z 1

z 1
5

(b) Direct form - II

(a) Direct form - I

Fig. 4.3 Direct form - I and II realization of example 4.4.6 (iii)

Example 4.4.7
Solution :
i) y( n) = 0.5 y( n 1) 0.25 y( n 2) + x( n) + 0.4 x( n 1)
TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

y(n)

Digital Signal Processing Applications

4-3

x(n)

Digital Filter Structures

y(n)

z
0.4

0.5

Direct form - I
1

z
0.25

x(n)

y(n)

z
0.5

0.4

Direct form - II

z
0.25

Fig. 4.4 Direct form I and II realizations of 4.4.7 (i)

ii) y( n) = 0.1 y( n 1) + 0.2 y( n 2) +3x( n) +3.6 x( n 1) + 0.6 x( n 2)


3

x(n)
z

z
3.6

y(n)

0.1

Direct form - I

z
0.6

x(n)

0.2

+
z
+

y(n)

0.1

3.6

Direct form - II

0.6

0.2

Fig. 4.5 Direct form-I and II realizations of 4.4.7 (ii)

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

4-4

Digital Filter Structures

iii) y( n) = 0.1 y( n 1) + 0.72 y( n 2) + 0.7 x( n) - 0.252 x( n 2)


0.7

x(n)

y(n)

z
0.1

Direct form - I

z
0.252

x(n)

0.72

0.7

+
z

y(n)

0.1

Direct form - II
z

0.252

0.72

Fig. 4.6 Direct form-I and II realizations of 4.4.7 (iii)

Example 4.4.8
Solution :
Direct form - I and II
Here b 0 = 0.7 , b 1 = 0, b 2 =- 0.2 and a 1 = 0.1, a 2 = 0.72
0.7

x (n)

y (n)

z1

z1
+

0.1

z1

z1
0.2

0.72

Fig. 4.7 Direct form - I realization

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

Digital Signal Processing Applications

4-5

Digital Filter Structures

0.7

y (n)

z1
+

0.1

z1
0.72

0.2

Fig. 4.7 (a) Direct form - II realization

Cascade structure
Y(z) = - 0.1z-1Y(z)- 0 . 72 z-2 Y(z)+ 0 . 7X(z)- 0 . 2 z-2 X(z)
H(z)=

0.7(z 2 - 0.285)
0 . 7 - 0 . 2 z-2
Y(z)
0.7z 2 - 0.2
=
= 2
= 2
X(z)
1+ 0 . 1z-1 + 0. 72 z-2
z + 0.1 z+ 0.72
z + 0.1z+ 0.72
=

0.7(z- 0 . 53)(z+ 0 . 53)


z 2 + 0 . 1z+ 0.72

= 0.7

Here,

H 1 (z) = 0.7

0.7(1- 0 . 53z-1 )(1+ 0 . 53z-1 )

1+ 0.1z-1 + 0.72 z-2

1- 0 . 53z-1
1+ 0 . 1z-1 + 0 . 72 z-2
1- 0 . 53z-1

(1+ 0 . 53 z-1 ) = H 1 (z)H 2 (z)


and

1+ 0.1 z-1 + 0 . 72 z-2

1
H 2 (z) = 1 + 0.53 z

Fig. 4.8 shows the cascade realization of H 1 (z) and H 2 (z).


x (n)

0.7

z1
+

0.1

y (n)

z1
0.53

0.53

z1
0.72

Fig. 4.8 Cascade realization

qqq

TM

TECHNICAL PUBLICATIONS - An up thrust for knowledge

You might also like