www.EasyEngineering.

net

3
Z-Transforms
ww
w .Ea
3.1 INTRODUCTION
syE
A linear time-invariant discrete-time system is represented by difference equations. The

ngi
direct solution of higher order difference equations is quite tedious and time consuming. So
usually they are solved by indirect methods. The Z-transform plays the same role for

nee
discrete-time systems as that played by Laplace transform for continuous-time systems. The
Z-transform is the discrete-time counterpart of the Laplace transform. It is the Laplace
transform of the discretized version of the continuous-time signal x(t). To solve the

rin
difference equations which are in time domain, they are converted first into algebraic
equations in z-domain using Z-transform, the algebraic equations are manipulated in z-

g.n
domain and the result obtained is converted back into time domain using inverse Z-
transform. The Z-transform has the advantage that it is a simple and systematic method and

e
the complete solution can be obtained in one step and the initial conditions can be introduced
in the beginning of the process itself. The Z-transform plays an important role in the analysis
and representation of discrete-time Linear Shift Invariant (LSI) systems. It is the
generalization of the Discrete-Time Fourier Transform (DTFT). The Z-transform may be
one-sided (unilateral) or two-sided (bilateral). It is the one-sided or unilateral Z-transform
t
that is more useful, because we mostly deal with causal sequences. Further, it is eminently
suited for solving difference equations with initial conditions.
The bilateral or two-sided Z-transform of a discrete-time signal or a sequence x(n) is
defined as:
‡
X ( z) Ç
n ‡
x (n) z  n

where z is a complex variable.

179

www.EasyEngineering.net
 www.EasyEngineering.net

180 h Digital Signal Processing

The one-sided or unilateral Z-transform is defined as:

‡
X ( z) Ç
x (n) z  n
n 0

If x(n) = 0, for n < 0, the one-sided and two-sided Z-transforms are equivalent.
In the z-domain, the convolution of two time domain signals is equivalent to
multiplication of their corresponding Z-transforms. This property simplifies the analysis of
the response of an LTI system to various signals.

Region of convergence (ROC)

ww
For any given sequence, the Z-transform may or may not converge.
The set of values of z or equivalently the set of points in z-plane, for which X(z)
converges is called the region of convergence (ROC) of X(z). In general ROC can be

w
Rx  < z < R x  where R  can be as small as zero and Rx can be as large as infinity.

.Ea
x
If there is no value of z (i.e. no point in the z-plane) for which X(z) converges, then the
sequence x(n) is said to be having no Z-transform.

syE
3.1.1 Advantages of Z-transform

ngi
1. The Z-transform converts the difference equations of a discrete-time system into
linear algebraic equations so that the analysis becomes easy and simple.
2. Convolution in time domain is converted into multiplication in z-domain.

Transform (DTFT) does not exist.

nee
3. Z-transform exists for most of the signals for which Discrete-Time Fourier

4. Also since the Fourier transform is nothing but the Z-transform evaluated along the

rin
unit circle in the z-plane, the frequency response can be determined.

3.2 RELATION BETWEEN DISCRETE-TIME FOURIER TRANSFORM (DTFT)

AND Z-TRANSFORM g.n
The Discrete-Time Fourier Transform (DTFT) of a sequence x(n) is given by

X(ejw) or X (X )
‡
Ç
x (n) e  jX n
e t
n ‡

For the existence of DTFT, the above summation should converge, i.e. x(n) must be
absolutely summable. The Z-transform of the sequence x(n) is given by
‡
Z [ x (n)] X ( z)
n ‡
Ç
x (n) z  n

where z is a complex variable and is given by

z = re jw
where r is the radius of a circle.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 181

‡
\ X (z ) X (re jX )
n
LJ x(n) (re X )
j n

n
LJ[ x(n) r  ] e X
n j n

For the existence of Z-transform, the above summation should converge, i.e. x(n) r–n must be
absolutely summable, i.e.
‡

n ‡
Ç
| x (n) r  n |  ‡

ww
The above equation represents the Discrete-Time Fourier Transform of a signal x(n) r–n.
Hence, we can say that the Z-transform of x(n) is same as the Discrete-Time Fourier

w
Transform of x(n) r–n.

.Ea Ç
For the DTFT to exist, the discrete sequence x(n) must be absolutely summable, i.e.
‡
| x (n)|  ‡

syE n ‡

So for many sequences, the DTFT may not exist but the Z-transform may exist. When

ngi
r = 1, the DTFT is same as the Z-transform, i.e. the DTFT is nothing but the Z-transform
evaluated along the unit circle centred at the origin of the z-plane.

EXAMPLE 3.1
radius r.
nee
Prove that, for causal sequences, the ROC is the exterior of a circle of

rin
Solution: Causal sequences are the sequences defined for only positive integer values of n
and do not exist for negative times, i.e.

Consider a causal sequence,

x(n) = 0 for n < 0
g.n
x(n) = rn u(n)
From the definition of Z-transform of x(n), we have
‡
e t
X ( z)
n ‡
Ç
x ( n) z  n

n ‡
Ç
r n u ( n) z  n

‡
Ç
r n z n
n 0
‡
Ç (rz  )
n 0
1 n

www.EasyEngineering.net
 www.EasyEngineering.net

182 h Digital Signal Processing

The above summation converges for

| rz 1 |  1, i.e. for | z | ! r
Hence, for the causal sequences, the ROC is the exterior of a circle of radius r.

EXAMPLE 3.2 Prove that the sequences

(a) x ( n) n
a u( n) and (b) x ( n)  a n u( n  1)
have the same X(z) and differ only in ROC. Also plot their ROCs.
Solution:
(a) The given sequence an u(n) is a causal infinite duration sequence, i.e.

ww x ( n)
ÑÎa ,
Ï
n
n•0
because u( n)
ÎÑ 1 for n • 0
Ï
ÑÐ0 for n  0

w \
.Ea
Z [ x ( n)]
ÑÐ0, n0

Z[ a n u( n)]
n
‡
LJ a u(n) z 
n n

Çs Ç
‡ ‡
"
yE n 0
an z  n
n 0
[ az 1 ]n 1  az 1  ( az 1 ) 2  ( az 1 )3 

ngi
This is a geometric series of infinite length, and converges if |az–1 | < 1, i.e. if
|z | > |a |.

\ X ( z)
1  az
1
1 za nee
z
; ROC; | z | ! | a |

which implies that the ROC is exterior to the circle of radius a as shown in
Figure 3.1(a)
rin
a n u(n) Ž  1  1az
ZT
1 z  a ; ROC; | z | ! a
z
g.n
e t

Figure 3.1 (a) ROC of an u(n) (b) ROC of –an u(–n – 1).

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 183

(b) The given signal x(n) = –an u(–n – 1) is a non-causal infinite duration sequence,
i.e.
ÎÑ  a n , n 1 ÎÑ1 for n 1
x ( n) Ï because u(  n  1) Ï
ÑÐ 0, n•0 ÑÐ0 for n • 0
‡ 1
\ X (z ) Z[ a n u (  n  1)]
n
LJ a u(n  1) z  Ç ‡a z 
n n

n
n n

‡ ‡
Ç  a z n n
 Ç (a  z ) 1 n

ww n 1 n 1

This is a geometric series of infinite length and converges if |a–1z| < 1 or |z| < |a|.
Hence

w .Ea
Ë ‡
X ( z)  Ì Ç
Û
( a1 z )n  1Ü 1 
Ìn 0
Í
‡
( a1 z ) n
ÝÜ
Ç
n 0

a 1 z
syE1
1
1  a 1 z

1  a 1 z 1  az 1
1
za
z
; ROC; | z |  | a |

ngi
That is, the ROC is the interior of the circle of radius a as shown in Figure 3.1(b).
From this example, we can observe that the Z-transform of the sequences
anu(n) and –anu(–n – 1) are same, even though the sequences are different. Only

nee
ROC differentiates them. Therefore, to find the correct inverse Z-transform, it is
essential to know the ROC. The ROCs are shown in Figure 3.1[(a) and (b)].
In general, the ROC of a causal signal is | z | > a and the ROC of a non-causal
signal is | z | < a, where a is some constant.
rin
EXAMPLE 3.3 Find the Z-transform of the sequence
n
ÈQ
g.n
Also sketch the ROC and pole-zero location.
x ( n)
È 1Ø
É Ù
Ê 4Ú
cos É
Ê 3
Ø
n Ù u( n)
Ú
e t
ÈQ Ë e j (Q /3) n  e  j (Q /3) n Û
n n
È 1Ø Ø È 1Ø
Solution: Given x (n) É Ù cos É n Ù u( n) É Ù Ì Ü u( n)
Ê 4Ú Ê 3 Ú Ê 4Ú 2
ÍÌ ÜÝ

ËÈ 1 n n
1 j (Q /3) Ø È 1  j (Q /3) Ø Û
ÌÉ e Ù  É e Ù Ü u( n)
2 Ê4 Ú Ê4 Ú Ü
ÍÌ Ý
We have
‡
X (z )
n
Ç ‡ x(n) z  n

www.EasyEngineering.net
 www.EasyEngineering.net

184 h Digital Signal Processing

‡ 1 ËÈ 1
Ç
n n
j (Q /3) Ø È 1  j (Q /3) Ø Û n
ÌÉ e Ù É e Ù Ü u( n) z
n ‡ 2 ÍÌ Ê 4 Ú Ê4 Ú Ü
Ý
‡ 1 È1 ‡ 1 È1
Ç Ç
n n
e j (Q /3) Ø n 
 j (Q /3) Ø z  n
É Ù z É e Ù
n 0 2 Ê 4 Ú Ê
n 0 2 4
Ú

1 ‡ Ë 1 j (Q /3) 1 Û 1 ‡ Ë 1  j (Q /3) 1 Û
Ç Ç
n n
e z  e z
2 n 0 ÌÍ 4 Ü
Ý 2 n 0 ÌÍ 4 Ü
Ý

1È 1 Ø 1È 1 Ø

ww É
2 Ê 1  (1/4)e j (Q /3) 1
z Ú
Ù  É
2 Ê 1  (1/4) e  j (Q /3) 1
z Ú
Ù

w
The first series converges if

1 1  j (Q /3)
z e .Ea 1
1 1 j (Q /3)
4
z e

1
 1, i.e. | z | ! . So the ROC is z ! .
1
 1, i.e. | z | ! and the second series converges if
4

syE
4 4 4

\ 1È z Ø 1È z Ø
X (z ) 
2 ÉÊ z j ( Q /3) Ù 2 ÉÊ z Ù
e  j ( Q / 3) Ú

ngi
 (1/4) e Ú  (1/4)
z [ z  (1/4) cos (Q /3)] z [ z  (1/8)]
j ( Q /3)  j ( Q /3)
[z  (1/4) e ] [z  (1/4) e

nee
] [z

The poles are at z = (1/4) e j(p /3) and z = (1/4) e –j(p /3) and the zeros are at z = 0 and
 (1/8)  j ( 3 / 8)] [ z  (1/8)  j ( 3/8)]

z = (1/8). The pole-zero location and the ROC are sketched in Figure 3.2.

rin
g.n
e t
Figure 3.2 Pole-zero location and ROC for Example 3.3.

EXAMPLE 3.4 Find the Z-transform and ROC of

n 2n
È 5Ø È1Ø
x ( n) 2É Ù u (  n  1) 3É Ù u( n)
Ê 6Ú Ê2Ú

Sketch the ROC and pole-zero location.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 185

n 2n n n
È 5Ø È1Ø È 5Ø È1Ø
Solution: Given x (n) 2É Ù u (  n  1)  3 É Ù u( n) 2É Ù u(  n  1)  3 É Ù u( n)
Ê 6Ú Ê 2Ú Ê 6Ú Ê 4Ú
We have
‡ ‡
Ç ‡ Ç ‡
Ë È 5Øn n Û
x ( n) z  n z n
È 1Ø
X (z ) Ì2 É Ù u(  n  1)  3 É Ù u(n) Ü
n n ÌÍ Ê 6 Ú Ê 4Ú ÜÝ

‡ ‡ 1 ‡
Ç Ç ‡ 3 14 Ç ‡ 2 Ç3
n n n n
u(  n  1) z  n  u( n) z  n z n z n
È 5Ø È Ø È 5Ø È 1Ø
2É Ù É Ù É Ù  É Ù
n ‡
Ê 6Ú n
Ê Ú n
Ê 6Ú n 0
Ê 4Ú

wwÇ ‡Ë È 5 Ø 1 Û ‡ È 1 Øn
n

2 ÌÉ Ù z Ü 
n 1 ÌÍ
Ê 6 Ú ÜÝ n 0
3 É z 1 Ù
Ê4 Ú
Ç
w
The first series converges if | (5/6)–1z | < 1 or | z | < (5/6) and the second series converges if

.Ea
|(1/4) z–1| < 1 or |z| > (1/4).
So the region of convergence for X(z) is (1/4) < |z| < (5/6), i.e. it is a ring with
(1/4) < |z| < (5/6).

\
Î
Ñ
s
Ç yEÇ
2 Ï1 
‡ ËÈ 5 Ø 1 Û
nÞ
Ñ ‡
ß
È1 Ø
3 É z 1 Ù
n
X ( z)
ÑÐ
ngi
n 0
ÌÉ Ù z Ü
Ê Ú
ÍÌ 6 ÝÜ Ñ
à
n 0
Ê4 Ú

ÑÎ Ë
Ï2 Ì 1 
ÑÐ Í nee 1

Û
1 Ü
1  (5/6) z Ý
3
1

Þ
Ñ
1ß
1  (1/4) z Ñà
2
z
z  (5/6)
3
z
z  (1/4)

2z 2  ( z/2)  3z 2  (5z / 2) z ( z  2)

rin
[ z  (5/6)] [ z  (1/4)] [ z  (5/6)] [ z  (1/4)]
g.n
e
The ROC and the pole-zero plot are shown in Figure 3.3.

Figure 3.3 Pole-zero plot and ROC for Example 3.4.

www.EasyEngineering.net
 www.EasyEngineering.net

186 h Digital Signal Processing

EXAMPLE 3.5 Consider the sequence

ÑÎ a N  1, a  0
n
0 n
x ( n) Ï
ÑÐ0 otherwise
Find X(z).
‡ N 1 N 1
Solution: Z[ x (n)] X (z )
n
Ç ‡ x(n) z  Ç a z  Ç (az  )
n

n 0
n n

n 0
1 n

1  (az 1 ) N 1  ( a/z) N ( z N  a N )/z N z N  aN

1  az 1 1  ( a/z) ( z  a)/z z N 1 ( z  a)

ww
\ X ( z)
1
z N 1
zN  aN
za

w .Ea
3.3 Z-TRANSFORM AND ROC OF FINITE DURATION SEQUENCES

syE
Finite duration sequences are sequences having a finite number of samples. Finite duration
sequences may be right-sided sequences or left-sided sequences or two-sided sequences.

3.3.1 Right-sided Sequence

ngi
but finite. The Z-transform of such a sequence is X ( z )
‡
nÇ ee
A right-sided sequence is one for which x(n) = 0 for n < n0, where n0 is positive or negative

x ( n) z  n . The ROC of the above

n n0
series is the exterior of a cirle. If n0 ³ 0, the resulting sequence is a causal or a positive time
rin
sequence. For a causal or a positive finite time sequence, the ROC is entire z-plane except at
z = 0.
g.n
EXAMPLE 3.6 Find the ROC and Z-transform of the causal sequence
x (n) {1, 0,  2, 3, 5, 4}

e t
Solution: The given sequence values are:
x (0) 1, x (1) 0, x (2)  2, x (3) 3, x (4) 5 and x (5) 4.
We know that
‡
X (z )
n
Ç ‡ x(n) z  n

For the given sample values,

X ( z) x (0)  x (1) z 1  x (2) z 2  x (3) z 3  x (4) z 4  x (5) z 5
\ Z [ x ( n)] X ( z ) 1  2 z 2  3z 3  5z 4  4 z 5
The X(z) converges for all values of z except at z = 0.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 187

EXAMPLE 3.7 A finite sequence x(n) is defined as x(n) = {5, 3, –2, 0, 4, –3}. Find X(z)
and its ROC.
Solution: Given x(n) = {5, 3, –2, 0, 4, –3}
\ x ( n) 5E ( n)  3E (n  1)  2E (n  2)  4E ( n  4)  3E (n  5)
The given sequence is a right-sided sequence. So the ROC is entire z-plane except at z = 0.
Taking Z-transform on both sides of the above equation, we have
X (z ) 5  3z 1  2 z 2  4 z 4  3z 5
ROC: Entire z-plane except at z = 0.

ww
3.3.2 Left-sided Sequence

w
A left-sided sequence is one for which x(n) = 0 for n ³ n0 where n0 is positive or negative,

.Ea
but finite. The Z-transform of such a sequence is X ( z ) Ç
n0

n ‡
x (n) z  n . The ROC of the

syE
above series is the interior of a cirle. If n0 £ 0, the resulting sequence is anticausal sequence.
For an anticausal finite duration sequence, the ROC is entire z-plane except at z = ¥.

EXAMPLE 3.8
ngi
Find the Z-transform and ROC of the anticausal sequence.
x ( n) {4, 2, 3, 1,  2, 1}

Solution: The given sequence values are: nee 

We know that
x (5) 4, x (4) 2, x ( 3) 3, x ( 2)  1, x( 1)  2, x(0) 1
rin
X ( z)
‡
LJ x(n) z n
g.n
e
n
For the given sample values, X(z) is:

\
X (z ) x ( 5) z 5  x ( 4) z 4  x ( 3) z 3  x ( 2) z 2  x ( 1) z  x (0)
Z[ x (n)] X ( z) 4 z  2 z  3z  z  2 z  1
5 4 3 2
t
The X(z) converges for all values of z except at z = ¥.

3.3.3 Two-sided Sequence

A sequence that has finite duration on both the left and right sides is known as a two-sided
sequence. A two-sided sequence is one that extends from n = – ¥ to n = + ¥. In general, we
‡ 1 ‡
can write X ( z )
n ‡
Ç
x (n) z  n =
n ‡
Ç
x (n) z  n +
n 0
Ç
x (n) z  n . The first series converges for

www.EasyEngineering.net
 www.EasyEngineering.net

188 h Digital Signal Processing

z < R x and the second series converges for z > R x . So the ROC of such a sequence
Rx  < z < R x  is a ring. For a two-sided finite duration sequence, the ROC is entire
z-plane except at z = 0 and z = ¥.

EXAMPLE 3.9 Find the Z-transform and ROC of the sequence

x (n) {2, 1,  3, 0, 4, 3, 2, 1, 5}

Solution: The given sequence values are:
x ( 4) 2, x ( 3) 1, x ( 2)  3, x ( 1) 0, x (0) 4, x (1) 3, x (2) 2, x (3) 1, x (4) 5

ww
We know that
X (z )
‡
Ç ‡ x(n) z  n

w
For the given sample values,
X (z )
.Ea
n

x ( 4) z 4  x (3) z 3  x (2)z 2  x (1)z  x (0)  x (1)z 1  x (2) z 2  x (3)z 3  x (4)z 4

2 z 4  z 3  3z 2  4  3z 1  2 z 2  z 3  5z 4

syE
The ROC is entire z-plane except at z = 0 and z = ¥.

EXAMPLE 3.10
(a) u (n)  u ( n  4) ngi
Find the Z-transform of the following sequences:
(b) u (  n)  u (  n  3) (c) u (2  n)  u ( 2  n)
Solution:
(a) The given sequence is: nee
x ( n) u ( n)  u ( n  4)
rin
g.n
From Figure 3.4, we notice that the sequence values are:
x ( n) 1, for 0 n 3
0, otherwise
e t

Figure 3.4 Sequences (a) u(n), (b) u(n – 4) and (c) u(n) – u(n – 4).

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 189

We know that
‡
X (z )
n
LJ x(n) z  n

Substituting the sequence values, we get

X ( z) 1  z 1  z 2  z 3
The ROC is entire z-plane except at z = 0.
(b) The given sequence is:
x (n) u( n)  u(  n  3)

ww From Figure 3.5, we notice that the sequence values are:

x (n) 1, for  2 n 0

w .Ea
0, otherwise

syE
ngi
nee
rin
Figure 3.5
g.n
Sequences (a) u(–n), (b) u(–n – 3) and (c) u(–n) – u(–n – 3).

‡
We know that X ( z)

Substituting the sequence values, we get

n
LJ x(n) z  n

e t
X ( z) 1  z  z 2
The ROC is entire z-plane except at z = ¥.
(c) The given sequence is:
x ( n) u(2  n)  u( 2  n)
From Figure 3.6, we notice that the sequence values are:
x (n) 1, for 1 n 2
0, otherwise

www.EasyEngineering.net
 www.EasyEngineering.net

190 h Digital Signal Processing

Substituting the sequence values, we get

X (z ) z  1  z 1  z 2
The ROC is entire z-plane except at z = 0 and z = ¥.

ww
w .Ea
syE
Figure 3.6 Sequences (a) u(2 – n), (b) u(–2 – n) and (c) u(2 – n) – u(–2 – n).

3.4 PROPERTIES OF ROC

ngi
2. The ROC cannot contain any poles.
nee
1. The ROC is a ring or disk in the z-plane centred at the origin.

3. If x(n) is an infinite duration causal sequence, the ROC is | z | > a, i.e. it is the
exterior of a circle of radius a.
rin
If x(n) is a finite duration causal sequence (right-sided sequence), the ROC is entire
z-plane except at z = 0.

g.n
4. If x(n) is an infinite duration anticausal sequence, the ROC is | z | < b, i.e. it is the
interior of a circle of radius b.

entire z-plane except at z = ¥. e

If x(n) is a finite duration anticausal sequence (left-sided sequence), the ROC is

5. If x(n) is a finite duration two-sided sequence, the ROC is entire z-plane except at
z = 0 and z = ¥.
t
6. If x(n) is an infinite duration, two-sided sequence, the ROC consists of a ring in the
z-plane (ROC; a < | z | < b) bounded on the interior and exterior by a pole, not
containing any poles.
7. The ROC of an LTI stable system contains the unit circle.
8. The ROC must be a connected region. If X(z) is rational, then its ROC is bounded
by poles or extends up to infinity.
9. x(n) = d (n) is the only signal whose ROC is entire z-plane.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 191

3.5 PROPERTIES OF Z-TRANSFORM

The Z-transform has several properties that can be used in the study of discrete-time signals
and systems. They can be used to find the closed form expression for the Z-transform of a
given sequence. Many of the properties are analogous to those of the DTFT. They make the
Z-transform a powerful tool for the analysis and design of discrete-time LTI systems. In
general, both one-sided and two-sided Z-transforms have almost same properties.

3.5.1 Linearity Property

The linearity property of Z-transform states that, the Z-transform of a weighted sum of two

ww
signals is equal to the weighted sum of individual Z-transforms. That is, the linearity
property states that
Ž  X ( z), with ROC = R
w
ZT
If x1 (n) 1 1

x ( n ) Ž  X ( z ) , with ROC = R
and

Then .Ea 2
ZT
2

ax1 ( n)  bx2 ( n) Ž  aX1 ( z )  bX 2 ( z ), with ROC = R1 © R2

ZT
2

Proof: We know that syE Ç Z[ x (n)] X (z)

‡

n ‡
x(n) z  n

Ç n
‡
\

Ç
g
Z[ ax1 ( n)  bx 2 ( n)]

inÇ n ‡
‡
[ ax1 (n)  bx2 ( n)] z  n

n ‡
‡
ax1 (n) z  n 

Ç
n ‡
‡
e
bx 2 ( n) z  n

Ç eri
a
n ‡
x1 ( n) z  n  b
n ‡
x 2 ( n) z  n
ng.
ax1 (n)  bx 2 (n)
aX1 ( z )  bX 2 ( z); ROC; R1 © R2

Ž  aX (z)  bX (z)
ZT
1 2
net
The ROC for the Z-transform of a sum of sequences is equal to the intersection of the ROCs
of the individual transforms.

3.5.2 Time Shifting Property

The time shifting property of Z-transform states that

If x ( n) Ž  X (z), with zero initial conditions with ROC = R

ZT

x ( n  m) Ž  z  m X ( z )
ZT
Then
with ROC = R except for the possible addition or deletion of the origin or infinity.

www.EasyEngineering.net
 www.EasyEngineering.net

192 h Digital Signal Processing

Proof: We know that

‡
Z[ x ( n)] X ( z)
n
Ç ‡ x(n) z  n

‡
\ Z [ x (n  m)]
n
LJ x(n  m) z  n

Let n – m = p in the summation, then n = m + p.

‡
\ Z [ x ( n  m)] Ç
x ( p) z  ( m  p )

ww p ‡

z m
‡

p ‡
x ( p) z  p Ç
w .Ea Ž 
 m
z X (z )

x ( n  m)
ZT
z m X (z )

Similarly,
syE x ( n  m) Ž ZT  z m X ( z )

(a) Time delay ngÇi

If the initial conditions are not neglected, we have

Z [ x (n  m)] z  m X (z )  z  m
m
x ( k ) z k

i.e. Z [ x ( n  m)] nee k 1

z  m X ( z )  z ( m  1) x ( 1)  z ( m  2) x ( 2)  z ( m  3) x ( 3)  "

(b) Time advance Z [ x(n  m)] z m X (z )  z m Ç x(k) z
m 1
k
rin
Z[ x (n  m)]
k 0

z m X ( z )  z m x (0)  z m 1 x (1)  z m 2 x (2)

g.n
i.e.
This time shifting property is very useful in finding the output y(n) of a system
described in difference equation for an input x(n).
e t
Z [ x (n  1)] zX ( z)  z x (0)
Z [ x ( n  2)] z 2 X ( z)  z 2 x (0)  zx (1)
Z [ x ( n  3)] z 3 X ( z)  z 3 x (0)  z 2 x (1)  zx (2)
Z [ x (n  1)] z 1 X ( z)  x ( 1)
Z [ x ( n  2)] z 2 X ( z)  z 1 x ( 1)  x ( 2)
Z [ x ( n  3)] z 3 X ( z)  z 2 x ( 1)  z 1 x ( 2)  x ( 3)

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 193

3.5.3 Multiplication by an Exponential Sequence Property

The multiplication by an exponential sequence property of Z-transform states that

If x ( n) Ž  X (z )
ZT
with ROC = R

ZT È zØ
Then a n x ( n) Ž  XÉ Ù with ROC = | a | R
Ê aÚ

where a is a complex number.

Proof: We know that

ww Z[ x (n)] X ( z)
n
‡
LJ x(n) z  n

\ w .Ea Ç Z [ an x (n)]
n
‡
LJ a x(n) z 
‡
n n

n

syE n ‡
ÈzØ
x ( n) É Ù
Ê aÚ

ngi XÉ
È zØ
Ù
Ê aÚ

nee an x ( n) Ž
ZT
 XÉ
È zØ
Ù
Ê aÚ

If X(z) has a pole at z = z1, then X(z/a) will have a pole at z = az1. In general, all the pole-
zero locations are scaled by a factor a. rin
Note: e jX n x (n) Ž
ZT

È z Ø
X É jX Ù
Êe Ú
X (e  jX z) g.n
e  jX n x ( n) Ž
ZT
 XÉ
È z Ø
Ê e  jX Ú
Ù X (e jX z) e t
3.5.4 Time Reversal Property
The time reversal property of Z-transform states that
If x ( n) Ž  X (z), with ROC = R
ZT

ZT È 1Ø 1
Then x (  n) Ž  XÉ Ù, with ROC =
Ê zÚ R

www.EasyEngineering.net
 www.EasyEngineering.net

194 h Digital Signal Processing

Proof: We know that

‡
Z[ x (n)] X ( z)
n
LJ x(n) z  n

‡
\ Z[ x ( n)]
n
Ç ‡ x(n) z n

Let p = –n in the above summation, then

‡
Z [ x (  n)]
p
LJ x( p) z p

ww ‡
Ç ‡ x( p) (z  ) 1 p

w
p

X ( z 1 )
È 1Ø

.Ea Ž  XÉ Ù
Ê zÚ

syE x (  n) ZT
X ( z 1 )

ngi
3.5.5 Time Expansion Property
The time expansion property of Z-transform states that

If Ž  n eex ( n)
ZT
X ( z), with ROC = R

Then Ž  x k ( n)
ZT
X ( z k ), with ROC = R1/k
rin
where xk ( n) xÉ
È nØ
Ù,
Ê kÚ
if n is an integer multiple of k g.n
0, otherwise
xk(n) has k – 1 zeros inserted between successive values of the original signal.
e t
Proof: We know that
‡
Z [ x ( n)] X ( z)
n
Ç ‡ x(n) z  n

‡ ‡
\ Z [ x k ( n)]
n
Ç ‡ x k ( n) z  n
n
Ç ‡ x È nØ
É Ù
ÊkÚ
zn

n
Let p
k

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 195

‡
\ Z [ x k ( n)]
p
Ç ‡ x( p) z  pk

p
Ç ‡ x( p) (z ) k p

X (z k )

È nØ ZT
xÉ Ù Ž  X (z k )
Ê kÚ

ww
3.5.6 Multiplication by n or Differentiation in z-domain Property

If w
The multiplication by n or differentiation in z-domain property of Z-transform states that

.Ea Ž  X ( z), with ROC = R

x ( n)
ZT

nx ( n) Ž   z X ( z), with ROC = R

d

syE
ZT
Then
dz
Proof: We know that
‡
LJ x(n) z 
Z [ x ( n)] X ( z)
ngi
n
n

Differentiating both sides with respect to z, we get

d d Ë ‡
Ç ‡ x(n) z 
n
Ç ee
Û ‡ d n

rin
n
X ( z) Ì Ü x ( n) (z )
dz dz Ì
Ín ÜÝ n ‡ dz

n
‡
LJ x ( n) (  n) z  n  1
g.n
z

1
1
n
‡
Ç ‡[nx(n)] z  n
e t
 z Z[ nx ( n)]

d
\ Z [ nx ( n)] z X (z )
dz

nx (n) Ž   z dzd X (z)

ZT

d k X (z )
In the same way, Z [nk x (n)] (1)k z k
dz k

www.EasyEngineering.net
 www.EasyEngineering.net

196 h Digital Signal Processing

3.5.7 Convolution Property

The convolution property of Z-transform states that the Z-transform of the convolution of
two signals is equal to the multiplication of their Z-transforms, i.e.

If Ž  X (z), with ROC = R

x1 ( n)
ZT
1 1

x ( n) Ž  X ( z) with ROC = R
ZT
and 2 2 2

x (n) * x (n) Ž  X ( z) X ( z), with ROC = R

ZT
Then 1 2 1 2 1 Ç R2

ww
Proof: We know that

x1 (n) * x2 (n)
‡
Ç ‡ x (k) x (n  k)
w
1 2
k

Let
We have .Ea x ( n) x1 (n) * x2 (n)

syE Z [ x ( n)] X ( z)
n
‡
LJ x(n) z  n

\ Ç
Z [ x1 (n) x2 (n)]
‡

‡
ngi
[ x1 (n) * x2 ( n)] z  n

nee
n

‡ Ë ‡ Û
LJ ÌÌÍ Ç x1 (k ) x2 (n  k ) Ü z  n
n

‡
k

‡
‡

LJ Ç ‡ x (k) x (n  k) z 
ÜÝ

(n  k ) rin
z k
n k
1 2

g.n
Interchanging the order of summations,

X (z )
‡

k ‡
Ç
x1 ( k ) z  k
‡

n ‡
x2 ( n  k ) z ( n  k )Ç
e t
Replacing (n – k) by p in the second summation, we get
‡ ‡
X ( z)
k
LJ x (k) z  LJ x ( p) z 
1
k

p
2
p

X1 ( z ) X 2 ( z )

x1 ( n) * x2 ( n) Ž  X (z) X (z); ROC; R © R

ZT
1 2 1 2

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 197

3.5.8 The Multiplication Property or Complex Convolution Property

The multiplication property or complex convolution property of Z-transform states that

If x1 ( n) Ž  X ( z)
ZT
1 and x2 (n) Ž  X ( z)
ZT
2

ZT 1 È z Ø 1
2Q j v
Then x1 ( n) x2 (n) Ž  Ô X1 (v) X 2 ÉÊ ÙÚ v dv
c
v

Proof: From the definition of Z-transform, we have

‡
Ç
[ x1 (n) x2 (n)] z  n

ww
Z [ x1 (n) x2 (n)]
n ‡

But from the inverse Z-transform, we have

w .Ea
x ( n)
1
2Q j v
Ô X (z ) z

Changing the complex variable z to v and substituting back, we get

n 1
dz

syÇ E
Z [ x1 ( n) x2 ( n)]
‡

‡
Ë 1
Ì
Í 2Q j
vÔ X1 ( v) v n 1 dv Ü x2 ( n) z  n
Û

ngi
n

Ë ‡
1 1  n Û 1
v X
Ô 1 ( v ) Ì Ç x2 ( n) ( v z ) Ü v dv
2Q j

1
2Q j v
Ìn  ‡
Í

È z Ø 1 nee
Ô X1 (v) X 2 ÉÊ v ÙÚ v dv
Ü
Ý

rin
x1 ( n) x2 ( n) Ž
ZT

1
2Q j v
Ô X1 ( v)
c
ÈzØ
X2 É Ù
Ê vÚ
1
v dv
g.n
3.5.9 Correlation Property e t
The correlation property of Z-transform states that

If x1 (n) Ž  X ( z)
ZT
1 and x 2 ( n) Ž  X ( z)
ZT
2

Then R12 (n) x1 ( n) ¦ x 2 ( n) Ž ZT  X1 ( z) X 2 ( z 1 )

Proof: From the definition of the Z-transform, we have

‡
Z [ x1 (n) ¦ x2 (n)]
n ‡
Ç
[ x1 (n) ¦ x2 (n)] z  n

www.EasyEngineering.net
 www.EasyEngineering.net

198 h Digital Signal Processing

But from the correlation sum, we have

‡ ‡
x1 ( n) ¦ x 2 ( n)
l ‡
Ç
x1 (l ) x2 (l  n) or x1 ( n) ¦ x 2 ( n)
l
LJ x (l  n) x (l)
1 2

Substituting back, we get

‡ ‡
LJ Ç ‡ x (l) x (l
Ë Û
Z [ x1 ( n) ¦ x2 ( n)] Ì 1 2  n) Ü z  n
n ÌÍ l ÜÝ

Interchanging the order of summation, we have

ww Z [ x1 (n) ¦ x2 (n)]
l
‡
LJ x (l) LJ x (l
1
Ë
Ì
ÌÍ n
‡
2  n) z  n Ü
Û

Ü
Ý

w
Letting l – n = m in the second summation, we have n = l – m.

.Ea Ç ‡ ‡
LJ x (m) z 
Ë
\ (l  m) ÛÜ

syÇE
Z [ x1 (n) ¦ x2 (n)] x1 (l) Ì 2
l ‡ ÌÍ m ÜÝ

‡ ‡
Ç ‡ x (m) z
Ë ÛË Û
Ì
ÌÍ l ‡
ngi
x1 (l) z l Ü Ì
ÜÝ ÌÍ m
2
m
Ü
ÜÝ

Çn
‡ ‡
LJ x (l) z 
Ë ÛË Û
Ì
Ì
Íl

X1 ( z) X2 ( z 1 )
1

ee l
ÜÌ
ÜÝ ÌÍ m ‡
x2 ( m) ( z 1 )  m Ü

rin
ÜÝ

R12 ( n) x1 ( n) ¦ x2 ( n) ŽZT X1 ( z ) X 2 ( z 1 ) g.n

3.5.10 Parseval’s Theorem or Relation or Property e t
The Parseval’s relation or theorem or property of Z-transform states that
If x1 ( n) Ž  X ( z)
ZT
1 and x2 ( n) Ž  X (z )
ZT
2

‡
Then
n
LJ x (n) x
1
*
2 (n)
1
2Q j v
* * 1 1
Ô X1 ( z) X2 [( z ) ] z dz for complex x1(n) and x2(n)

Proof: Substituting the relation of the inverse Z-transform at LHS, we get

‡
LHS =
n
LJ Ë 1
Ì
Í 2Q jv
Ô X1 ( z) z
n 1
dz Ü x *2 ( n)
Ý
Û

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 199

Ë ‡
1
2Q j v
Ô X1 ( z ) Ì
Í n ‡
Ì
x *
2 (Çn) ( z 1 ) n ÛÜ z 1 dz
ÜÝ

ÎÑ ‡ Þ*
1
Ô X1 ( z ) Ï
2Q j v ÐÑn ‡
Ç 1  n Ñ 1
x2 (n) [(z*) ] ß z dz
àÑ

Ô X1 ( z ) ^ X2 [( z ) ]` z dz
1 * 1 1 *

2Q j v
1 * * 1 1
2Q j v
ww Ô X1 ( z ) X2 [( z ) ] z dz RHS

‡
LJ x (n) x (n) 1 * 1 1

w
* *
2Q j v
Ô X1 ( z ) X 2 [( z ) ] z dz
1 2

.Ea
n

3.5.11 Initial Value Theorem

syEŽ 
The initial value theorem of Z-transform states that, for a causal signal x(n)
If

Then ngi x ( n)

Lt x (n)
0
ZT

x (0)
X (z )

Lt X (z)
‡
Proof: nee "
We know that for a causal signal
n z

Ç
Z [ x ( n)] X ( z )
‡

rin
x ( n) z  n x (0)  x (1) z 1  x (2) z 2 

g.n
n 0

" x (0) 
x (1) x (2)
 2 

Lt X ( z ) Lt
Ë
e
Taking the limit z ® ¥ on both sides, we have

Ì x (0) 
z

x (1)
z


x (2)

x (3)
 "Û
x (0)  0  0  " x (0)
t
z ‡ z ‡ Í z z2 z3
Ü
Ý

\ Lt x (n) x (0) Lt X (z)

n 0 z ‡
x (0) Lt X ( z )
z ‡
This theorem helps us to find the initial value of x(n) from X(z) without taking its inverse
Z-transform.

www.EasyEngineering.net
 www.EasyEngineering.net

200 h Digital Signal Processing

3.5.12 Final Value Theorem

The final value theorem of Z-transform states that, for a causal signal

If x ( n) Ž  X (z )
ZT

and if X(z) has no poles outside the unit circle, and it has no double or higher order poles on
the unit circle centred at the origin of the z-plane, then

Lt x (n) x (‡) Lt ( z  1) X (z )
n ‡ z 1

ww
Proof: We know that for a causal signal
‡
Z [ x ( n)] X (z ) Ç x ( n) z  n

w .Ea Z [ x ( n  1)] zX ( z)  zx (0)

n 0
‡
Ç x(n  1) z 
n 0
n

\
syE Ç Ç
Z [ x (n  1)]  Z [ x ( n)] zX ( z )  zx (0)  X ( z)
n 0
‡
x ( n  1) z  n 
‡

n 0
x ( n) z  n

Çn
‡
[ x ( n  1)  x (n)] z  n
i.e.

gin
( z  1) X ( z)  z x (0)

"
n 0

i.e.
ee"r
( z  1) X ( z )  z x (0) [ x (1)  x (0)] z 0  [ x (2)  x (1)] z 1  [ x (3)  x (2)] z 2 

Taking limit z ® 1 on both sides, we have

z 1 ing
Lt ( z  1) X ( z )  x (0) [ x (1)  x (0)  x (2)  x (1)  x (3)  x(2)   x(‡)  x(‡  1)

\ .ne
x (‡)  x (0)
x(‡) Lt (z  1) X (z)

or x (‡ )
z 1
Lt (1  z 1 ) X ( z)
z 1
t
x (‡ ) Lt ( z  1) X (z)
z 1
This theorem enables us to find the steady-state value of x(n), i.e. x(¥) without taking the
inverse Z-transform of X(z).
Some common Z-transform pairs are given in Table 3.1. The properties of Z-transform
are given in Table 3.2.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 201

TABLE 3.1 Some Common Z-transform Pairs

Sequence x(n) Z-transform X(z) ROC

1. d (n) 1 All z
1
2. u(n) z /( z  1) 1/(1  z ) z !1
1
1 1 z
3. u(–n)   z 1
1 z z 1 1  z 1
4. u(–n – 1) z /( z  1) 1/(1  z 1 ) z 1
z 1
2
5. u(–n – 2) z /(z – 1)

z 1
k

ww
6. u(–n – k) z /(z – 1)
7. d (n – k) z –k
All z except at z = 0 (if k > 0)
All z except at z = ¥ (if k < 0)

w
8. a n u (n)
9. – a n u (–n)
10. – a n u (–n – 1)
.Ea
( z /( z  a)) 1/(1  az 1 )
a/(z – a)
z /( z  a) 1/(1  az 1 )
1 1 2
|z| > |a|
|z| < |a|
|z| < |a|

syE  z /( z  1)  [ z /(1  z ) ]
2
11. nu (n) |z| > 1
1 1 2
12. na n u (n) az /( z  a) 2
az /(1  az ) |z| > |a|
az /( z  a) 2 az 1/(1  az 1 ) 2

ngi
13. – na n u (–n – 1) |z| < |a|
–jw n  jX 1  jX
14. e u (n) z /( z  e ) 1/(1  z e ) |z| > |1|
z ( z  cos X ) cos X

nee
1
1 z
15. cos w n u(n) |z| > 1
z  2 z cos X  1
2
1  2z 1
cos X  z 2
z sin X z 1 sin X
16. sin w n u(n)
z  2z cos X  1
2

z( z  a cos X )
1  2 z 1 cos X  z 2
1  z 1a cos X rin |z| > 1

17. an cos w n u(n)

z  2az cos X  a
2

az sin X
2
1  2az 1
cos X  a z
az 1 sin X
2

g.n |z| > |a|

18. an sin w n u(n)

19. (n + 1) an u(n)
20. –nu(–n – 1)
z 2  2az cos X  a2
z 2 /( z  a)2
z /( z  1)
1  2az 1 cos X  a2 z 2
1/(1  az 1 )2
z /(1  z )
2 1 1 2
e|z| > |a|

|z| > |a|

|z| < |1|
t
21. nan u(n) z/(z – a)2 |z| > |a|
22. [ n ( n  1) a n  2 u( n)]/2! z/(z – a)3 |z| > |a|

23.
n ( n  1) ! [ n  (k  2)] a n  k 1
u( n) z/(z – a)k |z| > |a|
( k  1)!
24. 1/n, n > 0  ln (1  z 1 ) |z| > |1|
n
È dØ 1
25. n k a n , k  0  É z |z| < |a|
Ê dz ÙÚ 1  az 1
26. a| n | for all n (1  a2 )/[(1  az ) (1  az 1 )] | a | < | z | < | 1/a |

www.EasyEngineering.net
 www.EasyEngineering.net

202 h Digital Signal Processing

TABLE 3.2 Properties of Z-transforms

Property Response ROC

Ž  X (z)
x1 ( n)
ZT
1 R1

x (n) Ž  X ( z)
2
ZT
2 R2

Linearity ax1 ( n)  bx2 ( n) ŽZT aX1 ( z )  bX2 ( z ) R1 Ç R2

x ( n  m) Ž  z  m X ( z )
ZT
Time shifting Same as X(z) except z = 0

x (n  m ) Ž  z m X ( z )
ZT
Same as X(z) except z = ¥

ww
Multiplication by
exponential sequence
or scaling in z-domain a n u( n) Ž
ZT
 XÉ
È zØ
Ù a R1  z  a R2

w
Ê aÚ

Time reversal
.Ea x (  n) Ž  X ( z 1 )
ZT 1
R2
 z 
1
R1

syE
È nØ ZT
Time expansion xÉ Ù Ž  X (z k )
ÊkÚ

Differentiation in z-domain Ž   z dzd X (z)

nx (n)
ZT
R1  z  R2

Conjugation x* (n) Ž  X* ( z )*
ZT
ngi R1  z  R2

Accumulation Ç x(k ) Ž  1 1z X (z)

n

k ‡
ZT

nee 1

Convolution x1 ( n) * x2 (n) Ž  X (z) X (z)

ZT
1 2

rin
At least the intersection
of R1 and R2
Correlation Rx1 x2 ( n) x1( n) ¦ x 2 ( n) Ž  X1 ( z) X2 ( z )
ZT 1

g.n
At least the intersection of
the ROC of X1(z) and X2(z–1)

Multiplication x1 ( n) x2 (n) Ž
ZT

1
v
È Ø 1
Ô X1( v) X 2 ÉÊ ÙÚ v dv
2Q j c v
z
At least
e
R1l R2 l  z  R1u R2u t
‡
Parseval’s theorem
n
LJ x (n) x
1 2 * ( n)
1
2Q j vÔc
X1 ( v) X 2 * É Ù v 1 dv
È 1 Ø
Ê v *Ú

Initial value theorem x (0) Lt x (n) Lt X ( z )

n 0 z ‡
Final value theorem x (‡) Lt x (n) Lt ( z  1) X ( z ),
n ‡ z 1
–1
If (1 – z ) has no pole on
or outside the unit circle

Note: The initial value theorem and the final value theorem hold true only for causal signals.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 203

EXAMPLE 3.11 Using properties of Z-transform, find the Z-transform of the following signals:
(a) x (n) u( n) (b) x ( n) u(  n  1)
(c) x (n) u(  n  2) (d) x ( n) 2 n u( n  2)
Solution:
(a) Given x (n) u ( n)
z 1
We know that Z [u(n)] ; ROC; z ! 1
z 1 1  z 1
Using the time reversal property,

ww Z [u ( n)]
z
z 1 z (1 / z )
1/z
(1/z )  1
1
1 z

z1
1
; ROC; z  1

w(b) Given

\
.Ea Z [ x ( n)]
x ( n)

X ( z)
u(  n  1)

Z [u (  n  1)]
u [ (n  1)]

Z {u [ ( n  1)]}

syE
1 1
z 1 Z[u(  n)] z 1 
1 z z ( z  1)

u(  n  2) u [  (n  2)]
(c) Given

\ Z [ x (n)]
x (n)

X (z ) ngi
Z [u(  n  2)] Z {u [ (n  2)]}

z 2 Z [u(  n)] nee

1 z
z2

z2
z 1

(d) Given x ( n) 2n u (n  2)
rin
Z [ u(n  2)] z 2 Z [ u(n)] z 2
z
z 1
z
z 1
1
1
z ( z  1) g.n
Z[2 n u(n  2)] Z[u( n  2)] z ( z/2)
1
z ( z  1)
z ( z/2)
1
e
( z/2) [( z/2)  1]
4
z ( z  2) t
EXAMPLE 3.12 Using properties of Z-transform, find the Z-transform of the sequence
ÎÑ1, for 0 n N 1
(a) x ( n) B n 2 u (n  2) (b) x (n) Ï
ÑÐ0, elsewhere
Solution:
(a) The Z-transform of the sequence x(n) = a n u(n) is given by
z
X ( z) ; ROC; z ! B
z B

www.EasyEngineering.net
 www.EasyEngineering.net

204 h Digital Signal Processing

Using the time shifting property of Z-transform, we have

Z[ x ( n  m)] z  m X ( z)
In the same way,
z 1
Z [B n 2 u( n  2)] z 2 Z[B n u(n] z 2 ; ROC; z ! B
z B z (z  B )

ÎÑ1, for 0 n N  1
(b) Given x ( n) Ï
ÑÐ0, elsewhere
implies that x(n) = u(n) – u(n – N).

ww We know that Z [u( n)]

z
z 1

w Using the time shifting property, we have

.Ea Z [u(n  N )] z  N Z[u( n)] zN

z
z 1

syE
Using the linearity property, we have

Z [u(n)  u( n  N )] Z[u( n)]  Z[u( n  N )]

z
z 1
 zN
z 1
z z
z 1
[1  z  N ]

EXAMPLE 3.13 ngi

Using appropriate properties of Z-transform, find the Z-transform of the
signal.

x ( n) n2 n sin É nee ÈQ
Ê 2
Ø
nÙ u( n)
Ú

ÈQ Ø rin
Solution:

We know that
Given x ( n) n2 n sin É
Ê 2
nÙ u( n)
Ú
g.n
Ë
Z Ìsin É
Í
ÈQ
Ê 2
Ø
nÙ u( n) Ü
Ú
Û

Ý z 2
 2z
z sin (Q /2)
cos (Q /2)  1 z 2
z
1
e t
Using the multiplication by an exponential property, we have

Ë ÈQ Ø Û Ë ÈQ Ø Û
Z Ì 2n sin É n Ù u( n) Ü Z Ìsin É nÙ u( n) Ü
Í Ê2 Ú Ý Í Ê 2 Ú Ýz  ( z/2)
z z/2
 ( z/2)
2
z 1 z ( z/2)2  1

2z
2
z  4

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 205

Using differentiation in z-domain property, we have

Ë ÈQ Ø Û d Î Ë n ÈQ Ø ÛÞ
Z Ì n2n sin É nÙ u( n) Ü z Ï Z Ì 2 sin É nÙ u( n) Ü ß
Í Ê2 Ú Ý dz Ñ
Ð Í Ê2 Ú ÝÑà
d È 2z Ø Ë ( z 2  4)(2)  2 z(2 z) Û
z z Ì Ü
dz ÉÊ z 2  4 ÙÚ ÍÌ ( z 2  4)2 ÝÜ
Ë 2 z 2  8 Û 2 z( z 2  4)
z Ì 2 2Ü
ÌÍ ( z  4) ÜÝ ( z 2  4)2

ww
EXAMPLE 3.14
of Z-transforms.
Find the Z-transform of the following signal using convolution property

w .Ea x ( n)
È1Ø
É Ù
Ê 2Ú
n

n
È 1Ø
u( n) * É Ù
Ê 4Ú
n
u( n)

Solution: Let
syE x1 ( n)
È 1Ø
É Ù
Ê 2Ú
u( n)

ngi
z 1
\ X1 ( z ) ; ROC; | z | !
z  (1/2) 2

and x2 ( n)
È 1Ø
É Ù
Ê 4Ú
n
u( n)
nee
\ X2 (z )
z
z  (1/4)
; ROC; z !
1
4 rin
Now, x ( n) x1 ( n) * x2 (n)
g.n
\ Z [ x ( n)] X (z )

z
Z [ x1 (n) * x2 ( n)]

z
; ROC; z !
1
X1 ( z) X 2 ( z); ROC; z !
1
2 e t
z  (1/2) z  (1/4) 2

EXAMPLE 3.15 Find the Z-transform of the signal

ËÈ 1 Ø n È 1Ø
n Û
x (n) n ÌÉ Ù u(n) * É Ù u(n) Ü
Ì Ê 2Ú Ê 3Ú Ü
Í Ý
n
È1Ø
Solution: Let x1 (n) É Ù u( n)
Ê2Ú

www.EasyEngineering.net
 www.EasyEngineering.net

206 h Digital Signal Processing

1 z 1
\ X1 ( z ) 1
; ROC; z !
1  (1/2) z z  (1/2) 2
n
È 1Ø
and x 2 ( n) É Ù u( n)
Ê 3Ú

1 z 1
\ X2 (z ) 1
; ROC; z !
1  (1/3) z z  (1/3) 3

Using convolution in the time domain property, we have

ww ËÈ 1 Ø n
Z [ x1 ( n) x2 (n)]

È 1Ø
n Û
X1 ( z ) X 2 ( z )

z z
\

w .Ea
Z ÌÉ Ù
ÌÍ Ê 2 Ú
u( n) * É

Using differentiation in z-domain property, we have

Ù
Ê 3Ú
u( n) Ü
ÜÝ z  (1/2) z  (1/3)

ÎÑ ËÈ 1 Ø

Ð ÌÍ
Ñ Ê 2Ú
n

syE È 1Ø
Ê 3Ú
n Û ÞÑ
Z Ïn ÌÉ Ù u( n) * É Ù u( n) Ü ß
ÜÝ Ñà
z
d
dz
ÎÑ Ë È 1 Ø n
Ì
Ï É Ù
Z
ÑÐ ÌÍ Ê 2 Ú
u ( n )
È 1Ø
Ê 3Ú
n ÛÞÑ
* É Ù u( n) Ü ß
ÜÝÑà

z
d
ngi
ÎÑ
Ï
z2 ÞÑ
ÐÑ[ z  (1/2)] [ z  (1/3)] àÑ
ß

nee
dz

Ë [ z 2  (5/6) z  (1/6)] 2 z  z 2 [2 z  (5/6)] Û

z Ì Ü
ÌÍ [ z 2  (5/6) z  (1/6)]2
rin
Ë 2 z 3  (5/3) z 2  (1/3) z  2 z 3  (5/6) z 2 Û
ÜÝ

 zÌ
ÍÌ [ z 2  (5/6) z  (1/6)]2
g.n
Ü
ÝÜ
(5/6) z 2 [ z  (2/5)]
[ z  (1/2)]2 [ z  (1/3)]2 e t
EXAMPLE 3.16 Using Z-transform, find the convolution of the sequences
x1 ( n) {2, 1, 0, 1, 3}; x2 ( n) {1, 3, 2}

Solution: From the convolution property of Z-transforms, we have

Z{x1 ( n) * x2 ( n)} X1 ( z) X 2 ( z) which implies that

x1 ( n) * x2 ( n) Z 1 [ X1 ( z) X2 ( z)]

Given x1 ( n) {2, 1, 0, 1, 3}

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 207

\ X1 ( z ) 2  z 1  z 3  3z 4

and x2 ( n) {1, 3, 2}

\ X2 ( z ) 1  3z 1  2z 2

\ X1 ( z ) X 2 ( z ) (2  z 1  z 3  3z 4 ) (1  3z 1  2 z 2 )
2  5z 1  z 2  z 3  6 z 4  11z 5  6 z 6
Taking inverse Z-transform on both sides,

ww
EXAMPLE 3.17
x (n) {2,  5, 1, 1, 6, 11, 6}

Find the convolution of the sequences

w .Ea x1 ( n)
È 1Ø
É Ù
Ê 2Ú
n
u( n) and x2 ( n)

using (a) Convolution property of Z-transforms and (b) Time domain method.
È 1Ø
É Ù
Ê 3Ú
n 2
u ( n  2)

Solution:
syE
ngi
n n 2
È 1Ø È 1Ø
(a) Given x1 ( n) É Ù u( n) and x2 ( n) É Ù u ( n  2)
Ê 2Ú Ê 3Ú

\ X1 (z )
ËÈ 1 Ø n
Z ÌÉ Ù
ÌÍ Ê 2 Ú
u(n)Ü
Û

ÜÝ
1
1  (1/2) z nee
1
z
z  (1/2)
; ROC; z !
1
2

ËÈ 1 Ø n  2 Û ËÈ 1 Ø n
z 2 Z Ì É
Û
rin
and X2 ( z) Z ÌÉ
Ì
Í
Ù
Ê 3Ú
u( n  2) Ü
ÜÝ

z 1
Ù
ÌÍ Ê 3 Ú
u( n) Ü
ÜÝ
g.n
We know that
z 2
1  (1/3) z
1

x ( n)
1 z  (1/3)

x1 (n) * x2 (n)
; ROC; z !
1
3
e t
\ Z[ x (n)] X (z ) Z [ x1 (n) * x2 (n)] X1 ( z ) X 2 ( z )

z z 1 1
\ Z [ x1 ( n) * x2 ( n)]
z  (1/2) z  (1/3) [ z  (1/2)][ z  (1/3)]
Î 1 Þ Ë 1 1 Û
\ x ( n) Z 1 Ï ß Z 1 Ì  Ü6
Ð[ z  (1/2)][ z  (1/3)] à Í z  (1/2) z  (1/3) Ý
ËÈ 1 Ø n 1 È 1Ø
n 1 Û
6 ÌÉ Ù u(n  1)  É Ù u( n  1) Ü
ÌÍÊ 2 Ú Ê 3Ú ÜÝ

www.EasyEngineering.net
 www.EasyEngineering.net

208 h Digital Signal Processing

Ç x (k ) x (n  k )
n
(b) x1 (n) * x2 (n) 1 2
k 0

n2 n2k

Ç
k
È 1Ø È 1Ø
É Ù u (k ) É Ù u( n  2  k )
Ê Ú
k 0 2
Ê 3Ú

n2 n2k n2 2 k

Ç Ç
k k n
È 1Ø È 1Ø È 1Ø È 1Ø È 1Ø È 1Ø
É Ù É Ù É Ù É Ù É Ù É Ù
Ê Ú Ê 3Ú
k 0 2
Ê Ú
k 0 2
Ê 3Ú Ê 3Ú Ê 3Ú

n n2 n 1 Û
Ë1  (3/2) n 1 Û
Ç
k n n
È 1Ø È 3Ø È 1Ø È 1Ø Ë È 3Ø

ww 9É Ù
Ê 3Ú
k 0

Ë È 1 Ø n 1
É Ù
Ê 2Ú
9É

n 1
Ù
Ê 3Ú

n 1 Û
Ì
Ì 1  (3/2) ÝÜ
Í
Ü  18 É Ù Ì1  É Ù
Ê 3Ú Ì
Í
Ê 2Ú
Ü
ÜÝ

w
È 1Ø È 3Ø
 6 ÌÉ Ù u(n  1)  É Ù É Ù u(n  1) Ü
Ì Ê 3Ú Ê 3Ú Ê 2Ú ÜÝ

.Ea Í

ËÈ 1 Ø n 1
 6 ÌÉ Ù
È 1Ø
u(n  1)  É Ù
n 1 Û
u( n  1) Ü

syE ÌÍ Ê 3 Ú Ê 2Ú ÜÝ

ngi
EXAMPLE 3.18 Determine the cross correlation sequence rx x (l) of the sequences
1 2

x1 ( n) = (1,2,3, 4)

nee
x2 ( n) = (4,3, 2,1)

Solution: The cross correlation sequence can be obtained using the correlation property of
Z-transforms.
Therefore, for the given x1(n) and x2(n), we have
rin
and
X1 ( z) = 1 + 2z 1 + 3z 2 + 4z 3

X2 ( z) = 4 + 3z 1 + 2z 2 + z 3 g.n
Thus, we have

Now
X2 (z 1 ) = 4 + 3z + 2z 2 + z 3

Rx1 x2 (z ) = X1 ( z) X2 ( z 1 ) = (1 + 2z 1 + 3z 2 + 4z 3 )(4 + 3z + 2z 2 + z 3 )
e t
or Rx1 x2 (z ) = z 3 + 4z 2 + 10z + 20 + 25z 1 + 24 z 2 + 16z 3

Therefore, rx1 x2 (l) = Z 1[ Rx1 x2 (z )] = {1, 4,10, 20, 25, 24,16}

Another method
The cross correlation can be written as

rx1 x2 (l) =  x1 (n) x2 (n  l)
n 

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 209

or equivalently, we have

rx1 x2 (l) =  x1 (n + l) x 2 (n)
n

where l is the time shift index.


For l = 0, we have rx1 x2 (0) =  x1 (n) x2 (n)
n

The product sequence P0(n) = x1(n) x2(n) is

ww
and therefore, the sum of all values of P0(n) is
P0 ( n) = {4, 6, 6, 4}

w .Ea
rx1 x2 (0) = 20


For l = 1, we have rx1 x2 (1) =  x1 (n) x2 ( n  1)

syE
The product sequence P1(n) = x1(n) x2(n – 1) is
n

and therefore, the sum of all values of P1(n) is ngi

P1 (n) = {8,9,8}

nee
rx1 x2 (1) = 25

Similarly, rx1 x2 (2) = 24, rx1 x2 (3) = 16

rin
For l < 0, rx 1 x2 (1) = 10, rx 1 x2 (2) = 4, rx 1 x2 (3) = 1
g.n
Therefore, the cross correlation of sequences x1(n) and x2(n) is

rx1 x2 (l) = {1, 4, 10, 20, 25, 24, 16}


e t
EXAMPLE 3.19 Using final value theorem, find x(¥), if X(z) is given by
z 1 z2 2z  3
(a) (b) (c)
( z  0.6) 2 4( z  1) ( z  0.7) ( z  1)( z  3)( z  1)

Solution:
z 1
(a) Given X (z )
( z  0.6)2

www.EasyEngineering.net
 www.EasyEngineering.net

210 h Digital Signal Processing

Looking at X(z), we notice that the ROC of X(z) is | z | > 0.6 and (z – 1) X(z) has no
poles on or outside the unit circle. Therefore,
z 1
x (‡) Lt ( z  1) X ( z) Lt ( z  1) 0
z 1 z 1 ( z  0.6)2
z2
(b) Given X ( z)
4( z  1) ( z  0.7)
z 2
( z  1) X ( z)
4 ( z  0.7)

ww (z – 1) X(z) has no poles on or outside the unit circle.

\ x (‡) Lt ( z  1) X ( z )
1
Lt

Ë z 2 Û
Ì Ü
3
0.44

w(c) Given
.Ea
z

X ( z)
z

2z  3
1 Í 4( z  0.7) Ý

( z  1) ( z  3)( z  1)
6.8

syE
( z  1) X ( z)
( z  1) (2 z  3)
( z  1)( z  3)( z  1)
2z  3
( z  1) ( z  3)

ngi
(z – 1) X(z) has one pole on the unit circle and one pole outside the unit circle. So
x(¥) tends to infinity as n ® ¥.

EXAMPLE 3.20 Find x(0) if X(z) is given by

z  2z  2
nee z 3

rin
2
(a) (b)
( z  1) ( z  0.5) ( z  1)( z  2)
Solution:
z 2  2z  2 1  (2/z)  (2/z 2 ) g.n
(a) Given

x (0)
X ( z)
( z  1)( z  0.5)

Lt X ( z )
z ‡
Lt
[1  (1/z)] [1  (0.5/z)]

[1  (2/z)  (2/z 2 )]
z ‡ [1  (1/z )][1  (0.5/z )]
1
e t
z3 z [1  (3/z)] 1 1  (3/z)
(b) Given X ( z)
( z  1) ( z  2) z [1  (1/z )] [1  (2/z )]
2 z [1  (1/z )] [1  (2/z )]

1 1  (3/z )
x (0) Lt X ( z) Lt 0
z‡ z ‡ z [1  (1/z )] [1  (2/z )]

EXAMPLE 3.21 Prove that the final value of x(n) for X(z) = z2/(z – 1)(z – 0.2) is 1.25 and
its initial value is unity.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 211

z2
Solution: Given X ( z)
( z  1) ( z  0.2)
The final value theorem states that
Lt x (n) x (‡) Lt ( z  1) X (z )
n ‡ z 1
z2 z2
x (‡) Lt ( z  1)
1
\ Lt 1.25
z 1 ( z  1) ( z  0.2) z  1 z  0.2 1  0.2
The initial value theorem states that

ww n
Lt x (n)
0
x (0)
z
Lt X (z)
‡

w
1
\ x (0) Lt 1
z  ‡ [1  (1/z )] [1  (0.2/z )]

.Ea
3.6 INVERSE Z-TRANSFORM

syE
The process of finding the time domain signal x(n) from its Z-transform X(z) is called the
inverse Z-transform which is denoted as:
x ( n)
ngi
Z 1 [ X ( z )]
We have

X ( z) X (re jX )
‡
Ç nee [ x (n) r  n ] e  jX n
n ‡
This is the DTFT of the signal x(n) r–n. Hence the Inverse Discrete-Time Fourier Transform rin
(IDTFT) of X(re jw) must be x(n) r–n. Therefore, we can write
Q g.n
x ( n) r  n
1
2Q ÔQ

1
X (re jX ) e jX n dX

Q
e t
i.e. x ( n) X (re jX ) (re jX ) n dX
2Q ÔQ
We have
z = re jw

jre jX , i.e. dX
dz dz
\
dX jre jX
1
\ X ( z ) z n 1 dz
2Q j vÔc
x ( n)

www.EasyEngineering.net
 www.EasyEngineering.net

212 h Digital Signal Processing

where the symbol v

Ô denotes integration around the circle of radius | z | = r in a counter
c
clockwise direction.
This is the direct method of finding the inverse Z-transform of X(z). It is quite tedious.
So inverse Z-transform is normally found using indirect methods. The Z-transform X(z) is a
ratio of two polynomials in z given by

b0 z M  b1 z M  1  b2 z M  2  "b
"  a
M
X ( z) N 1 N 2
z  a1 z
N
 a2 z N

The roots of the numerator polynomial are those values of z for which X(z) = 0 and are

ww
referred to as the zeros of X(z). The roots of the denominator polynomial are those values of
z for which X(z) = ¥ and are referred to as poles of X(z). In z-plane, zero locations are
denoted by • (a small circle) symbol and the pole locations with × (cross) symbol.

w Basically, there are four methods that are often used to find the inverse Z-transform.
They are:
(a)
.Ea
Power series method or long division method

syE
(b) Partial fraction expansion method
(c) Complex inversion integral method (also known as the residue method)
(d) Convolution integral method

ngi
The long division method is simple, but does not give a closed form expression for the
time signal. Further, it can be used only if the ROC of the given X(z) is either of the form
| z | > a or of the form | z | < a, i.e. it is useful only if the sequence x(n) is either purely

nee
right-sided or purely left-sided. The partial fraction expansion method enables us to
determine the time signal x(n) making use of our knowledge of some basic Z-transform pairs
and Z-transform theorems. The inversion integral method requires a knowledge of the theory
rin
of complex variables, but is quite powerful and useful. The convolution integral method uses
convolution property of Z-transforms and can be used when given X(z) can be written as the
product of two functions.
g.n
3.6.1 Long Division Method
The Z-transform of a two-sided sequence x(n) is given by
e t
‡
X (z )
n
Ç ‡ x(n) z  n

The X(z) has both positive powers of z as well as negative powers of z. We cannot obtain a
two-sided sequence by long division. If the sequence x(n) is causal, then
‡
X ( z) Ç x(n) z  n
x (0) z 0  x (1) z 1  x (2) z 2  "
n 0

has only negative powers of z, with ROC; | z | > a.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 213

If the sequence x(n) is anticausal, then

Ç x(n) z "  x(3) z  x(2) z

0
n
X (z ) 3 2
 x( 1) z1  x(0) z 0
n ‡

has only positive powers of z, with ROC; | z | < a.

Since determination of the inverse Z-transform of X(z) is only the determination of
x(n), i.e. x(0), x(1), x(2), ¼ if it is a causal signal or x(0), x(–1), x(–2), ¼ if it is an
anticausal signal, to determine the inverse Z-transform, if X(z) is a ratio of the polynomials

N (z ) b0  b1 z 1  b2 z 2  " b z M

" a z
M
X (z )
1  a1 z 1  a2 z 2 N

ww D( z )

We can generate a series in z by dividing the numerator of X(z) by its denominator.

If X(z) converges for | z | > a, we obtain the series
N

w .Ea X (z ) x (0)  x (1) z 1  x (2) z 2 

–n
We can identify the coefficients of z as x(n) of a causal sequence.
"
If X(z) converges for | z | < a, we obtain the series

syE X ( z) x (0)  x ( 1) z1  x ( 2) z 2  "

–n

ngi
We can identify the coefficients of z as x(n) of a non-causal sequence.
For getting a causal sequence, first put N(z) and D(z) either in descending powers of z
or in ascending powers of z–1 before long division.

nee
For getting a non-causal sequence, first put N(z) and D(z) either in ascending powers of
z or in descending powers of z–1 before long division. This method is best illustrated by the
following examples.
rin
EXAMPLE 3.22 Find the inverse Z-transform of

X (z ) z 3  2 z 2  z  1  2 z 1  3z 2  4 z 3 g.n
Solution: We know that

X (z )
‡
Ç ‡ x(n) z  ! x(3) z  x(2) z
n 3 2
e
 x ( 1) z1  x (0)  x (1) z 1
t
n

 x (2) z 2  x (3) z 3  "

Comparing this X(z) with the given X(z), we have
x ( n) {1, 2, 1, 1,  2,  3, 4}

Alternatively, taking inverse Z-transform of X(z), we have
x (n) E (n  3)  2E (n  2)  E (n  1)  E (n)  2E (n  1)  3E (n  2)  4E (n  3)

www.EasyEngineering.net
 www.EasyEngineering.net

214 h Digital Signal Processing

EXAMPLE 3.23 Determine the inverse Z-transform of

1 1
(a) X ( z) ; ROC; z ! a (b) X ( z ) ; ROC; z ! a
za 1  az 1
1
(c) X ( z) ; ROC; z ! 1
1  z 4
Solution:
1
(a) Given X (z ) ; ROC; z ! a
za

")
ww 1
z (1  az ) 1
z 1 (1  az 1 ) 1

‡
z 1 (1  az 1  a2 z 2  a3 z 3 

" Ç a  z  Ç a  u(n  1) z 
‡

w \ .Ea
z 1  az 2  a2 z 3 

x ( n)
n 1

a n 1 u(n  1)
n 1 n

n 0
n 1 n

(b) Given
syE X ( z)
1
1  az 1
; ROC; z ! a

By Taylor’s series expansion, we have

1
1  az 1  a2 z 2  a3 z 3 
ng"i Ç Ç ‡
an z  n
‡
a n u( n) z  n
X (z )

Therefore,
1  az 1
x(n) = a u(n) nee
n
n 0 n 0

(c) From infinite sum formula, we have

‡ rin
k 0
Bk
1
1B
; Ç g.n
B 1

Given X ( z)
1
1  z 4
Ç
‡

k 0
(z 4 )k Ç
k 0
‡
z 4 k e Ë| z 4 |  1,
Í
i.e. | z | ! 1ÛÝ
t
Taking inverse Z-transform on both sides, we get
‡
x ( n) E (n  4 k ) Ç
k 0

\ x (n) 1, when n = 4k , i.e. when n is an integer multiple of 4

0, otherwise

EXAMPLE 3.24 Determine the inverse Z-transform of the sequences:

(a) X (z ) cos (3z ); ROC; z  ‡ (b) X (z ) sin (z ); ROC; z  ‡

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 215

Solution:
(a) Given X(z) = cos (3z)
The corresponding x(n) must be a left-sided sequence because ROC is | z | < ¥.
From the trigonometric series or Taylor series, we have
‡
cos R 1 
R2 R4 R6

2! 4! 6!
  ( 1)k
R 2k
(2k )!
" Ç
k 0

‡
\ X (z ) cos (3z) Ç ( 1) k
(3z )2 k
(2 k )!
1
(3z) 2 (3z ) 4 (3z) 6
2!

4!

6!
 "
k 0

ww "  8081 z 6

27 4 9 2
8
z  z 1
2

w Therefore, the inverse Z-transform is:

.Ea x ( n)
Ë
Ì
Ì
", 
81
80
27 9 Û
, 0, , 0,  , 0, 1Ü
8 2
Ü

syE Ì
Í

(b) Given X(z) = sin (2z). The corresponding x(n) must be a left-sided sequence
 ÜÝ

sin R R
R3 R5 R7
  
n"gÇi
because ROC is | z | < ¥. From the trigonometric series or Taylor series, we have
‡
( 1)k
R 2k 1
3! 5! 7!
nee k 0 (2 k  1)!

\ X (z ) sin (2z ) Ç
k 0
‡
( 1)k
(2 z )2 k  1
(2k  1)!
(2z ) 
3! rin "
(2z )3 (2 z)5 (2 z ) 7

5!

7!


"  3158 z 7

4 5 4 3
15
z  z  2z
3 g.n
Therefore, the inverse Z-transform is:

Ë
", 
8 4 4 Û
, 0, , 0,  , 0, 2, 0 Ü
e t
Ì
x ( n) 315 15 3
Ì Ü
ÌÍ  ÜÝ

EXAMPLE 3.25 Determine the inverse Z-transform of the following transformed signals:

X ( z) = log10 (1 + az 1 ); ROC; z > a

Solution: Given
X ( z ) = log10 (1 + az 1 )

www.EasyEngineering.net
 www.EasyEngineering.net

216 h Digital Signal Processing

Since the given signal is of base 10, manipulating, we get

loge (1 + az 1 )
X (z ) = log10 (1 + az 1 ) =
loge (10)

=
1
loge (10)
 1 (az 1 )2 ( az 1 )3 (az 1 ) 4 ( az 1 ) 5
az 
2
+
3

4
+
5
 "
 
 
( az 1 ) n (  a) n
=  =  z n
n 1
n loge (10) n 1
n loge (10)

ww
Therefore, the inverse Z-transform is:

Ë a a2 a3 a4
"
Û

w x ( n) =

.Ea
Ì 0,
Ì
Ì
͏
loge (10)
, ,
2 loge (10) 3 loge (10)
,
4 loge (10)
, Ü
Ü
Ü
Ý

syE
=
(  a) n
n loge (10)
u( n  1)

EXAMPLE 3.26
of ngi
Using power series expansion method, determine the inverse Z-transform

X ( z)
nee
ln (1  z 1 ); ROC; z ! 0

ln (1  z 1 )

rin
Solution: Given X (z )
The corresponding x(n) must be right-sided sequence because ROC is | z | > 0. We know that

ln (1  R ) R 
R
2
2
R
 
3
R
4

3 4 ‡
" Ç (1)  k 1 R
k
k
; if R  1 g.n
\ X (z ) ln (1  z 1 ) Ç
k 1
‡
(1)k 1
k 1

(z 1 )k
k
Ç
k 1
‡
(1)k 1
z k
k
e t
Taking inverse Z-transform on both sides, we get
‡
x (n)
k 1
( 1)k 1
k
Ç
E (n  k )

EXAMPLE 3.27 Determine the inverse Z-transform of

È 1 Ø È 1 Ø
(a) X ( z ) loge É ; ROC; z  a (b) X ( z ) loge ; ROC; z ! a
1 Ù É 1 Ù
Ê1  a zÚ Ê 1  az Ú

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 217

Solution:
È 1 Ø
(a) Given X (z ) loge É ; ROC; z  a
1 Ù
Ê1  a zÚ

È Ø
1
X ( z) loge É 1 Ù
 loge (1  a 1 z )
Ê1  a zÚ

Ì
1
 Ìa z 
( a 1 z)2
2

( a 1 z)3
3

(a 1 z) 4
4
 " Û
Ü
Ü
Í Ý

" Ç
1 È n Ø

ww \ X ( z) 
4 a4
1
z4 
3a3
1
z3 
1
2a2
z2 
1
a
z
n ‡
É
Ê
a
 n Ù z
Ú
n

w Hence

that is, .Ea

x ( n)

x ( n)


È an Ø
É Ù
an
n
for n  0

u(  n  1)

syE È 1 Ø
Ê n Ú

ngi
(b) Given X ( z) loge É ; ROC; z ! a
1 Ù
Ê 1  az Ú

È 1 Ø
X (z ) loge É 1 Ù
Ê 1  az Ú

( az 1 )2
nee"
 log e (1  az 1 )

( az 1 )3 ( az 1 ) 4
Ë

Ì
Í
1
 Ì  az 

1 2
2


1 3
3


1 4
4
ri
" Ç ng

Û
Ü
Ü
Ý
‡ (az 1 ) k ‡ ak
Ç
\ X (z ) az 1 
( az )
2

(az )
3

(az )
4

.ne
k 1 k k 1 k
zk

Taking inverse Z-transform, we have

x ( n)
‡ ak

k 1 k
Ç
E (n  k )
‡ an

n 1 n
Ç an
n
u( n  1)
t
EXAMPLE 3.28 Determine the inverse Z-transform of
X ( z ) = log e (1 + az 1 ); ROC; z ! a

Solution: Given X ( z ) = loge (1 + az 1 )

X ( z) = loge (1 + az 1 ) = az 1 
(az 1 )2 ( az 1 )3 (az 1 ) 4
2
+
3

4
+ "
 
( az 1 ) n (  a) n  n ( a) n
=  =  z = u( n  1)
n 1
n n 1
n n

www.EasyEngineering.net
 www.EasyEngineering.net

218 h Digital Signal Processing

(  a) n
\ x ( n) =  u (n  1)
n
EXAMPLE 3.29 Using long division, determine the inverse Z-transform of

z 2  2z
X ( z) ; ROC; z ! 1
z 3  3z 2  4 z  1

Solution: Since ROC is | z | > 1, x(n) must be a causal sequence. For getting a causal
sequence, the N(z) and D(z) of X(z) must be put either in descending powers of z or in
ascending powers of z –1 before performing long division.

ww In the given X(z) both N(z) and D(z) are already in descending powers of z.

z–1 + 5z–2 + 11z–3 + 12z–4 – 13z–5

w z3 – 3z2 + 4z + 1

.Ea
z2 + 2z
z2 – 3z + 4 + z–1
5z – 4 – z–1

syE 5z – 15 + 20z–1 + 5z–2

11 – 21z–1 – 5z–2

ngi
11 – 33z–1 + 44z–2 + 11z–3
12z–1 – 49z–2 – 11z–3

ne"e
12z–1 – 36z–2 + 48z–3 + 12z–4
–13z–2 – 59z–3 – 12z–4

\
\
X ( z) = z 1  5z 2  11z 3  12 z 4  13z 5
x (n) = {0, 1, 5, 11, 12, 13, …} rin
Writing N(z) and D(z) of X(z) in ascending powers of z–1, we have
g.n
X (z )
N ( z)
D( z )
z2  2z
z 3  3z 2  4 z  1
z 1  2 z 2
1  3z 1  4z 2  z 3

z–1 + 5z–2 + 11z–3 + 12z–4 – 13z–5

e t
1 – 3z–1 + 4z–2 + z–3 z–1 + 2z–2
z–1 – 3z–2 + 4z–3 + z–4
5z–2 – 4z–3 – z–4
5z–2 – 15z–3 + 20z–4 + 5z–5
11z–3 – 21z–4 – 5z–5
11z–3 – 33z–4 + 44z–5 + 11z–6
12z–4 – 49z–5 – 11z–6
12z–4 – 36z–5 + 48z–6 + 12z–7
–13z–5 – 59z–6 – 12z–7

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 219

\ X ( z ) = z 1 + 5z 2 + 11z 3 + 12 z 4  13z 5 "

\ x (n) = {0, 1, 5, 11, 12, 13, …}
Observe that both the methods give the same sequence x(n).

EXAMPLE 3.30 Using long division, determine the inverse Z-transform of

z2  z  2
X (z ) ; ROC; z  1
z 3  2 z 2  3z  4

ww
Solution: Since ROC is |z| < 1, x(n) must be a non-causal sequence. For getting a non-
causal sequence, the N(z) and D(z) must be put either in ascending powers of z or in
descending powers of z–1 before performing long division.

w .Ea
X ( z)
z2  z  2
z 3  2z 2  3z  4
2  z  z2
4  3z  2z 2  z 3

syE
4 + 3z – 2z2 + z3
1 1
2 8
19
 z  z2 
32
2 + z + z2
81 3 411 4
128
z 
512
z

3 1
ngi
2  z  z 2  z3
2 2
1 1
 z  2z2  z3
2 2 nee
1 3 1
 z  z 2  z3  z 4
2 8 4
1
8 rin
19 2 3 3 1 4
8
z  z  z
4 8 g.n
19 2 57 3 19 4 19 5
8
z 
32
81
z  z 
16
21
 z3  z 4  z5
32
19
z
e t
32 16 32
81 243 4 81 5 81 6
 z3  z  z  z
32 128 64 128
411 4 81 6
z  129 z 5  z
128 128

\ X (z )
1 1
2 8
19
 z  z2 
32
81 3 411 4
128
z 
512
z !

www.EasyEngineering.net
 www.EasyEngineering.net

220 h Digital Signal Processing

\
Ñ , !
Î 411
,
81 19 1 1 Þ
, , , Ñ
x ( n) Ï 512 128 32 8 2 ß
Ñ Ñ
Ð à

2  z  z2 2 z 3  z 2  z 1
Also X ( z)
4  3z  2 z 2  z 3 4 z 3  3z 2  2 z 1  1

1 1
2 8
19
 z  z2 
32
81 3 411 4
128
z 
512
z  "

ww 4z–3 + 3z–2 – 2z–1 + 1 2z–3 + z–2 + z–1

3
2 z 3  z 2  z 1 
2
1
2

w .Ea
1 2
 z  2 z 1 
2
1 2 3 1 1 1
 z  z   z
1
2

syE 2 8
19 1 3 1
8
4 8
z   z
4 8

ngi
19 1 57 19
8
z 
19
 z  z2
32 16 32

nee 81 21
  z  z2
32 16
19
32

rin
18 243 81 2 81 3
  z z  z
32 128 64 128

g.n
411 119 2 81 3
z z  z
128 69 128

\ X ( z)
1 1
2 8
19
 z  z2 
32
81 3 411 4
128
z 
512
z  "
e t
\
x ( n)
Î
!
Ï ,
Ð
411
,
81 19 1 1Þ
, , , ß
512 128 32 8 2 à

Observe that both the methods give the same sequence x(n).
We can say from the above examples that, this method does not give x(n) in a closed
form expression in terms of n, and hence, is useful only if one is interested in determining
the first few terms of the sequence x(n).

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 221

3.6.2 Partial Fraction Expansion Method

To find the inverse Z-transform of X(z) using partial fraction expansion method, its
denominator must be in factored form. It is similar to the partial fraction expansion method
used earlier for the inversion of Laplace transforms. However, in this case, we try to obtain
the partial fraction expansion of X(z)/z instead of X(z). This is because, the Z-transform of
time domain signals have z in their numerators. This method can be applied only if X(z)/z is
a proper rational function (i.e. the order of its denominator is greater than the order of its
numerator). If X(z)/z is not proper, then it should be written as the sum of a polynomial and
a proper function before applying this method. The disadvantage of this method is that, the
denominator must be factored. Using known Z-transform pairs and the properties of

ww
Z-transform, the inverse Z-transform of each partial fraction can be found.
Consider a rational function X(z)/z given by

"
w .Ea
X ( z ) b0 z M + b1 z M 1 + b2 z M  2 + + bM
z
When M < N, it is a proper function.
= N
z + a1 z N 1 + a2 z N  2 + + aN "

X (z )

s
When M ³ N, it is not a proper function, so write it as:

yE " N
c0 z N  M  c1 z N  M  1   cN  M 
N1 ( z )
z
ngi polynomial
D ( z)
Proper rational function

nee
There are two cases for the proper rational function X(z)/z.
CASE 1 X(z)/z has all distinct poles.
When all the poles of X(z)/z are distinct, then X(z)/z can be expanded in the form

"  z C P
rin
X ( z)
z
C1 C
 2 
z  P1 z  P2
N

N
g.n
The coefficients C1, C2, ¼, CN can be determined using the formula

Ck ( z  Pk )
X (z )
z z Pk
,k 1, 2, !, N
e t
CASE 2 X(z)/z has l-repeated poles and the remaining N-l poles are simple. Let us say the
kth pole is repeated l times. Then, X(z)/z can be written as:

X ( z) C1 C
 2  "  z C P k1

Ck 2
 "  (z C P )
kl
z

z  P1 z  P2
(N -l ) terms
k (z  Pk ) 2
k
l

X (z )
where Ckl ( z  Pk )l
z z Pk

www.EasyEngineering.net
 www.EasyEngineering.net

222 h Digital Signal Processing

1 d l i Ë X ( z) Û
In general, Cki l i Ì
( z  Pk )l
(l  i)! dz Í z ÜÝ z Pk

If X(z) has a complex pole, then the partial fraction can be expressed as:
X (z ) C1 C*
 1 *
z z  P1 z  P1
where C1* is complex conjugate of C1 and P1* is complex conjugate of P1.
In other words, complex conjugate poles result in complex conjugate coefficients in the
partial fraction expansion.

ww
EXAMPLE 3.31 Find the inverse Z-transform of

w .Ea
X ( z)
z 1
3  4 z 1  z 2

z 1
; ROC; z ! 1

syE
z
Solution: Given X (z ) 1 2
3  4z z 3z  4 z  1
2

z 1 z
3 [ z  (4 z/3)  (1/3)]
2

ngi 3 ( z  1) [ z  (1/3)]

\
X (z )
z
1 1
3 ( z  1) [ z  (1/3)]
nee z 1
A

B
z  (1/3)
where A and B can be evaluated as follows:
rin
g.n
X (z ) 1 1 1 1 1
A (z  1) (z  1)
z z 1 3 (z  1) [ z  (1/3)] z 1
3 1  (1/3) 2

B
È
Éz 
Ê
1 Ø X (z )
Ù
3Ú z z 1/3
È
Éz 
Ê
1Ø 1
Ù
1
3 Ú 3 ( z  1) [ z  (1/3)] z 1/3
1 1
3 (1/3)  1e 
1
2
t
X (z ) 1 1 1 1
\ 
z 2 z  1 2 z  (1/3)

1Ë z z Û
or X ( z) Ì  Ü ; ROC; | z | > 1
2 Í z  1 z  (1/3) Ý
Since ROC is |z| > 1, both the sequences must be causal. Therefore, taking inverse Z-transform,
we have
n
1Ë È 1Ø Û
x ( n) Ìu(n)  É Ù u( n) Ü ; ROC; z !1
2 ÍÌ Ê 3Ú
ÝÜ

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 223

EXAMPLE 3.32 Find the inverse Z-transform of

z ( z  1)
X ( z) ; ROC; z ! 2
( z  1)3 ( z  2)

z ( z  1)
Solution: Given X ( z) ; ROC; z ! 2
( z  1)3 ( z  2)

X (z ) z 1 C1 C2 C3 C
\    4
z (z  1) (z  2)
3 z  1 ( z  1) 2
(z  1)3 z2

ww
where the constants C1, C2, C3 and C4 can be obtained as follows:

w C4 ( z  2)
X ( z)

.Eaz z

X (z )
2
z 1
( z  1)

z 1
3
z 2
2  1
( 2  1)3

1  1
3

C3 ( z  1)3

1 d Ë
syE
z z 1

X (z ) Û
( z  2) z 1
1  2

d È z 1Ø
2

( z  2) (1)  ( z  1) (1)
C2 Ì
1! dz Í
( z  1)3
z ÜÝ z 1
ngi
dz ÉÊ z  2 ÙÚ z 1 ( z  2) 2 z 1
3

C1
1 d
2! dz 2
2
Ë 3 X ( z) Û
ÌÍ( z  1) z ÜÝ
z 1
2

nee
1 d È z  1Ø
2! dz 2 ÉÊ z  2 ÙÚ z 1

1 d Ë 3 Û
Ì Ü
1 3 – 2( z  2) 3(1  2)
3 rin
2! dz Í ( z  2) 2 Ý
z 1
2 ( z  2) 4 z 1 ( 1  2)3
g.n
\
X (z )
z
3z 3z
3

3
z  1 ( z  1)
2z
2

2
( z  1)
3z
3

3
z2 e t
\ X (z )    ; ROC; | z | > 2
z  1 ( z  1) 2
( z  1) 3 z2

Since ROC is | z | > 2, all the above sequences must be causal. Taking inverse Z-transform
on both sides, we have
x ( n) 3(1)n u(n)  3n(1)n u(n)  2(n) (n  1) (1)n u(n)  3( 2)n u(n)
[ 3  3n  2n( n  1)] ( 1) n u( n)  3( 2) n u( n)

www.EasyEngineering.net
 www.EasyEngineering.net

224 h Digital Signal Processing

EXAMPLE 3.33 Determine all possible signals x(n) associated with Z-transform.

(1/4) z 1
X (z ) =
[1  (1/2) z 1 ] [1  (1/4) z 1 ]

(1/4) z 1
Solution: Given X (z ) =
[1  (1/2) z 1 ] [1  (1/4) z 1 ]

Multiplying the numerator and denominator with z2, we obtain

(1/4) z

ww X (z )
[ z  (1/2)] [ z  (1/4)]
Now, X(z) has two poles, one at z = (1/2) and the other at z = 1/4 as shown in Figure 3.7.

w
The possible ROCs are:

(a) ROC; z !
1
2
.Ea (b) ROC; z 
1
4
(c) ROC;
1
4
z 
1
2

Now,
X (z ) syE
Hence there are three possible signals x(n) corresponding to these ROCs.
1/ 4 C1

C2 1

1

or
z [ z  (1/2)] [ z  (1/4)]

X ( z)
z
 ngi
z  (1/2) z  (1/4)
z
z  (1/2) z  (1/4)

(a) ROC; z !
1 nee
z  (1/2) z  (1/4)

2
Here both the poles, i.e. z = (1/2) and z = (1/4) correspond to causal terms.
rin
\ x ( n)
È 1Ø
É Ù
Ê 2Ú
n
u( n)
È 1Ø
É Ù
Ê 4Ú
n
u( n)
g.n
(b) ROC; z 
1
4
Here both the poles must correspond to anticausal terms.
n n
e t
È1Ø È 1Ø
\ x ( n) É Ù u (  n  1)  É Ù u(  n  1)
Ê2Ú Ê 4Ú

1 1
(c) ROC; z 
4 2
Here the pole at z = (1/4) must correspond to causal term and the pole at z = (1/2)
must correspond to anticausal term.
n n
È1Ø È 1Ø
\ x ( n) É Ù u (  n  1)  É Ù u( n)
Ê 2Ú Ê 4Ú

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 225

The ROCs are shown in Figure 3.7.

ww
w .Ea Figure 3.7 ROCs for Example 3.33.

EXAMPLE 3.34
syE
Find all possible inverse Z-transforms of the following function:

X (z )
z( z 2  4 z  5)

z ( z 2  4 z  5)
ngi
z 3  6 z 2  11z  6

z ( z 2  4 z  5)
Solution: Given X (z )
z 3  6 z 2  11z  6
nee( z  1) ( z  2) ( z  3)

The poles of X(z) are at z = 1, z = 2 and z = 3. So the possible ROCs are:

(a) | z | > 3 (b) | z | < 1 (c) 2 < | z | < 3 and rin (d) 1 < | z | < 2
Using partial fraction method, we have
g.n
or
X (z )
z
z 2  4z  5
( z  1)( z  2)( z  3)

X (z )
A
z 1
z



B
z 2
z



C
z 3
z
1
z 1

1
z 2

e 1
z 3
t
z 1 z 2 z 3
Therefore, the possible inverse Z-transforms are:
(a) x ( n) u( n)  2 n u( n)  3 n u( n); ROC; z ! 3
(b) x ( n) u(  n  1)  2n u( n  1)  3n u(  n  1); ROC; z  1
(c) x ( n) u(n)  2 n u(n)  3 n u(  n  1); ROC; 2  z  3
(d) x ( n) u( n)  2 n u(  n  1)  3n u(  n  1); ROC; 1  z  2

www.EasyEngineering.net
 www.EasyEngineering.net

226 h Digital Signal Processing

EXAMPLE 3.35 Determine the causal signal x(n) having Z-transform

z2 + z
X ( z) =
[z  (1/2)]2 [z  (1/4)]

z2 + z z (z + 1)
Solution: Given X ( z ) = =
[z  (1/2)] [z  (1/4)] [z  (1/2)]2 [z  (1/4)]
2

X (z )
Taking partial fractions of , we have
z

ww X (z )
z
=
(z + 1)
=
A
[z  (1/2)] [z  (1/4)] [z  (1/2)]
2 2
+
[ z 
B
(1/2)]
+
[z 
C
(1/4)]

w =
.Ea
6
[z  (1/2)]2

20
+
20
[ z  (1/2)] [z  (1/4)]

\ X (z ) = 6 syEz
[ z  (1/2)]2
 20
z
[ z  (1/2)]
+ 20
z
[z  (1/4)]

ngi
Taking inverse Z-transform on both sides, we have the causal signal

1
x (n) = 6n  
2
n 1
1
2nee n
1
u(n)  20   u(n) + 20   u( n)
4
n

3.6.3 Residue Method rin

The inverse Z-transform of X(z) can be obtained using the equation:
g.n
x (n)
1
2Q j v
Ô X (z ) z
c
n 1
dz

where c is a circle in the z-plane in the ROC of X(z). The above equation can be evaluated
e t
by finding the sum of all residues of the poles that are inside the circle c. Therefore,

x ( n) Ç Residues of X (z)z n 1
at the poles inside c

Ç ( z  z ) X ( z) z z
i
i
n 1
zi

If X(z) zn–1 has no poles inside the contour c for one or more values of n, then x(n) = 0 for
these values.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 227

EXAMPLE 3.36 Using residue method, find the inverse Z-transform of

1  2z 1
X ( z) ; ROC; z ! 3
1  4 z 1  3z 2

1  2 z 1 z( z  2) z ( z  2)
Solution: Given X (z ) 1 2 ( z  1)( z  3)
1  4z  3z z  4z  3
2

\ x ( n) Ç Residues of X (z) z at the poles of X (z) z within c

n 1 n 1

Ç Residues of (zz(z1)2)(zz 3) (zz 1)(

n 1
( z  2) n

ww z  3)
at the poles of same within c

Ç Residues of (zz(1)z (z2) 3) at poles z

n
\
w x ( n)

.Ea ( z  1)
z n ( z  2)

1 and z 3

( z  3) z n ( z  2)

syE
1
( z  1) ( z  3)

1
( 1) n u( n)  ( 3) n u( n)
z 1
( z  1) ( z  3)
z 3

2 2
ngi
nee
EXAMPLE 3.37 Determine the inverse Z-transform using the complex integral

3 z 1 1
X ( z) ; ROC; z !

rin
1 2
[1  (1/2) z ] 4

Solution:
equation:
We know that the inverse Z-transform of X(z) can be obtained using the

g.n
x ( n)
1
2Q j v
Ô X ( z) z
c
n 1

at the poles inside c where c is a circle in the z-plane in the ROC of X(z).
dz
e t
This can be evaluated by finding the sum of all residues of the poles that are inside the
circle c. Therefore, the above equation can be written as:

x ( n) Ç Residues of X (z) z n 1
at the poles inside c

Ç ( z  z ) X ( z) z
i
i
n 1
z zi

If there is a pole of multiplicity k, then the residue at that pole is:

1 d k 1
[( z  zi ) k X ( z) z n 1 ] at the pole z = zi
(k  1)! dz k 1

www.EasyEngineering.net
 www.EasyEngineering.net

228 h Digital Signal Processing

3 z 1 3z
Given X ( z) 1 2
[1  (1/2) z ] [ z  (1/2)]2

The given X(z) has a pole of order 2 at z = 1/2.

x ( n) Ç Residues of X (z) z at its poles

n 1

Ç Residue of 3z /[z  (1/ 2)] at the pole z = (1/2) of multiplicity 2

n 2

ËÈ 2 Û n 1
1 d 1Ø 3z n È1Ø
\ x ( n) ÌÉ z  Ù Ü 3nz n 1 3n É Ù u( n)
ÌÍ Ê 2 Ú [ z  (1/ 2)] ÝÜ Ê 2Ú
1! dz 2 z 1/2

ww
EXAMPLE 3.38
z 1/2

Find all possible inverse Z-transforms of X ( z)

z
using

w
contour integration (residue) method.

Solution: Given .Ea X ( z)

z
( z  1) ( z  2)3
2

syE ( z  1) ( z  2)3
2

X(z) has two poles, one of order 2 at z = –1 and the second one of order 3 at z = –2. So
there are three possible inverse Z-transforms:
(a) with ROC; |z| > 2
(b) with ROC; |z| < 1 ngi
(c) with ROC; 1 < |z| < 2
nee
We know that x(n) is given by the sum of the residues of X(z) zn–1 at the poles of X(z).

Residue of X ( z ) z n 1
zn
at the pole z = –1 of order 2 is: rin
1 d Ë
( z  1)2 ( z  2)3

zn Û d Ë zn Û g.n
e
2
Ì( z  1) Ü Ì Ü
1! dz ÍÌ ( z  1) 2 ( z  2)3 ÝÜ z 1
dz ÍÌ ( z  2)3 ÝÜ
z 1

( z  2)3 ( nz n 1 )  z n 3( z  2)2
( z  2)6 z 1
t
n
 ( n  3) ( 1)

zn
Residue of X ( z ) z n 1 at the pole z = –2 of order 3 is:
( z  1)2 ( z  2)3

1 d2 Ë 3 zn Û 1 d2 Ë zn Û
Ì ( z  2) Ü Ì Ü
2
2! dz ÍÌ 2 3 2 dz 2 ÍÌ ( z  1)2 ÝÜ
( z  1) ( z  2) ÝÜ
z 2 z 2
n 2
( 2) [0.125n  1.125n  3]

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 229

(a) When ROC; | z | > 2, both the residues must be positive.

\ x ( n)  (n  3)(1)n u(n)  (0.125n2  1.125n  3)( 2)n u(n)
(b) When ROC; | z | < 1, both the residues must be negative.
\ x ( n) (n  3) ( 1) n u(  n  1)  (0.125n2  1.125n  3)( 2) n u(  n  1)
(c) When ROC; 1 £ | z | £ 2, residue at pole z = –1 is positive and the residue at pole
at z = –2 is negative.
\ x ( n)  (n  3)(1)n u(n)  (0.125n2  1.125n  3)(2)n u(  n  1)

ww
3.6.4 Convolution Method
The inverse Z-transform can also be determined using convolution method. In this method,

w
the given X(z) is splitted into X1(z) and X2(z) such that X(z) = X1(z) X2(z). Then, x1(n) and
x2(n) are obtained by taking the inverse Z-transform of X1(z) and X2(z) respectively. Then,

.Ea
x(n) is obtained by performing convolution of x1(n) and x2(n) in time domain.
Z [ x1 ( n) x2 (n)] X1 ( z ) X 2 ( z ) X (z )

\ x ( n) syE
Z 1 [ X ( z )] Z 1 [ Z {x1 ( n) * x2 ( n)}] x1 (n) * x2 (n) Ç x (k ) x (n  k )
n
1 2

ngi
EXAMPLE 3.39 Explain how the analysis of discrete time-invariant system can be
k 0

obtained using convolution properties of Z-transform.

nee
Solution: One of the most important properties of Z-transform used in the analysis of
discrete-time systems is convolution property. According to this property, the Z-transform of
the convolution of two signals is equal to the multiplication of their Z-transforms. The
rin
analysis of a discrete-time system means the input to the system x(n) and its impulse
response h(n) are known and we have to determine the output y(n) of the system. If the input
sequence to the LTI system is x(n) and the impulse response is h(n), then first determine the g.n
Z-transforms of x(n) and h(n).
Let Z [x(n)] = X(z) and Z [h(n)] = H(z), then obtain the product of these Z-transforms,
i.e. Y(z) = H(z) X(z). We know that the output y(n) = x(n) * h(n).
As per the linear convolution property,
e t
Y ( z) Z[ x (n) * h(n)] Z[ x ( n)] Z[ h( n)] X ( z) H ( z)

So having obtained Y(z) = X(z) H(z), take the inverse Z-transform of X(z) H(z). This gives
y(n) which is the response of the system for the input x(n).

z2
EXAMPLE 3.40 Find the inverse Z-transform of X ( z ) using convolution
( z  2) ( z  3)
property of Z-transforms.

www.EasyEngineering.net
 www.EasyEngineering.net

230 h Digital Signal Processing

z2
Solution: Given X (z )
( z  2) ( z  3)
z z
Let X ( z) X1 ( z) X 2 ( z )
z 2 z 3
È z Ø
\ x1 ( n) Z 1 [ X1 ( z )] Z 1 É Ù 2 n u( n)
Êz  2Ú

È z Ø
x 2 ( n) Z 1 [ X 2 ( z )] Z 1 É Ù 3n u( n)

ww
\ x1 ( n) * x 2 ( n) Ç x (k ) x (n  k )
n
1 2
Ê z  3Ú

w .Ea
k 0

k
Ç 2 u( k ) 3
n

0
k nk
u( n  k )

syÇE 3n
k 0
n
È 2Ø
ÉÊ 3 ÙÚ
k
3n Ì
Ë 1  (2/3)n 1 Û
ÌÍ 1  (2/3) ÜÝ
Ü

Ë È 2Ø
3n 1 Ì1  É Ù
ÌÍ Ê 3 Ú ngi
n 1 Û
Ü 3n 1 u(n)  2 n 1 u(n)
ÜÝ

nee z

rin
EXAMPLE 3.41 Find the inverse Z-transform of X ( z) using
( z  1) [ z  (1/ 2)]
convolution property of Z-transforms.

Solution: Given X ( z)
z
( z  1) [ z  (1/ 2)] g.n
Let X ( z) X1 ( z ) X2 ( z )
z 1
( z  1) [ z  (1/2)]
e t
È z Ø
\ x1 ( n) Z 1 [ X1 ( z)] Z 1 É Ù u( n)
Êz  1Ú
n 1
Ë 1 Û È 1Ø
and x 2 ( n) Z 1 [ X2 ( z)] Z 1 Ì Ü É Ù u(n  1)
Í z  (1/2) Ý Ê 2Ú

Ç x (k) x (n  k )
n
\ x1 (n) * x2 ( n) 1 2
k 0

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 231

n 1 n 1 k

Ç u( k )
k 0
È1Ø
É Ù
Ê2Ú
u( n  1  k )

n 1 Ë n 1 k n 1 n 1 Ë 1 k
È1Ø
É
Ê Ù
Ú
2
Ì Ç È 1Ø Û
É Ù
ÌÍ k 0 Ê 2 Ú
Ü
Ü
Ý
È1Ø
É
Ê Ù
2Ú
Ç
k 0 Ì
Í
ÌÉ Ù
È1Ø Û
Ê 2Ú
Ü
Ü
Ý

n 1 n 1
È 1Ø ÑÎ1  [(1/2) 1 ]n ÑÞ È 1 Ø Ë 1  (1/2)  n Û
ÉÊ 2 ÙÚ Ï 1 ß ÉÊ 2 ÙÚ Ì Ü
Ñ 1  (1/2) à
Ð Ñ ÍÌ 1 ÝÜ
1 1

ww
n n
È 1Ø È 1Ø È 1Ø È1Ø
ÉÊ 2 ÙÚ É Ù É Ù 2u(n)  2 É Ù u( n)
Ê 2Ú Ê 2Ú Ê 2Ú

w .Ea
3.7 TRANSFORM ANALYSIS OF LTI SYSTEMS
The Z-transform plays an important role in the analysis and design of discrete-time LTI systems.

syE
3.7.1 System Function and Impulse Response

ngi
Consider a discrete-time LTI system having an impulse response h(n) as shown in Figure 3.8.

nee
Figure 3.8 Discrete-time LTI system.

Let us say it gives an output y(n) for an input x (n). Then, we have rin
y( n) x ( n) * h(n)
g.n
Taking Z-transform on both sides, we get

where
Y ( z) X ( z) H ( z ) e t
Y(z) = Z-transform of the output y(n)
X(z) = Z-transform of the input x(n)
H(z) = Z-transform of the impulse response h(n)

Y ( z)
\ H ( z)
X ( z)
H(z) is called the system function or the transfer function of the LTI discrete system and is
defined as:
The ratio of the Z-transform of the output sequence y(n) to the Z-transform of the input
sequence x(n) when the initial conditions are neglected.

www.EasyEngineering.net
 www.EasyEngineering.net

232 h Digital Signal Processing

If the input x(n) is an impulse sequence, then X(z) = 1. So Y(z) = H(z). So the transfer
function is also defined as the Z-transform of the impulse response of the system.
The poles and zeros of the system function offer an insight into the system
characteristics. The poles of the system are defined as the values of z for which the system
function H(z) is infinity and the zeros of the system are the values of z for which the system
function H(z) is zero.

3.7.2 Relationship between Transfer Function and Difference Equation

In terms of a difference equation, an nth order discrete-time LTI system is specified as:

ww Ç
N

k 0
ak y (n  k ) Ç b x(n  k )
M

k 0
k

w
Expanding it, we have

.Ea
a0 y(n)  a1 y(n  1)  a2 y(n  2)  "a N y (n  N ) b0 x(n)  b1 x(n  1)  b2 x (n  2)
"
  bM x (n  M )

sy"E
Taking Z-transform on both sides and neglecting the initial conditions, we obtain

a0Y ( z )  a1 z 1Y ( z)  a2 z 2Y ( z )   aN z  N Y ( z) b0 X (z)  b1 z 1 X (z )  b2 z 2 X (z )

ngi" "   bM z  M X ( z)

i.e. Y ( z) [ a0  a1 z 1  a2 z 2  " a Nz
N
]
n
X ( z)[b0  b1 z 1  b2 z 2 

" ee
 bM z  M ]

\ Y ( z)
X (z )
b0  b1 z 1  b2 z 2 
a0  a1 z 1  a2 z 2 " rin  b M zM
  aN z  N

Çb z
M
k g.n
e
k
k 0

Ça z
N

k 0
k
k

Now, Y(z)/X(z) = H(z) is called the transfer function of the system or the system function.
t
The frequency response of a system is obtained by substituting z = e jw in H(z).

3.8 STABILITY AND CAUSALITY

We know that the necessary and sufficient condition for a causal linear time-invariant
discrete-time system to be BIBO stable is:
‡

n 0
Ç h(n)  ‡

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 233

i.e. an LTI discrete-time system is BIBO stable if its impulse response is absolutely
summable.
We also know that for a system to be causal, its impulse response must be equal to
zero for n < 0 [i.e. h(n) = 0 for n < 0]. Alternately, if the system is causal, then the ROC for
H(z) will be outside the outermost pole.
For a causal LTI system to be stable, all the poles of H(z) must lie inside the unit circle
in the z-plane, i.e. for a causal LTI system to be stable, the ROC of the system function must
include the unit circle.

1
EXAMPLE 3.42 Consider an LTI system with a system function H ( z ) .
1  (1/2) z 1

ww
Find the difference equation. Determine the stability.

Y ( z) 1 z

w
Solution:

That is
Given

.Ea
H ( z)
X ( z)

Y ( z) 
1  (1/2) z

1 1
z Y ( z)
1
=
z  (1/2)

X ( z)

syE 2

Taking inverse Z-transform on both sides (applying the time shifting property), we get the
difference equation

y( n) 
1 ngi
y( n  1) x ( n)
2
nee
The only pole of H(z) is at z = 1/2, i.e., inside the unit cirle. So the system is stable.

EXAMPLE 3.43 A causal system is represented by

rin
H (z )
z2
2 z 2  3z  4 g.n
Find the difference equation and the frequency response of the system.

Solution: Given H (z )
z2
e t
2 z 2  3z  4

As the system is causal, H(z) is expressed in negative powers of z.

Y (z )
\ H (z )
X ( z)
z2 z 1  2 z 2
2 z 2  3z  4 2  3z 1  4 z 2

www.EasyEngineering.net
 www.EasyEngineering.net

234 h Digital Signal Processing

i.e. 2Y ( z )  3z 1 Y ( z)  4 z 2Y ( z) z 1 X ( z )  2 z 2 X ( z)

Taking inverse Z-transform on both sides, we have

2 y(n)  3y(n  1)  4 y(n  2) x (n  1)  2 x (n  2)
which is the required difference equation.
Putting z = e jw in H(z), we get the frequency response H(w) of the system.

z 2 e jX  2
H (X )
2z 2  3z  4 z e jX
2e j 2X  3e jX  4

ww 2  cos X  j sin X
4  (2 cos 2X  3 cos X )  j(2 sin 2X  3 sin X )

w .Ea
EXAMPLE 3.44 Determine the system function of a discrete-time system described by the
difference equation

syE y( n) 
1
3
1
y( n  1)  y( n  2)
5
x (n)  2 x ( n  1)

Solution:

Y ( z) 
1 ngi
Taking Z-transform on both sides of the given difference equation, we get

1 1
z Y ( z)  z 2Y ( z) X ( z)  2 z 1 X ( z )
3 5
nee
Hence the system function or transfer function of the given system is:

H (z )
Y (z ) 1  2 z 1 z ( z  2)
rin
g.n
1 2
X ( z) 1  (1/3) z  (1/5) z z  (1/3) z  (1/5)
2

EXAMPLE 3.45 Plot the pole-zero pattern and determine which of the following systems
are stable:
(a) y( n) y( n  1)  0.8 y ( n  2)  x (n)  x ( n  2)
(b) y( n) 2 y(n  1)  0.8 y ( n  2)  x ( n)  0.8 x ( n  1)
e t
Solution:
(a) Given y( n) y( n  1)  0.8 y ( n  2)  x ( n)  x ( n  2)
Taking Z-transform on both sides and neglecting the initial conditions, we have

Y ( z) z 1Y ( z)  0.8z 2Y ( z )  X ( z)  z 2 X ( z)

i.e. Y ( z) [1  z 1  0.8z 2 ] X ( z ) (1  z 2 )

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 235

The transfer function of the system is:

Y ( z)
H (z )
X (z )
1  z 2 z2  1
1  z 1  0.8z 2 z 2  z  0.8
( z  j )( z  j )
( z  0.5  j 0.74)( z  0.50  j 0.74)
The zeros of H(z) are z = +j1 and z = –j1.

ww The poles of H(z) are z = 0.5 – j0.74 and z = 0.5 + j0.74

The pole-zero plot is shown in Figure 3.9(a).
All the poles are inside the unit circle. Hence, the system is stable.

w .Ea
syE
ngi
nee
Figure 3.9 Pole-zero plots for Example 3.45.

2 y (n  1)  0.8 y (n  2)  x (n)  0.8 x (n  1) rin

(b) Given y(n)
Taking Z-transform on both sides and neglecting the initial conditions, we have g.n
i.e.
Y ( z) 2 z 1Y ( z )  0.8z 2Y ( z)  X ( z)  0.8z 1 X ( z )

Y ( z) [1  2 z 1  0.8z 2 ] X ( z ) [1  0.8z 1 ]
e t
The transfer function of the system is:
Y ( z) 1  0.8z 1 z ( z  0.8)
H ( z) 1 2
X ( z) 1  2z  0.8z z  2 z  0.8
2

z ( z  0.8)
( z  1.445)( z  0.555)
The zeros of H(z) are z = 0 and z = –0.8.
The poles of H(z) are z = 1.445 and z = 0.555.
The pole-zero plot is shown in Figure 3.9(b).
One pole is outside the unit circle. Therefore, the system is unstable.

www.EasyEngineering.net
 www.EasyEngineering.net

236 h Digital Signal Processing

EXAMPLE 3.46 A causal system has input x(n) and output y(n). Find the system function,
frequency response and impulse response of the system if
1 1
x ( n) E (n)  E (n  1)  E ( n  2)
6 6
2
and y( n) E ( n)  E ( n  1)
3
Also assess the stability.
1 1
Solution: Given x ( n) E (n)  E (n  1)  E ( n  2)

ww
6 6
2
and y( n) E ( n)  E ( n  1)
3

w
Taking Z-transform of the above equations, we get

.Ea X ( z) 1 
1 1 1 2
z  z

syE
6 6

2 1
and Y ( z) 1  z

ngi
3
The system function or the transfer function of the system is:

Y (z )
X ( z)
H ( z)
1  (2/3) z 1
1  (1/6) z 1
nee
 (1/6) z 2
z [ z  (2/3)]
[ z  (1/3)] [ z  (1/2)]

The frequency response of the system is:

rin
H (X )
z [ z  (2/3)]
[ z  (1 / 3)][ z  (1/2)] z e jX
e jX [e jX  (2/3)]
[e jX  (1/3)][e jX  (1/2)]g.n
By partial fraction expansion, we have

H (z ) [ z  (2/3)] A

B 2 / 5

7/5
e t
z [ z  (1/3)][ z  (1/2)] z  (1/3) z  (1/2) z  (1/3) z  (1/2)

2Ë z Û 7Ë z Û
\ H ( z)  Ü Ì
5 Í z  (1/3) Ý 5 Í z  (1/2) ÜÝ
Ì

Taking inverse Z-transform on both sides, we get the impulse response as:
n n
2 È 1Ø 7 È 1Ø
h( n)  É Ù u( n)  É  Ù u( n)
5 Ê 3Ú 5 Ê 2Ú
Both the poles of H(z) are inside the unit circle. So the system is stable.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 237

EXAMPLE 3.47 We want to design a causal discrete-time LTI system with the property
n n 1 n
È 1Ø 1 È 1Ø È 1Ø
that if the input is x ( n) É Ù u( n)  É Ù u(n  1) , then the output is y( n) É Ù u( n).
Ê 3Ú 5 Ê 3Ú Ê 2Ú

Determine the transfer function H(z), the impulse response h(n) and frequency response H(w)
of the system that satisfies this condition. Also assess the stability.

n n 1
È 1Ø 1 È 1Ø
Solution: Given x ( n) É Ù u( n)  É Ù u(n  1)
Ê 3Ú 5 Ê 3Ú

ww
and y ( n)
È 1Ø
É Ù
Ê 2Ú
n
u( n)

\ w
We want to design a causal system.

.Ea h(n) = 0 for n < 0

Taking Z-transform on both sides of the above equations, we have

X ( z) syE
z 1
 z 1
z  (1/3) 5
z
z  (1/3)
z

1 1
z  (1/3) 5 z  (1/3)
z  (1/5)
z  (1/3)

and Y (z )
z
z  (1/2)
ngi
The system function H(z) is: nee
H ( z)
Y ( z)
X (z ) rin
z
š
z  (1/5)
z  (1/2) z  (1/3)
z [ z  (1/3)]
[ z  (1/2)] [ z  (1/5)] g.n
By partial fraction expansion of H(z)/z, we get

\
H ( z) z  (1/3) A

B 5/9

4/9
e t
z [ z  (1/2)] [ z  (1 / 5)] z  (1/2) z  (1/5) z  (1/2) z  (1/5)

5Ë z Û 4 Ë z Û
\ H (z ) Ì Ü Ì Ü
9 Í z  (1/2) Ý 9 Í z  (1/5) Ý
Taking inverse Z-transform of H(z), we get the impulse response h(n) as:
n n
5 È1Ø 4 È 1Ø
h( n) É Ù u( n)  É Ù u( n)
Ê
9 2 Ú 9 Ê 5Ú

www.EasyEngineering.net
 www.EasyEngineering.net

238 h Digital Signal Processing

The frequency response is:

z [ z  (1/3)] e jX [e jX  (1/3)]
H (X )
[ z  (1/2)][ z  (1/5)] z e jX [e jX  (1/2)][e jX  (1/5)]

Both the poles of H(z) are inside the unit circle. So the system is stable.

EXAMPLE 3.48 A causal LTI system is described by the difference equation

y( n) y( n  1)  y( n  2)  x (n)  2 x (n  1)
Find the system function and frequency response of the system. Plot the poles and zeros and

ww
indicate the ROC. Also determine the stability and impulse response of the system.
Solution: The given difference equation is:

w .Ea
y(n) y(n  1)  y(n  2)  x (n)  2 x (n  1)
Taking Z-transform on both sides, we have

z 1Y ( z )  z 2Y ( z )  X ( z )  2 z 1 X ( z)

i.e. syE Y ( z)

Y ( z) (1  z 1  z 2 ) X ( z ) (1  2 z 1 )
The system function is:
1  2 z 1
ngi
z ( z  2) z ( z  2)
Y ( z)
X (z )
H (z )
1 z 1
z 2
nee
z  z 1
2
( z  1.62)( z  0.62)
The frequency response of the system is:

H (X )
z ( z  2) e jX (e jX  2) rin
( z  1.62) ( z  0.62) z e jX (e jX  1.62) (e jX  0.62)
g.n
H(z) has the zeros at z = 0 and z = –2.
H(z) has the poles at z = 1.62 and z = –0.62. One of the pole is outside the unit circle.
So the system is unstable. The poles and zeros and the ROC are shown in Figure 3.10.
To find the impulse response h(n), partial fraction expansion of H(z)/z gives
e t
H ( z) z 2 A B 1.62 0.62
 
z ( z  1.62) ( z  0.62) z  1.62 z  0.62 z  1.62 z  0.62

È z Ø È z Ø
\ H ( z ) 1.62 É Ù  0.62 É Ù
Ê z  1.62 Ú Ê z  0.62 Ú

Taking inverse Z-transform, the impulse response is:

h( n) 1.62(1.62) n u( n)  0.62 ( 0.62)n u( n)

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 239

ww
w .Ea Figure 3.10 Pole-zero plot and ROC for Example 3.48.

EXAMPLE 3.49

(a) H (z )
syE
Determine whether the following systems are both causal and stable.
3  z 1
(b) H ( z)
1  2 z 1

Solution:
1  z 1  (4/9) z 2
ngi 1  (6/5) z 1  (9/25) z 2

(a) Given H ( z )
3z 1

1  z 1  (4/9) z 2 nee
z ( z  3)
z 2  z  (4/9)
z( z  3)
[ z  (1/3)] [ z  (4/3)]
The poles of H(z) are: z = 1/3 and z = – 4/3.
For a causal system to be stable, the ROC must include the unit circle. Now, for a rin
causal H(z), the ROC is | z | > 4/3. Since one pole is lying outside the unit circle,
the given system is not both causal and stable. g.n
(b) Given H (z )
1  2 z 1
1  (6/5) z 1

The location of the poles is at z = –3/5.

 (9/25) z 2
z ( z  2)
z  (6/5) z  (9/25)
2 e
z( z  2)
[ z  (3/5)]2
t
Since all the poles are lying inside the unit circle in the z-plane, the system is both
causal and stable.

EXAMPLE 3.50 An LTI system is described by the difference equation.

9 1
y( n)  y(n  1)  y( n  2) x (n)  3 x ( n  1)
4 2
Specify the ROC of H(z), and determine h(n) for the following conditions:
(a) The system is stable.
(b) The system is causal.

www.EasyEngineering.net
 www.EasyEngineering.net

240 h Digital Signal Processing

9 1
Solution: Given y( n)  y(n  1)  y( n  2) x (n)  3 x ( n  1)
4 2
Taking Z-transform on both sides, we have

9 1 1
Y ( z)  z Y ( z)  z 2Y ( z) X ( z )  3z 1 X ( z )
4 2

È 9 1 1 2 Ø
i.e. Y ( z ) É1  z  z Ù X ( z) (1  3 z 1 )
Ê 4 2 Ú

The system function is:

ww Y ( z)
X (z )
H (z )

w .Ea
1  (9/4) z
1  3z 1
1
 (1/2)z
By partial fraction expansion of H(z)/z, we have
2
z ( z  3)
z  (9/4) z  (1/2)
2
z ( z  3)
[ z  (1/4)] ( z  2)

H ( z)
z
syE ( z  3)
[ z  (1/4)] ( z  2)
A

B
z  (1/4) z  2
11 z

4 z
7 z  (1/4) 7 z  2

The system function has poles at z = 1/4 and z = 2.

ngi
is
1
 z  2. nee
(a) For the system to be stable, its ROC must include the unit circle. Therefore, ROC

4
Taking the inverse Z-transform of H(z), we have the impulse response as:
rin
h( n)
11 È 1 Ø
n
4 n
É Ù u(n)  (2) u(  n  1)
7 Ê 4Ú 7 g.n
Here the system is stable but non-causal.
(b) For the system to be causal, the ROC should be outside the outermost pole.
\ ROC is |z| > 2
e t
n
11 È 1 Ø 4 n
\ h( n) É Ù u( n)  (2) u( n)
7 Ê 4Ú 7
Here the system is causal, but unstable.

EXAMPLE 3.51 An LTI system is described by the equation

y( n) x (n)  0.81x (n  1)  0.81x (n  2)  0.45 y(n  2)
Determine the transfer function of the system. Sketch the poles and zeros on the z-plane.
Assess the stability.

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 241

Solution: The given difference equation is:

y(n) x (n)  0.81x ( n  1)  0.81x ( n  2)  0.45 y( n  2)
Taking Z-transform on both sides and neglecting initial conditions, we have
Y ( z) X ( z)  0.81z 1 X ( z)  0.81z 2 X ( z)  0.45z 2Y ( z)

i.e. Y ( z)[1  0.45z 2 ] X ( z)[1  0.81z 1  0.81z 2 ]

The transfer function of the system H(z) is:

Y ( z) 1  0.81z 1  0.81z 2 z 2  0.81z  0.81

ww H ( z)
X (z ) 1  0.45z 2
Equating the denominator to zero, we have z2 + 0.45 = 0, i.e. z2 = –0.45
z 2  0.45

\ w .Ea z
So the poles are z = +j0.67 and z = –j0.67.
“ j 0.45 “ j 0.67

syE
Equating the numerator to zero, we have

z 2  0.81z  0.81 0

\ z
0.81 “ 0.656  3.24
2 ngi
0.81 “ 3.896
2
nee
 0.81 “ 1.974  1.39, 0.582

The zeros are at z = –1.39 and z = 0.58.

rin
\
Y (z )
X (z )
( z  1.39) ( z  0.582)
( z  j 0.67) ( z  j 0.67) g.n
The pole-zero plot is shown in Figure 3.11.
The system is stable as both the poles are inside the unit circle. e t

Figure 3.11 Pole-zero plot for Example 3.51.

www.EasyEngineering.net
 www.EasyEngineering.net

242 h Digital Signal Processing

EXAMPLE 3.52 Define stable and unstable systems. Test the condition for stability of the
first order infinite impulse response (IIR) filter governed by the equation
y(n) = x(n) + bx(n – 1)
Solution: A stable system is a system for which the impulse response is absolutely
summable. For a stable system all the poles of the system function must lie inside the unit
circle centered at the origin of the z-plane. An unstable system is a system for which the
impulse response is not absolutely summable. For an unstable system one or more poles of
the system will lie either on the unit circle or outside the unit circle.
Given difference equation is:

ww
Taking Z-transform on both sides, we have
y(n) = x(n) + bx(n – 1)

w .Ea
Y ( z) X ( z )  bz 1 X ( z)
X ( z ) [1  bz 1 ]

syE
The transfer function of the system H(z) is:
Y ( z)
H ( z)
X ( z)

ngi
1  bz 1
zb
z

nee
The system has a zero at z = –b and a pole at z = 0. So the system is stable for all values of b.

EXAMPLE 3.53 Determine the impulse response and step response of the causal system
given below and discuss on stability rin
y( n)  y( n  1)  2 y( n  2) x ( n  1)  2 x ( n  2)
g.n
Solution: The given difference equation is:
y( n)  y( n  1)  2 y( n  2)

Taking Z-transform on both sides, we have

x ( n  1)  2 x ( n  2) e t
Y ( z)  z 1 Y ( z )  2z 2 Y ( z ) z 1 X ( z)  2 z 2 X ( z )

i.e. Y ( z) [1  z 1  2z 2 ] X ( z ) [ z 1  2z 2 ]
The transfer function of the system H(z) is:
Y (z )
H (z )
X ( z)
z 1  2z 2 ( z  2) ( z  2)
1 2 ( z  2)( z  1)
1 z  2z z z2
2

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 243

The impulse response of the system is:

Ë z2 Û È A B Ø
Z 1 [ H ( z)] Z 1 Ì Z 1 É 
Ü Ê z  2 z  1ÙÚ
h( n)
Í (z  2)( z  1) Ý
È4 1 Ø È1 1 Ø
Z 1 É Ù  Z 1 É
Ê 3 z  2Ú Ê 3 z  1ÙÚ
4 n 1 1
(2) u( n  1)  ( 1) n 1 u(n  1)
3 3

ww
For step response, x(n) = u(n)

\ X ( z)
z
z 1

\ w .Ea Output Y ( z) H ( z) X ( z )
z 2 z
( z  2)( z  1) z  1
Y (z )
z syE z2
( z  2) ( z  1) ( z  1)
A
z 2

B
z 1

C
z 1
4/3

1/6

3/2
z  2 z 1 z 1

\ Y ( z)
4È z Ø
3 ÉÊ z  2 ÙÚ

ngi
1È z Ø
6 ÉÊ z  1ÙÚ

3È z Ø
2 ÉÊ z  1ÙÚ

Taking inverse Z-transform on both sides, we have the step response

nee
y ( n)
4 n
3
1 3
(2) u( n)  ( 1) n u(n)  u( n)
6 2
rin
One pole of H(z) is on the unit circle and another pole is outside the unit circle. So the
system is unstable. g.n
EXAMPLE 3.54 Consider a discrete-time linear time invariant system described by the
difference equation
3 1 1
e t
y( n)  y(n  1)  y( n  2) x (n)  x ( n  1)
4 8 3
where y(n) is the output and x(n) is the input. Assuming that the system is relaxed initially,
obtain the unit sample response of the system.
Solution: The given difference equation is:
3 1 1
y( n)  y(n  1)  y( n  2) x ( n)  x ( n  1)
4 8 3
Taking Z-transform on both sides and assuming that the system is relaxed initially,
i.e. neglecting the initial conditions, we have

www.EasyEngineering.net
 www.EasyEngineering.net

244 h Digital Signal Processing

3 1 1 1
Y ( z)  z Y ( z )  z 2 Y ( z ) X ( z)  z 1 X ( z)
4 8 3

È 3 1 1 2 Ø È 1 Ø
i.e. Y ( z ) É1  z  z Ù X ( z) É1  z 1 Ù
Ê 4 8 Ú Ê 3 Ú

The system function H(z) is:

Y (z )
H (z )
X (z )

ww 1  (3/4) z
1  (1/3) z 1
1
 (1/8) z 2
z [ z  (1/3)]
z  (3/4) z  (1/8)
2

w
Taking partial fractions of H(z)/z, we have

H ( z)
z .Ea [ z  (1/3)]
[ z  (1/2)] [ z  (1/4)]
A

B
z  (1/2) z  (1/4)
10/3

7/3
z  (1/2) z  (1/4)

\ syE H ( z)
10 z

7 z
3 z  (1/2) 3 z  (1/4)

ngi
Taking inverse Z-transform on both sides, the unit sample response of the system is:

h( n)
10 È 1 Ø
n

nee
7 È 1Ø
É Ù u( n)  É Ù u( n)
Ê
3 2 Ú 3 Ê 4Ú
n

EXAMPLE 3.55 Find the

rin
(a) Impulse response
(b) Output response for a step input applied at n = 0 of a discrete-time linear time-
g.n
e
invariant system whose difference equation is given by
y(n)
(c) Stability of the system
Solution:
y(n  1)  0.5y(n  2)  x (n)  x (n  1)
t
(a) The given difference equation is:
y ( n) y( n  1)  0.5y( n  2)  x (n)  x ( n  1)
Taking Z-transform on both sides and neglecting the initial conditions, we have

Y ( z) z 1Y ( z)  0.5z 2Y ( z)  X ( z)  z 1 X ( z)

i.e. Y ( z) (1  z 1  0.5z 2 ) X ( z ) (1  z 1 )

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 245

The system function H(z) is:

Y (z )
H (z )
X ( z)
1  z 1 z ( z  1)
1 2
1 z  0.5z z  z  0.5
2

1“ 1 2 1
z 2  z  0.5 0, z “ 0.75 0.5 “ 0.866 1.366,  0.366
2 2
z ( z  1)

ww \ H ( z)
( z  1.366) ( z  0.366)
To find the impulse response h(n), we have to take the inverse Z-transform of H(z).

w Taking partial fractions of H(z)/z, we have

H ( z)
z .Ea z 1
( z  1.366)(z  0.366)
A

B
z  1.366 z  0.366
1.366

0.366
z  1.366 z  0.366

\ syE H ( z) 1.366
z
z  1.366
 0.366
z
z  0.366

ngi
Taking inverse Z-transform, the impulse response of the system is:

(b) For a unit step input, x(n) = u(n) nee

h( n) 1.366 (1.366)n u(n)  0.366( 0.366) n u( n)

\ X ( z)
z
z 1 rin
Therefore, the output for a step input is:
g.n
Y ( z)
z ( z  1) z
( z  1.366) ( z  0.366) ( z  1)
Taking partial fractions of Y(z)/z, we have
e t
Y (z ) z ( z  1) A B C
 
z ( z  1.366) ( z  0.366) ( z  1) z  1.366 z  0.366 z  1
5 0.1 4
 
z  1.366 z  0.366 z  1
z z z
\ Y ( z) 5  0.1 4
z  1.366 z  0.366 z 1

www.EasyEngineering.net
 www.EasyEngineering.net

246 h Digital Signal Processing

Taking inverse Z-transform on both sides, the output response for a step input is:
y( n) 5(1.366) n u( n)  0.1( 0.366) n u( n)  4u(n)
(c) H(z) has one pole outside the unit circle. So the system is unstable.

z2  z + 1
EXAMPLE 3.56 Check the stability of the filter for H (z) =
1
z2  z +
2
z2  z + 1 z2  z + 1
Solution: Given H (z ) = =
( z  P1 ) (z  P2 )

ww
1
z2  z +
2

1  1 2 1 j

\
w
The poles are at

.Ea
z=

H (z ) =
2

z2  z + 1
=
2

syE 
z 

1
2
1 1 1
+ j z   j 
2  2 2

ngi
Since the magnitude of the poles z < 1, both the poles are inside the unit circle and the
filter is stable.

EXAMPLE 3.57 Test the stability of the system

nee
y(n)  y(n  1) = x (n) + x (n  1)

Solution: Given y(n)  y( n  1) = x (n) + x (n  1) rin

Taking Z-transform on both sides, we have
g.n
Y (z)  z 1Y (z) = X (z ) + z 1 X (z)

Y (z)[1  z 1 ] = X (z )[1 + z 1 ]
1
e t
\ Y ( z) 1 + z z+1
H (z ) = = 1
=
X (z ) 1  z z 1

The pole is at z = 1. So the system is unstable.

EXAMPLE 3.58 Check the stability condition for the DSP systems described by the
following equations.
n
(a) y(n) = a u(n)
(b) y(n) = x (n) + e y (n  1)
a

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 247

Solution:
(a) Given y(n) = a n u(n)
Taking Z-transform on both sides, we have
z
Y ( z) =
za
Here the pole is at z = a and hence for the system to be stable |a| < 1.
(b) Given y(n) = x (n) + e a y (n  1)
Taking Z-transform on both sides, we have

ww Y ( z) = X (z)  ea z 1Y (z)

w i.e.

or
.Ea Y ( z)
X ( z)
Y ( z)[1  e a z 1 ] = X (z )

= H (z ) =
1
a 1
1e z
=
z
z  ea

syE
Here the pole is at z = ea
Hence the condition for stability is | ea | < 1, i.e. a < 0.

ngi
EXAMPLE 3.59 Determine the impulse response of the system described by the difference
equation y( n)  3 y ( n  1)  4 y (n  2) x ( n)  2 x ( n  1) using Z-transform.

nee
Solution: Given the difference equation y( n)  3 y ( n  1)  4 y ( n  2) x (n)  2 x (n  1)
Taking Z-transform on both sides, we have
Y ( z)  3z 1Y ( z)  4 z 2Y ( z ) X ( z )  2 z 1 X ( z) rin
i.e. Y ( z) (1  3z 1  4z 2 ) X ( z) (1  2 z 1 )
g.n
The transfer function of the system is:

H ( z)
Y ( z)
X ( z)
e t
1  2 z 1 z ( z  2)
1 2
1  3z  4z z  3z  4
2

Taking partial fractions of H(z)/z, we have

H ( z) z2 z2 1.2 0.2

z z 2  3z  4 ( z  4) ( z  1) z  4 z 1

È z Ø È z Ø
\ H ( z) 1.2 É Ù  0.2 É Ù
Êz  4Ú Ê z  1Ú

www.EasyEngineering.net
 www.EasyEngineering.net

248 h Digital Signal Processing

Taking inverse Z-transform on both sides, we have the impulse response

h( n) 1.2(4) n u(n)  0.2( 1) n u( n)

EXAMPLE 3.60 Determine the impulse response and step response of the causal system
given below and discuss on stability
y(n) + y(n  1)  2 y(n  2) x (n  1)  2 x (n  2)

Solution: The given difference equation is

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

ww
Taking Z-transform on both sides, we have

Y ( z) + z 1 Y ( z)  2 z 2 Y ( z) z 1 X ( z)  2 z 2 X ( z)

w
i.e.

.Ea Y ( z) [1 + z 1  2 z 2 ]

The system transfer function H(z) is

X ( z ) [ z 1  2z 2 ]

syE
H ( z)
Y ( z)
X (z )
z 1  2 z 2
1 z 1
 2z 2
z z2
2
ngi
z2 z2
( z  2)( z  1)
=
1
z 1

Taking inverse Z-transform, the impulse response of the system is

nee
h(n) = u(n – 1)
For step response, x(n) = u(n)
rin
\ X ( z)
z 1
z
g.n
\ The output Y ( z) X ( z) H ( z)

Taking inverse Z-transform on both sides, the step response is

z 1
( z  1) z  1
=
z
(z  1)2 e t
y(n) = nu(n – 1)

3.9 SOLUTION OF DIFFERENCE EQUATIONS USING Z-TRANSFORMS

To solve the difference equation, first it is converted into algebraic equation by taking its
Z-transform. The solution is obtained in z-domain and the time domain solution is obtained
by taking its inverse Z-transform.
The system response has two components. The source free response and the forced
response. The response of the system due to input alone when the initial conditions are

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 249

neglected is called the forced response of the system. It is also called the steady state
response of the system. It represents the component of the response due to the driving force.
The response of the system due to initial conditions alone when the input is neglected is
called the free or natural response of the system. It is also called the transient response of the
system. It represents the component of the response when the driving function is made zero.
The response due to input and initial conditions considered simultaneously is called the total
response of the system.
For a stable system, the source free component always decays with time. In fact a
stable system is one whose source free component decays with time. For this reason the
source free component is also designated as the transient component and the component due
to source is called the steady state component.

ww When input is a unit impulse input, the response is called the impulse response of the
system and when the input is a unit step input, the response is called the step response of the
system.

w
EXAMPLE 3.61
.Ea A linear shift invariant system is described by the difference equation

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

syE
with y(–1) = 0 and y(–2) = –1.
4 8

ngi
Find (a) the natural response of the system (b) the forced response of the system for a
step input and (c) the frequency response of the system.
Solution:

nee
(a) The natural response is the response due to initial conditions only. So make
x(n) = 0. Then the difference equation becomes

y( n) 
3 1
y(n  1)  y(n  2) = 0 rin
Taking Z-transform on both sides, we have
4 8

g.n
Y (z ) 
3 1
4

È
1
[z Y ( z)  y(  1)] + [z 2 Y ( z)  z 1 y(  1) + y (  2)] = 0
8

3 1 1 2 Ø 1
e t
i.e. Y ( z ) É1  z  z Ù  =0
Ê 4 8 Ú 8

1/8 1/8z 2 1/8z 2

\ Y ( z) = = =
1  (3/4) z 1  (1/8) z 2 z 2  (3/4) z  (1/8) [ z  (1/2)] [ z  (1/4)]

The partial fraction expansion of Y(z)/z gives

Y (z ) (1/8) z A B 1/4 1/8
= = + = 
z [ z  (1/2)] [ z  (1/4)] z  (1/2) z  (1/4) z  (1/2) z  (1/4)

www.EasyEngineering.net
 www.EasyEngineering.net

250 h Digital Signal Processing

1 z 1 z
\ Y ( z) = 
4 z  (1/2) 8 z  (1/4)
Taking inverse Z-transform on both sides, we get the natural response as:
n n
1 1 11
y( n) =  2  u( n)  8  4  u(n)
4    
(b) To find the forced response due to a step input, put x(n) = u(n). So we have

3 1
y( n)  y(n  1) + y ( n  2) = u(n) + u(n  1)

ww 4 8
We know that the forced response is due to input alone. So for forced response, the

w initial conditions are neglected. Taking Z-transform on both sides of the above
equation and neglecting the initial conditions, we have

.Ea
Y (z ) 
3 1 1
z Y (z) + z 2Y (z ) = U ( z) + z 1U ( z) =
z
+
1
z 1 z 1

syE
4 8

 3 1  z+1
i.e. Y ( z )  1  z 1 + z 2  =
  z 1

ngi
4 8

z 1 z 2 ( z  1)
\ Y ( z)
( z  1) [1  (3/4) z 1  (1/ 8) z 2 )
z 2 ( z  1) nee ( z  1)[ z 2  (3/4) z  (1/8)]

( z  1) [ z  (1/2)] [ z  (1/4)]
rin
Taking partial fractions of Y(z)/z, we have
Y (z ) z( z  1) A

B

C g.n
\
z ( z  1) [ z  (1/2)] [ z  (1/4)]
16/3

6

5/3
z  1 z  (1/2) z  (1/4)
z 1 z  (1/2) z  (1/4)
e t
16 È z Ø Ë z Û 5Ë z Û
or Y ( z)  6Ì Ü
3 ÉÊ z  1ÙÚ Í z  (1/2) Ý 3 Í z  (1/4) ÜÝ
Ì

Taking the inverse Z-transform on both sides, we have the forced response for a
step input.
n n
16 È 1Ø 5 È 1Ø
y ( n) u(n)  6 É Ù u( n)  É Ù u( n)
3 Ê 2Ú 3 Ê 4Ú

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 251

(c) The frequency response of the system H(w) is obtained by putting z = e jw in H(z).
Y ( z) z ( z  1)
Here H ( z)
X (z ) z  (3/4) z  (1/8)
2

e jX (e jX  1)
Therefore, H (X )
(e jX )2  (3/4) e jX  (1/8)

EXAMPLE 3.62 (a) Determine the free response of the system described by the difference equation
5 1
y( n)  y( n  1)  y( n  2) x ( n) with y(–1) = 1 and y (–2) = 0

ww 6 6
(b) Determine the forced response for an input

w
Solution: .Ea
x ( n)
È 1Ø
É Ù
Ê 4Ú
n
u( n)

syE
(a) The free response, also called the natural response or transient response is the
response due to initial conditions only [i.e. make x(n) = 0].
So, the difference equation is:

y( n) 
5
6
1
ngi
y( n  1)  y(n  2)
6
0

Taking Z-transform on both sides, we get

nee
rin
5 1 1
Y ( z)  [ z Y ( z )  y( 1)]  [ z 2Y ( z)  z 1 y( 1)  y( 2)] 0
6 6
È
Y ( z ) É1 
Ê
5 1 1 2 Ø
6
z  z Ù
6 Ú

5
6
1 1
 z
6
0
g.n
\ Y ( z)
(5/6)  (1/6) z 1
1  (5/6) z 1
 (1/6) z 2
5/6[ z  (1/5)]z
z  (5/6) z  (1/6)
2 e
(5/6) z [ z  (1/5)]
[ z  (1 / 2)] [ z  (1/3)]
t
Taking partial fractions of Y(z)/z, we have
Y (z ) 5/6 [ z  (1/5)] A B 3/2 2/3
 
z [ z  (1/2)] [ z  (1/3)] z  (1/2) z  (1/3) z  (1/2) z  (1/3)
3 z 2 z
\ Y ( z) 
2 z  (1/2) 3 z  (1/3)
Taking inverse Z-transform on both sides, we get the free response of the system as:
n n
3 È 1Ø 2 È 1Ø
y ( n) É Ù u( n)  É Ù u( n)
2 Ê 2Ú 3 Ê 3Ú

www.EasyEngineering.net
 www.EasyEngineering.net

252 h Digital Signal Processing

(b) To determine the forced response, i.e. the steady state response, the initial conditions
are to be neglected.
The given difference equation is:
n
5 1 È1Ø
y( n)  y(n  1)  y( n  2) x ( n) É Ù u( n)
6 6 Ê 4Ú

Taking Z-transform on both sides and neglecting the initial conditions, we have
5 1 1 z
Y ( z)  z Y ( z )  z 2Y ( z)
6 6 z  (1/4)

ww i.e.,
È 5 1 Ø
Y ( z) É1  z 1  z 2 Ù
Ê 6 6 Ú
z
z  (1/4)

w \
.Ea
Y ( z)
z 1
z  (1/4) 1  (5/6) z 1  (1/6) z 2
z3
[ z  (1/4)] [ z  (1/2)] [ z  (1/3)]

Y (z )
syE
Partial fraction expansion of Y(z)/z gives

z2 A

B

C
z
3 8 6 ngi
[ z  (1/4)] [ z  (1/3)] [ z  (1/2)] z  (1/4) z  (1/3) z  (1/2)

 
z  (1/4) z  (1/3) z  (1/2)
nee
rin
Multiplying both sides by z, we get
z z z
8 6

g.n
Y ( z) 3
z  (1/4) z  (1/3) z  (1/2)
Taking inverse Z-transform on both sides, the forced response of the system is:

y ( n) 3É
È 1Ø
Ù
Ê 4Ú
n
È 1Ø
u( n)  8 É Ù
Ê 3Ú
n
u( n)  6 É
È1Ø
Ù
Ê 2Ú
n
u( n) e t
EXAMPLE 3.63 Find the impulse and step response of the system
y(n) 2 x (n)  3 x (n  1)  x (n  2)  4 x (n  3)

Solution: For impulse response, x(n) = d (n)

The impulse response of the system is:
y(n) 2E (n)  3E (n  1)  E (n  2)  4E (n  3)
For step response, x(n) = u(n)

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 253

The step response of the system is:

y(n) 2u(n)  3u(n  1)  u(n  2)  4u(n  3)

EXAMPLE 3.64 Solve the following difference equation

y(n)  2 y(n  1) x (n)
with x(n) = (1/3)n u(n) and the initial condition y(–1) = 1.
Solution: The solution of the difference equation considering the initial condition and input
simultaneously gives the total response of the system.
The given difference equation is:

ww y( n)  2 y(n  1) x ( n)
È 1Ø
É Ù
n
u( n) with y( 1) 1

w .Ea
Taking Z-transform on both sides, we get
Ê 3Ú

1
Y ( z)  2[ z 1Y ( z)  y( 1)]

syE
X ( z)
1  (1/3) z 1
Substituting the initial conditions, we have

Y ( z) (1  2 z 1 )
ngi
 2(1) 
1
1  (1/3) z 1

\ Y ( z)
2
1  2z 1
 nee
1
[1  (1/3) z ] [1  2 z 1 ]
1

2 z

z2 rin
z  2 [ z  (1/3)] ( z  2)

z2
g.n
Let Y1 ( z)
[ z  (1/3)] ( z  2)

Taking partial fractions of Y1(z)/z, we have

e t
Y1 ( z) z A B 1/7 6/7
 
z [ z  (1/3)] ( z  2) z  (1/3) z  2 z  (1/3) z  2
Multiplying both sides by z, we have
1 z 6 z
Y1 ( z ) 
7 z  (1/3) 7 z  2
2z 6 z 1 z 8 z 1 z
\ Y ( z)     
z  2 7 z  2 7 z  (1/3) 7 z  2 7 z  (1/3)

www.EasyEngineering.net
 www.EasyEngineering.net

254 h Digital Signal Processing

Taking inverse Z-transform on both sides, the solution of the difference equation is:
n
8 1 È 1Ø
y( n)  ( 2) n u( n)  É Ù u( n)
7 7 Ê 3Ú

EXAMPLE 3.65 Solve the following difference equation using unilateral Z-transform.
7 1
y ( n)  y(n  1)  y( n  2) x ( n) for n • 0
12 12
n
È 1Ø

ww
with initial conditions y( 1) 2, y( 2) 4 and x (n) É Ù
Ê 5Ú
u( n)

w
Solution: The solution of the difference equation gives the total response of the system
(i.e., the sum of the natural (free) response and the forced response)

.Ea
The given difference equation is:
n

syE
7 1 È 1Ø
y( n)  y( n  1)  y(n  2) x ( n) É Ù u( n)
12 12 Ê 5Ú

with initial conditions y(–1) = 2 and y(–2) = 4.

Taking Z-transform on both sides, we have
ngi
nee
7 1 1 1
Y ( z)  [ z Y ( z )  y( 1)]  [ z 2Y ( z)  z 1 y( 1)  y( 2)]
12 12 1  (1/5) z 1

i.e.
È
Y ( z) É1 
Ê
7 1 1 2 Ø
12
z  z Ù
12 Ú
7
12
1
12
1
(2)  (2 z 1 )  (4) 
12
1
1  (1/5) z 1 rin
i.e. È
Y ( z ) É1 
7 1 1 2 Ø
z  z Ù
5 È 1 1 Ø
É1  z Ù 
1 g.n
\ Y ( z)
Ê 12 12

(5/6) [1  (1/5) z 1 ]
[1  (7/12) z 1
Ú


6Ê 5 Ú 1  (1/5) z 1

1
 (1/12) z ] [1  (1/5) z ] [1  (7/12) z 1  (1/12)z 2 ]
2 1
e t
(5/6) [ z  (1/5)] z z3

[ z  (1/ 4)] [ z  (1/3)] [ z  (1/5)] [ z  (1/4)] [ z  (1/3)]
z [(11/6) z 2  (1/3) z  (1/30)]
[ z  (1/5)] [ z  (1/4)][ z  (1/3)]

Taking partial fractions of Y(z)/z, we have

Y (z ) A B C 6 1 1 1 100 1
   
z z  (1/5) z  (1/4) z  (1/3) 5 z  (1/5) 8 z  (1/4) 27 z  (1/3)

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 255

Multiplying both sides by z, we have

6 z 1 z 102 z
Y ( z)  
5 z  (1/5) 8 z  (1/4) 27 z  (1/3)
Taking inverse Z-transform on both sides, the solution of the difference equation is:
n n n
6 È 1Ø 1 È1Ø 102 È 1 Ø
y( n) É Ù u( n)  É Ù u( n)  É Ù u( n)
5 Ê 5Ú 8 Ê 4Ú 27 Ê 3 Ú

EXAMPLE 3.66 Using Z-transform determine the response of the LTI system described by

ww
y( n)  2r cos R y( n  1) + r 2 y(n  2) = x (n) to an excitation x(n) = anu(n).

Solution: Taking Z-transform on both sides of the difference equation, we have

w
i.e.
Y ( z)  2r cos R [z 1Y ( z) + y(1)] + r 2 [z 2Y ( z) + z 1 y(1) + y(  2)] = X ( z)

.Ea Y (z) [1  2r cos R z 1 + r 2 z 2 ] =

z
za

\ Y ( z) = syE z3
( z  a) ( z  re jR ) ( z  re  jR )

a2 z ngi r 2 e j 2R z
=
a  2ar cos R  r
2 2

r 2 e  j 2R z
+

nee jR  jR
z  a (re  a)(re  re ) z  re jR
jR

+
(re  jR  a)(re  re ) z  re jR
 jR jR

rin
\ y(n) =
a
a  2ar cos R + r
2
2

2
a n u(n) +
r

sin R 
n 1
 r sin (n + 1)R  a sin (n + 2)R 
a2  2ar cos R + r 2 g.n
 u(n)


EXAMPLE 3.67 Determine the step response of an LTI system whose impulse response
h(n) is given by h(n) = a  n u(  n); 0 < a < 1 .
e t
Solution: The impulse response is h(n) = a n u(  n); 0 < a < 1

1 1 1
\ H ( z) = = 
1  az a z  (1/a)
We have to find the step response

z
\ x (n) = u( n) and H ( z) =
z 1

www.EasyEngineering.net
 www.EasyEngineering.net

256 h Digital Signal Processing

The step response of the system is given by

y(n) = x (n)  h( n)

 1 z 1 1  z z 
\ Y (z ) = X ( z ) H ( z ) =    =   
 a  z  1 z  (1/a ) 1  a  z  1 z  (1/a ) 
So the step response is

1  
n
1
y ( n) =  u( n)    u( n) 
1  a  a 

ww
EXAMPLE 3.68 Determine the frequency response, magnitude response and phase
response for the system given by

w .Ea y( n) 
3
4
1
y( n  1)  y(n  2)
8
x ( n)  x ( n  1)

Solution:
syE
3
Given y( n) 
4
1
y( n  1)  y( n  2) x (n)  x (n  1)
8
Taking Z-transform on both sides and neglecting initial conditions, we have

Y ( z) 
3 1 1 ngi
z Y ( z ) + z 2Y (z ) = X (z )  z 1 X ( z)

i.e.
4 8

nee
Y (z) [1  (3/4) z 1 + (1/8) z 2 ] = X ( z)[1  z 1 ]

The transfer function of the system

rin
H ( z) =
Y (z )
=
1 z 1

X ( z) 1  (3/4)z 1 + (1/8)z 2
= 2
z( z  1)
z  (3/4)z + (1/8) g.n
The frequency response of the system

H (X ) =
e jX (e jX  1)
=
e j 2X  e jX
e t
e j 2X  (3/4) e jX + (1/8) e j 2X  (3/4) e jX + (1/8)

(cos 2X  cos X ) + j (sin 2X  sin X )

=
[cos 2X  (3/4) cos X + (1/8)] + j [sin 2X  (3/4) sin X ]
The magnitude response

(cos 2X  cos X )2 + (sin 2X  sin X )2

1/2
 
H (X ) =  2
[cos 2X  (3/4) cos X + (1/8)] + [sin 2X  (3/4) sin X ] 
2

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 257

The phase response

sin 2X  sin X sin 2X  (3/4) sin X
H (X ) = tan 1  tan 1
cos 2X  cos X cos 2X  (3/4) cos X + (1/8)

The magnitude response plot and phase response plot can be obtained by plotting H (X )
versus w plot and H (X ) versus w plots for various values of w.

EXAMPLE 3.69 A causal LTI system is defined by the difference equation

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

ww
Find the frequency response, magnitude response and phase response.

Solution: Given 2 y(n)  y( n  2) = x (n  1) + 3x ( n  2) + 2x ( n  3)

w
Taking Z-transform on both sides and neglecting initial conditions, we have

.Ea 2Y (z )  z 2Y ( z ) = z 1 X ( z ) + 3z 2 X ( z) + 2z 3 X ( z )

Y (z) [2  z 2 ] = X ( z )[z 1 + 3z 2 + 2z 3 ]
i.e.

syE
The transfer function of the system

Y (z )
X ( z)
= H ( z) =
2  z 2 ngi
z 1 + 3z 2 + 2z 3
=
[z 2 + 3z + 2]
[2z 3  z ]
The frequency response of the system
nee
e j 2X + 3e jX + 2
H (X ) =
2e j 3X  e jX
rin
The magnitude response of the system

(cos 2X + 3 cos X + 2) + j (sin 2X + 3sin X )

g.n
H (X ) =
(2cos 3X  cos X ) + j (2 sin 3X  sin X )

 (cos 2X + 3cos X + 2) 2 + (sin 2X + 3 sin X ) 2 

e
1/2
t
=  
 (2 cos 3X  cos X ) + (2 sin 3X  sin X )
2 2

Phase response
sin 2X + 3sin X 2sin 3X  sin X
H (X ) = tan 1  tan 1
cos 2X + 3cos X  2 2cos 3X  cos X

EXAMPLE 3.70 Determine the steady state response for the system with impulse response
n
1
h( n) =   u(n) for an input x(n) = [cos(Q n)] u(n) .
2

www.EasyEngineering.net
 www.EasyEngineering.net

258 h Digital Signal Processing

Solution: Let y(n) be the steady state (forced) response of the system which is given by the
convolution of x(n) and h(n).
Then y(n) = x(n) * h(n)
By the convolution property of Z-transforms we get

Y ( z) = X ( z ) H ( z )
\ y(n) = Z 1 [Y ( z )] = Z 1 [X ( z) H ( z )]

Given x(n) = cos(Q n) u(n)

z( z  cos Q )

ww
z( z + 1) z
\ X ( z) = = 2 =
z  2 z cos Q + 1 z + 2 z + 1 (z + 1)
2

w
n
1
Given h(n) =   u (n)

\ .Ea 2
z

syE
H (z ) =
z  (1/2)

z z z z
= (0.8  j 0.4)

ngi
Y (z ) = X ( z ) H ( z ) = + (0.2 + j 0.4)
z + 1 z  j 0.5 z +1 z  j 0.5

Taking inverse Z-transform on both sides, we have

nee
y(n) = (0.8  j 0.4) (1) n u( n) + (0.2 + j 0.4) ( j 0.5) n u( n)

EXAMPLE 3.71 Find the response of the time-invariant system with impulse response
h(n) = {1, 2, 1, –1} to an input signal x(n) = {1, 2, 3, 6}. rin
Solution: Let y(n) be the response or output of the system. We know that the response is
g.n
given by the convolution of input x(n) and impulse response h(n).

\
i.e. y(n) = x(n) * h(n)
Y(z) = X(z) H(z)
e t
Given x (n) = {1, 2, 3, 6} = E ( n) + 2E (n  1) + 3E ( n  2) + 6 E ( n  3)
\ X ( z ) = 1 + 2z 1 + 3z 2 + 6z 3
Given h( n) = {1, 2, 1, 1} = E ( n) + 2E (n  1) + E ( n  2)  E (n  3)

\ H (z ) = 1 + 2z 1 + z 2  z 3
\ Y ( z) = X ( z ) H ( z) = (1 + 2z 1 + 3z 2 + 6z 3 ) (1 + 2z 1 + z 2  z 3 )

= 1 + 4z 1 + 8z 2 + 8z 3 + 3z 4  2z 5  z 6

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 259

Taking inverse Z-transform on both sides, we have

y(n) = E ( n) + 4E ( n  1) + 8E ( n  2) + 8E (n  3) + 3E (n  4)  2E (n  5)  E (n  6)

i.e. y(n) = {1, 4, 8, 8, 3,  2,  1}

EXAMPLE 3.72 The step response of an LTI system is

n 2
1
s ( n) =   u ( n + 2)
3

ww
Find the impulse response of the system.
Solution: We have s(n) = h(n) * u(n)

w \

.Ea
S(z ) = H ( z ) U (z) = H ( z )

1
n 2
z
z 1

Given

syE 
s ( n) =  
3

n 2
u ( n + 2)

ngi
n
1 1 
S(z ) =   3  u(n + 2) z  n = 32   3z 
n   n2

1 
 
= 32   =
2

34 z 2 81z 3
nee
rin
3z
=
1
1 1
1  z 1  1
z  

g.n
3z 3  3
The system function H(z) is

H (z ) = S ( z )
z 1
z
=

81 z 3 z  1

z  

1 z
3
e t
81 z 2 (z  1) 81 z 3 81 z 2
= = 
 1
z
1
z
1
z  
 3 3 3

z z
= 81 z 2  81 z
1 1
z z
3 3

www.EasyEngineering.net
 www.EasyEngineering.net

260 h Digital Signal Processing

The impulse response of the system is

n 2 n 1
1 1
h( n) = 81  u(n + 2)  81   u(n + 1)
3 3
n n
1 1
= 9   u (n + 2)  27   u (n + 1)
3 3

EXAMPLE 3.73 Consider a causal linear shift-invariant system with system function

1  a1 z 1

ww H (z ) =
1  az 1
where a is real. Determine the range of values of a for which the system is stable? Show

w
analytically that this system is an all pass system.

.Ea 1  a 1 z 1
z
1

syE
Solution: Given H (z ) = = a
1
1  az za
The system function H(z) has one pole at z = a and a zero at z = 1/a. The ROC of H(z) is

ngi
exterior to the circle of radius a. For a stable filter all poles must be inside the unit circle.
Therefore for the system to be stable z > a < 1 .
For a < 1 , the frequency response is

H (X ) =
1  a 1 z 1 nee =
1  a1 e jX
1  az 1 z  e jX
1  ae  jX
rin
=
a  e jX
a (1  ae jX )
=
a  (cos X  j sin X )
a (1  a cos X + ja sin X ) g.n
=
a  cos X + jsin X
a(1  a cos X + ja sin X ) e t
( a  cos X ) + sin X2 2
H (X )
2
=
a {(1  a cos X )2 + a2 sin 2X}
2

a2  2 a cos X + cos2X + sin 2 X

=
a2 (1  2a cos X + a2 cos2 X + a2 sin 2 X )

1 + a2  2 a cos X 1
= = 2
a (1 + a  2 a cos X )
2 2
a

That is H (X ) =
1
for all values of w. Therefore the system is an all pass system.
a

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 261

3.10 DECONVOLUTION USING Z-TRANSFORM

Deconvolution is used to find the input x(n) applied to the system once the output y(n) and
the impulse response h(n) of the system are known. The Z-transform also can be used for
deconvolution operation.
We know that
Y ( z)
Y (z) = X ( z) H ( z ) or X (z ) =
H (z )

where Y(z), X(z) and H(z) are the Z-transforms of output, input and impulse response,
respectively. If y(n) and h(n) are given, we can determine their Z-transforms Y(z) and H(z).

ww
Knowing Y(z) and H(z), we can determine X(z) and knowing X(z), we can determine the
input x(n). Thus, the deconvolution is reduced to the procedure of evaluating an inverse

w
Z-transform.

.Ea
EXAMPLE 3.74 Find the input x(n) of the system, if the impulse response h(n) and the
output y(n) are as given below:

Solution: Given syE

h( n) = {2, 1, 0,  1, 3}; y(n) = {2,  5, 1, 1, 6,  11, 6}

h(n) = {2, 1, 0,  1, 3}

\ H ( z) = 

h(n)z  n ngi
n

 h ( n ) z  n = 2 + z 1
nee
 z 3 + 3z 4
=
n0
rin
y(n) = {2,  5, 1, 1, 6,  11, 6}
g.n
\ Y (z ) = 


n

6
y( n) z  n
e t
=  y ( n) z  n
n0

= 2  5z 1 + z 2 + z 3 + 6z 4  11z 5 + 6 z 6

Y ( z)
\ X (z ) =
X (z )
2  5z 1 + z 2 + z 3 + 6z 4  11z 5 + 6z 6
=
2 + z 1  z 3 + 3 z 4

www.EasyEngineering.net
 www.EasyEngineering.net

262 h Digital Signal Processing

1  3z 1  2 z 2
2  z 1  z 3  3z 4 2  5z 1  z 2  z 3  6 z 4  11z 5  6z 6
2  z 1  z 3  3z 4
 6 z 1  z 2  2z 3  3z 4  11z 5  6 z 6
 6 z 1  3z 2  3z 4  9z 5
4z 2  2 z 3  2 z 5  6 z 6

ww 4z 2  2 z 3  2 z 5  6 z 6
0

\
w X(z) = 1 – 3z–1 + 2z–2

.Ea
Taking inverse Z-transform, we have input

i.e. syE
x (n) = E (n)  3 E ( n  1) + 2 E (n  2)
x (n) = {1,  3, 2}

3.11 RELATION BETWEEN s-PLANE AND z-PLANE ngi

nee
Let x(t) be a continuous signal and x*(t) be the sampled version of x(t). Let the sampling
period be T.

\ x * (t ) =


 x (nT ) E (t  nT ) rin
The Laplace transform of x*(t) can be written as
n =0

g.n
X * ( s) =


 x(nT ) enTs
n 0
e t
The Z-transform of x*(t) is

X (z ) =  x(nT ) z n
n0

Therefore, the relation between s-plane and z-plane can be described by the equation
z = esT
jw
Let z = re and s = s + jw
\ z = re jX = e(T + jX )T = eT T e jX T

www.EasyEngineering.net
 www.EasyEngineering.net

Z-Transforms h 263

\ z = r = eT T and z = X T

If s
= 0, |z| = 1. For s < 0, |z| < 1 and for s > 0, |z| > 1
When s = 0, |z| = 1. So the jw axis of s-plane maps into the unit circle. The left half of
the s-plane where s < 0 maps into the inside of the unit circle (|z| < 1). The right half of the
s-plane where s > 0 maps into the outside of the unit circle (|z| > 1). The mapping is as
shown in Figure 3.12.

ww
w .Ea
syE
Figure 3.12 Relation between s-plane and z-plane.

ngi
From Figure 3.12 we can observe that a single horizontal strip of width 2p/T, from
w = –p/T to w = p/T completely maps into the inside of the unit circle and the horizontal

nee
strips between p/T and 3p/T between –p/T and –3p/T (In general between (2n–1)p/T and
(2n+1)p/T where n = 0, ±1, ±2, ...) are mapped again into the inside of the unit circle. Thus,
many points in the s-plane are mapped into a single point in the z-plane, causing aliasing

rin
effect. That is when we sample two sinusoidal signals of the frequencies which differ by a
multiple of the sampling frequency, we cannot distinguish between the results.

SHORT QUESTIONS WITH ANSWERS g.n

1. How are discrete-time systems analysed using Z-transforms?
e
Ans. Discrete-time systems are described by difference equations. The difference
equations, which are in time domain, are converted into algebraic equations in
t
z-domain using Z-transforms. Those algebraic equations are manipulated and result
is obtained in z-domain. Using inverse Z-transform, the result is converted into
time domain.
2. Define Z-transform.
Ans. The Z-transform of a bilateral discrete-time signal x(n) is defined as:
‡
X ( z)
n ‡
Çx (n) z  n

The Z-transform of a unilateral discrete-time signal x(n) is defined as:

www.EasyEngineering.net