Signals, Spectra and Signal Processing

THE
-TRANSFORM AND ITS APPLICATION TO THE ANALYSIS OF LTI SYSTEMS

3.0. Introduction

systems. This chapter introduces the z-transform, develop its properties and demonstrate its importance in
Transform techniques are an important tool in the analysis of signals and linear time-invariant (LTI)

the analysis and characterization of linear time-invariant systems.

The z-transform plays the same role in the analysis of discrete-time signals and LTI systems as the

z-transform can be applied to simplify the analysis of LTI systems, and the idea of poles and zeros provide
Laplace transform does in the analysis of continuous-time signals and LTI systems. Many properties of the

another perspective in characterizing the response of the system.

3.1. The z-Transform

The direct
-transform. The direct z-transform of a discrete-time signal xn is defined as the power
series

X z  xnz (3.1)

wherez is a complex variable ( re ). The direct z-transform transforms the time-domain signal xn into
its complex-plane representation Xz. The inverse procedure, that is, obtaining xnfrom Xz, is called
the inverse z-transform. The z-transform of a signal xn is denoted by

Xz  Zxn (3.2)

whereas the relationship between xn and Xz is indicated by


xn  Xz (3.3)

Since the z-transform is an infinite power series, it exists only for those values of z for which the series
converges. The region of convergence (ROC) of Xz is the set of all values of z for which Xz attains a
finite value. Thus, anytime that the z-transform of a function is cited, its ROC must also be cited.

Determine the z-transform of the following finite-duration signals.

Example 3.1

(a) x n  1 2 5 7 0 1

(b) x n  1 2 5

7 0 1

The z-Transform and Its Application to the Analysis of LTI Systems Page 1
Signals, Spectra and Signal Processing

(c) x! n  0 0 1 2 5 7 0 1

(d) x" n  δn

(e) x$ n  δn % k, k ' 0

(f) x( n  δn ) k, k ' 0

(a) X z  1 ) 2z ) 5z ) 7z )z ; ROC: the entire z-plane except z  0.

Answer:
  ! $

(b) X z  z  ) 2z ) 5 ) 7z 
) z ! ; ROC: the entire z-plane except z  0 and z  ∞.

(c) X! z  z 
) 2z !
) 5z "
) 7z $
) z + ; ROC: the entire z-plane except z  0.

(d) X" z  1; ROC: the entire z-plane.

(e) X$ z  z ,
; ROC: the entire z-plane the entire z-plane except z  0.

(f) X( z  z , ; ROC: the entire z-plane the entire z-plane except z  ∞.

Determine the closed-form expression of the z-transform of the signal

Example 3.2

1
xn  - . un
2

X z  ;ROC: |z| ' 

Answer:
 
0 20
 
1

The complex variable z expressed in polar form is z  re , where r  |z|(magnitude of z) and θ  5z
(angle or argument of z). Then Equation 3.1 can be expressed as

6Xz|  xnr e 
 789: (3.4)

In the ROC of Xz, |Xz| ; ∞. But,

|Xz|  < xnr e  <= ?xnr e  ? |xnr | (3.5)

The z-Transform and Its Application to the Analysis of LTI Systems Page 2
Signals, Spectra and Signal Processing

Hence, |Xz| is finite if the sequence xnr is absolutely summable. Thus, the problem of finding the
ROC is equivalent to determining the range of values of r for which the sequence xnr is absolutely
summable. To elaborate, express Equation 3.5 as

x  n
|Xz| = |xnr |) @ @
r
A
xn
|Xz| = |x%nr | ) @ @
r
(3.6)
 A

If Xz converges in some region of the complex plane, both summations of Equation 3.6 must be finite in
that region. If the first sum in 3.6 converges, there must exist values of r small enough such that the
product sequence x%nr , 1 = n = ∞ is absolutely summable. Therefore, the ROC for the first sum
consists of all points in a circle of some radius r , where r ; ∞ as illustrated below.

Figure 3.1. Region of convergence for anti-causal signal.

On the other hand, if the second sum of 3.6 converges, there must exist values of r large enough such that
the product sequence xn/r , 0 = n = ∞, is absolutely summable. Hence the ROC for the second sum
of 3.6 consists of all points outside a circle of radius r ' r as illustrated below in Figure 3.2.

For Xz to converge, both sums must be finite. Hence the ROC is generally specified as the annular
region in the z-plane, r ; r ; r which is the common region where both sums are finite. This region is
illustrated in Figure 3.3. On the other hand, if r ' r , there is no common region of convergence for the
two sums and hence Xz does not exist.

The z-Transform and Its Application to the Analysis of LTI Systems Page 3
Signals, Spectra and Signal Processing

Figure 3.2. Region of convergence for causal signals.

Figure 3.3. Region of convergence of a two-sided signal if both the sums of Equation 3.6 converges and CD ; C ; CE

Determine the z-transform of the signals

Example 3.3

(a) x n  a un

(b) x n  %a u%n % 1

(a) X z  ; ROC: |z| ' |a|

Answers:

 G20

(b) X z   ; ROC: |z| ; |a|


G20

The z-Transform and Its Application to the Analysis of LTI Systems Page 4
Signals, Spectra and Signal Processing

Figure 3.4. The exponential HE I  JI KI signal and its ROC.

Figure 3.5.The exponential signal HD I  %JI K%I % E and its ROC.

The previous example raises two very important issues with regards to the z-transform:

The closed-form expression of the z-transform alone does not uniquely determine the signal.
Specifying the ROC will remove the ambiguity. A discrete-time signal xn is uniquely determined
•

by its z-transform Xz and the region of convergence of Xz.

The ROC of a causal signal is the exterior of a circle of some radius r while the ROC of an anti-
causal signal is the interior of a circle of some radius r .
•

Determine the z-transform of the signal

Example 3.4

xn  a un ) b u%n % 1

If |b| ; |a|, Xz does not exist because the ROCs of the first and second terms of the signal do not
Answer:

overlap. If |b| ' |a|, Xz exists and is equal to

b%a
X z 
a ) b % z % abz 

with ROC of Xz is |a| ; |z| ; |b|.

The z-Transform and Its Application to the Analysis of LTI Systems Page 5
Signals, Spectra and Signal Processing

Figure 3.6. ROCs of the M-transform for Example 3.4.

To summarize, the ROC of the z-transform of a signal depends whether the signal has a finite or infinite
duration, and on whether it is causal, anti-causal or two-sided. These are summarized in Table 3.1 below.

It should be noted that the z-transform defined by 3.1 is referred to as the two-sided or bilateral z-
transform. The one-sided or unilateral z-transform is given by

X N z  x  n z (3.7)
A

Note that when the signal xn is causal, the two-sided and the one-sided z-transform are equal. In any
other case, they are different.

The z-Transform and Its Application to the Analysis of LTI Systems Page 6
Signals, Spectra and Signal Processing

The inverse
-transform. The procedure for transforming from the z-domain to the time-domain is called
the inverse z-transform. An inversion formula for obtaining xn from Xz can be derived using the
Cauchy integral theorem, which is an important theorem in the theory of complex variables.

To find xnfrom Xz, it can be shown that

1
xn  Q Xzz 
dz
j2π C
(3.8)

whereC is any contour that encloses the origin within the ROC of Xz and taken in counterclockwise

evaluation of inverse z-transform. A lookup table will be developed where the forward and inverse z-
direction. Although this integral provides the desired inversion formula, this shall not be used directly in the

transform will be evaluated.

The z-Transform and Its Application to the Analysis of LTI Systems Page 7
Signals, Spectra and Signal Processing

The z-Transform and Its Application to the Analysis of LTI Systems Page 8
Signals, Spectra and Signal Processing

3.2. Properties of z-Transform

 
Linearity. If x n  X zand x n  X z, then

xn  a x n ) a x n  Xz  a X z ) a X z (3.9)

or the z-transform of the linear combination of xn is the linear combination of their transforms.

The z-Transform and Its Application to the Analysis of LTI Systems Page 9
Signals, Spectra and Signal Processing

Determine the z-transform and the ROC of the signal

Example 3.5

xn  32 % 43 un

3 4
Answer:
X z  %
1 % 2z  1 % 3z 
with ROC |z| ' 3.

Determine the z-transform of the signals

Example 3.6

(a) xn  cos ωA nun

(b) xn  sin ωA nun

Answers:
(a) Xz  ; ROC: |z| ' 1
 20 \]^ _`
 20 \]^ _` N21

(b) Xz   ; ROC: |z| ' 1

20 ^a _`
20 \]^ _` N21


Time shifting.Ifxn  Xz, then

xn % k  z ,
X z (3.10)

or when shifting the signal xn by k number of samples, its z-transform is multiplied by the factor z ,

The ROC of z , Xz is the same as that of Xz except for z  0 if k ' 0 and z  ∞if k ; 0.
.

If x n  1 2 5 7 0 1, using its z-transform, determine the z-transform of

Example 3.7

(a) x  x n ) 2

(b) x!  x n % 2

(a) X z  z  ) 2z ) 5 ) 7z ) z ! ; ROC: the entire z-plane except z  0 and z  ∞.

Answers:


(b) X! z  z 
) 2z !
) 5z "
) 7z $
) z + ; ROC: the entire z-plane except z  0.

The z-Transform and Its Application to the Analysis of LTI Systems Page 10
Signals, Spectra and Signal Processing

Example 3.8
Determine the transform of the signal

Answer:


Scaling in the
-domain.If xn  Xz, ROC: r ; |z| ; r , then

a xn  Xa  z; ROC: |a|r ; |z| ; |a|r (3.11)

for any constant a, real or complex. To better understand the meaning and implications of the scaling
property, let a  rA e_` and z  re_ and let w  a  z . Thus, Zxn  Xz and
Za xn  Xw. It can be seen that

1
w  a z  - r. e_ _` 
rA
(3.12)

Thus, when a signal xn is multiplied by an exponential (real or complex) this causes the signal to be
scaled in z -domain, that is it shrinks when |a|  rA ' 1 or expands when |a|  rA ; 1 with a
corresponding rotation (if ωA c 2kπ) of the z-plane. This is illustrated in Figure 3.7.

Figure 3.7. Scaling in
-domain.

The z-Transform and Its Application to the Analysis of LTI Systems Page 11
Signals, Spectra and Signal Processing

Determine the z-transforms of the signals

Example 3.9

(a) xn  a cos ωA nun

(b) xn  a sin ωA nun

Answers:
(a) Xz  ; ROC: |z| ' |a|
 G20 \]^ _`
 G20 \]^ _` NG1 21

(b) Xz   ROC: |z| ' |a|

G20 ^a _`
G20 \]^ _` NG1 21
;


Time reversal.Ifxn  Xz, ROC r ; |z| ; r , then

x%n  Xz 
; ROC 7 ; |z| ; 7
 
(3.13)
1 0

Determine the z-transform of the signal

Example 3.10

xn  u%n

Xz   ;ROC |z| ; 1

Answer:



Differentiation in the
-domain.If xn  Xz, then

 dXz
nxn  %z
dz
(3.14)

Note that both transforms have the same ROC.

Determine the z-transform of the signal

Example 3.11

xn  na un

Answer:
Xz   ; ROC |z| ' |a|
G20
G201

The z-Transform and Its Application to the Analysis of LTI Systems Page 12
Signals, Spectra and Signal Processing

Determine the signal xn whose z-transform is given by

Example 3.12

Xz  log1 ) az 

and ROC |z| ' |a|

a
Answer:
xn  %1 N
u n % 1
n

 
Convolution of two sequences.Ifx n  X zand x n  X z, then

xn  x n f x n  Xz  X zX z (3.15)

The ROC of Xz is at least the intersection that for X zand X z. Note that convolving signals in time-
domain causes their z-transforms to be multiplied.

Compute the convolution xn of the signals

Example 3.13

Answer:

The convolution property is one of the most powerful properties of the z-transform because it converts the
convolution of two signals (time-domain) to multiplication of their transforms using the following steps:

1. Compute the z-transform of the signals to be convolved.

2. Multiply the two transforms.
3. Find the inverse of the product in (2).

This procedure is, in many cases, computationally easier than the direct evaluation of the convolution
summation.

Initial value theorem. If xn is causal, then

x0  lim Xz

h
(3.16)

The z-Transform and Its Application to the Analysis of LTI Systems Page 13
Signals, Spectra and Signal Processing

1. Determine the z-transform of the following signals:

Drill Problems 3.1

2. Determine the z-transform of the following signals.

3. Determine the z-transform of the following signals and sketch the ROC .

4. Determine the z-transform of the following signals

5. Express the z-transform of

y  n  xk
,
in terms of Xz.

The z-Transform and Its Application to the Analysis of LTI Systems Page 14
Signals, Spectra and Signal Processing

6. Compute the convolution of the following signals by means of the z-transform.

3.3. Rational z-Transforms

An important family of z-transforms are those for which Xz is a rational function, that is, a ratio of two
polynomials in z  (or z).

Poles and Zeros. The zeros of a z-transform Xz are the values of z for which Xz  0. The poles of a
z-transform are the values of z for which Xz  ∞. If Xz is a rational function, then

Nz bA ) b z 
) l ) bM z M ∑M
, A b, z
,
X z    N
Dz aA ) a z  )l) a z N ∑, A a, z ,
(3.17)
N

If aA c 0 and bA c 0, negative powers of z can be avoided by factoring out the terms bA z M

and aA z N
,
thus,

z M ) pq0 r z M ) l ) p qMr
q  q
Nz bA z M
X z   ` `
D z a A z N
zN ) pG
G0
r zN  ) l ) pG r
GN
(3.18)
` `

Since Nz and Dz are polynomials in z, they can be expressed in factored form as

Nz bA z % z z % z  l z % zM 
X z   z MNN
D z a A z % p z % p  l z % pN 

∏M, z % z, 
Xz  Gz
(3.19)
N M
∏ ,  z % p, 
N

where G  called the system gain. Thus, Xz has M finite zeros at z  z , z , l , zM (the roots of the
q`
G`
numerator polynomial), N finite poles at z  p , p , l , pN (the roots of the denominator polynomial) and
|N % M| zeros (if N ' z) or poles (if N ; z) at the origin z  0. Poles and zeros may also occur at
z  ∞. A zero exists at z  ∞ if X∞  0 and a pole exists at z  ∞ if X∞  ∞. If we count the
poles and zeros at zero and infinity, there will be exactly an equal number of poles as zeros.

The z-Transform and Its Application to the Analysis of LTI Systems Page 15
Signals, Spectra and Signal Processing

Xz can be represented graphically by a pole-zero plot (or pattern) in the complex plane, which shows the
location of poles by crosses ({ and the location of zeros by circles (|). The multiplicity of multiple-order
poles or zeros is indicated by a number close to the corresponding cross or circle.

Example 3.14

xn  a un
Determine the pole zero plot for the signal

for a ' 0.

Answer:

Example 3.15
Determine the pole-zero plot for the signal

Answer:

The z-Transform and Its Application to the Analysis of LTI Systems Page 16
Signals, Spectra and Signal Processing

Determine the z-transform and the signal that corresponds to the to the pole-zero plot below:
Example 3.16

1 % rz  cos ω]
Answer:
X z  G
1 % 2rz  cos ω] ) r  z 

with ROC |z| ' r. The signal corresponding to this is

xn  Gr cos ω] n un

As was shown before, the z-transform Xz is a complex function of the complex variable z  Rez )
j Imz. Its magnitude, |Xz| is a real and positive function of z. Since, z represents a point in the
complex plane, |Xz| is a two-dimensional function and describes a surface. For the transform

z %z 
X z 
1 ) 1.2732z  ) 0.81z 

the plot of its magnitude |Xz| is shown below.

Figure 3.8. Plot of the magnitude of 
  ENE.D‚ƒD
2E N„.

2E
2D
E
2D

The z-Transform and Its Application to the Analysis of LTI Systems Page 17
Signals, Spectra and Signal Processing

and for z  e_ ,

Figure 3.9. Plot of 
 in the previous Figure with
 †‡ˆ.

signals depends whether the poles of the transform are contained in the region |z| ; 1, or in the region
Pole Location and Time-Domain Behavior for Causal Signals. The characteristic behavior of causal

|z| ' 1 or on the circle |z|  1. Since the circle has a radius of 1, it is called the unit circle.

If a real signal has a z-transform with one pole, this pole has to be real. The only such signal is the real
exponential

 1
xn  a un  Xz 
1 % az 

whose ROC is |z| ' |a| having one zero at z  0 and one pole at p  a on the real axis. Figure 3.10
summarizes the time-domain behavior of a single real pole causal signal as a function of the pole location
with respect to the unit circle.

The z-Transform and Its Application to the Analysis of LTI Systems Page 18
Signals, Spectra and Signal Processing

Figure 3.10. Time-domain behavior of signals with single real pole.

A causal real signal with a double real pole has the form

xn  na un

and its behavior is summarized in figure 3.11.

Figure 3.11. Time-domain behavior of causal signals with double real pole.

The z-Transform and Its Application to the Analysis of LTI Systems Page 19
Signals, Spectra and Signal Processing

A causal signal with a complex conjugate pair of poles results in an exponentially weighted sinusoidal
signal, as shown in the Figure 3.12.

Figure 3.12. Time-domain behavior of signals with complex conjugate poles.

Two pairs of complex conjugate poles on the unit circle results in the following time-domain behavior.

Figure 3.13. Time-domain behavior of a causal signal with two pairs of complex cojugate poles on the unit circle.

The z-Transform and Its Application to the Analysis of LTI Systems Page 20
Signals, Spectra and Signal Processing

From the foregoing discussions, it can be observed that:

• A causal signal decays if its poles (whether single or double real, or complex conjugates) is inside
the unit circle, has a bounded value when the poles are on the unit circle except when there are
double poles in the unit circle, and grows when its poles are outside the unit circle.
• When the poles on the unit circle have multiplicity greater than one, the resulting signal is one that
grows over time.
• For the complex conjugate poles, the distance of the poles from the origin dictates the envelope of
the sinusoidal signal and their angle with the real positive axis dictates the relative frequency of the
signal.
• A signal with a pole or complex conjugate poles near the origin decays more rapidly than the one
associated with a pole near and inside the unit circle.

It should be remembered that everything that has been said about causal signals applies as well to causal
LTI systems since their impulse response is a causal signal. Hence if a pole of a system is outside the unit
circle, the impulse response of the system becomes unbounded and consequently, the system is unstable.

System Function of an LTI System. The output of a relaxed, LTI system to an input sequence xn can
be obtained by computing the convolution sum of xn and the unit sample response of the system. The
convolution property allowed the expression of this relationship in z-domain as

Yz  HzXz (3.20)

where Yz is the z-transform of the output sequence yn, Xz is the z-transform of the input sequence
xn and Hz is the z-transform of the unit sample response hn. Thus, if hn is the time-domain
characterization of the system, its z-domain characterization is Hz which is called the system function or
the transfer function.

If the system is described by a difference equation

N M

yn  % a, yn % k ) b, xn % k (3.21)

,  , A

the system function can be determined by taking the z -transform of 3.21 and solving for the ratio
Hz  X, thus
Y

Y z ∑M
k0 bk z
,
Hz  
Xz 1 ) ∑Nk1 ak z ,
(3.22)

Thus, an LTI system described by a constant coefficient difference equation has a rational system function.

The z-Transform and Its Application to the Analysis of LTI Systems Page 21
Signals, Spectra and Signal Processing

From 3.22, if a,  0 for 1 = k = N, this will be reduced to

M
1
Hz  M b, z M ,
z
(3.23)
, A

which has M zeros and M poles at the origin. This results to an impulse response that is finite in duration.

from 3.22, if b,  0 for 1 = k = M, the system function reduces to

A systemdescribed by this system function is called all-zero system or an FIR system. On the other hand,

bA z N
Hz 
∑N
, A a, z
N ,
(3.24)

with aA Ž 1. This reduces to an impulse response that is of infinite length because of the presence of non-
zero poles. A system described by this system function is called an all-pole system or an IIR system.

Example 3.17
Determine the system function and the unit sample response of the system described by the difference

1
equation
yn  yn % 1 ) 2xn
2

Answers:

2
The system function is
Hz 
1 % z
 

1
with the impulse response as
h  n  2 - . u  n
2

The z-Transform and Its Application to the Analysis of LTI Systems Page 22
Signals, Spectra and Signal Processing

Determine the z-transform of the following signals and sketch the pole-zero patterns.
Drill Problems 3.2

1. xn  1 ) nun

2. xn  a ) a un, a is real

3. xn  Ar cosω] n %  un, 0 ; r ; 1


4. xn   n ) n p!r u n % 1
 

5. xn  pr un % un % 10



3.4. Inversion of z-Transform

There are three methods that are often used for the evaluation of the inverse z-transform:

Direct evaluation by contour integration.

Expansion into a series of terms, in the variables z and z
•

• .
• Partial fraction expansion and table look-up.

In this section, the last two methods will be discussed.

Inverse
-transform by Power Series Expansion. The basic idea is this: Given a z-transform Xz with
its corresponding ROC, the function Xz can be expanded into a power series of the form

X z  c z (3.25)

which converges in the given ROC. Then, by uniqueness of the z-transform, xn  c for all n. When
Xz is rational, the expansion can be performed by long division.

Determine the inverse z-transform of

Example 3.18

1
X z 
1 % 1.5z  ) 0.5z 

when the ROC is (a) |z| ' 1 and (b) |z| ; 0.5.

Answers:

3 7 15 31
(a)
xn  ‘1, , , , , l “
2 4 8 16

The z-Transform and Its Application to the Analysis of LTI Systems Page 23
Signals, Spectra and Signal Processing

(b)
xn  ”l 62, 30, 14, 6, 2, 0, 0•


Note that long division will not provide a closed-form solution of the inverse z-transform except if the
resulting pattern is simple enough to infer the general term xn. Hence this method is used only if one
wished to determine the values of the first few samples of the signal.

Determine the inverse z-transform of

Example 3.19

Xz  log1 ) az 

using power series expansion.

a
Answer:
xn  %1 N
u n % 1
n

Expansion of irrational functions into power series can be obtained from tables.

Inverse
-transform by Partial Fraction Expansion. In the table lookup method, the function Xz will be
first expressed as a linear combination

Xz  a X z ) a X z ) a! X! z ) l ) a, X, z (3.26)

If such decomposition is possible, then xn can be found using the linearity property as

xn  a x n ) a x n ) a! x! n ) l ) a, x, n (3.27)

This approach is particularly useful if Xz is a rational function. Assume that aA Ž 1, then the general
form of the rational z-transform as

Nz bA ) b z 
) l ) bM z M
X z  
Dz 1 ) a z  ) l)a z N
(3.28)
N

A rational function of the form 3.28 is called proper if aN c 0 and M ; N. Otherwise, the function is called
improper (when M – N). An improper rational function can always be expressed as a sum of a polynomial
and a proper rational function.

The z-Transform and Its Application to the Analysis of LTI Systems Page 24
Signals, Spectra and Signal Processing

Example 3.20

1 ) 3z ) z ) z
Express the improper rational function
    !
X z  ( !
1) z ) z
$   
( (

z
Answer:
 
Xz  1 ) 2z 
) (
1) z ) z
$   
( (

By inspection, the inverse of the polynomial can be found easily. Thus, the expansion of the proper rational
function to its partial fraction components will be given emphasis. For the proper rational function

Nz bA ) b z 
) l ) bM z M
X z  
Dz 1 ) a z  ) l)a z N
(3.28)
N

where aA Ž 1 and M ; —, negative powers of z must be eliminated first by multiplying thru numerator
and denominators by z N , making 3.28 as

bA z N ) b z N  ) l ) bM z N M
X z 
z N ) a z N  ) l ) aN
(3.29)

and then dividing both sides by z to make 3.29 a proper rational function again, thus

Xz bA z N  ) b z N  ) l ) bM z N M 

z z N ) a z N  ) l ) aN
(3.30)

Factor the denominator of 3.30 into linear factors. The factors will contain the poles p , p , … pN of Xz.
Two cases will be specified.

Distinct poles. Suppose that the poles p , p , … pN are all different. Then the expansion will be of the
form

X z A A A,
 ) ) l)
z z % p z % p z % p,
(3.31)

The coefficients A , A , …, A, can be determined using the formula

The z-Transform and Its Application to the Analysis of LTI Systems Page 25
Signals, Spectra and Signal Processing

z % p, X z
A,  6 @
z  ˜™
(3.32)

for k  1, 2, 3, … , N. Of course, the old trick will still work, but this method is better when evaluating partial
fraction expansion of functions with complex poles.

Example 3.21

1
Determine the partial fraction expansion of the proper function
X z 
1 % 1.5z  ) 0.5z 

X z 2 1
Answer:
 %
z z % 1 z % 0.5

Example 3.22

1)z 
Determine the partial fraction expansion
X z 
1 % z  ) 0.5z 

Answer:
%j )j
 !  !
X z
    )    
z z% %j z% )j
   

Note that complex conjugate poles result in complex conjugate coefficients in the partial fraction expansion.

Multiple Poles. If Xz has a pole of multiplicity ℓ , that is it contains in its denominator the factor
z % p, ℓ , the expansion becomes

X z A A A,
 ) ) l )
z z % p, z % p,  z % p, ℓ
(3.33)

The method of determining the coefficients A , A , …, A, is illustrated in the following example.

Example 3.23

1
Determine the partial fraction expansion of
X z 
1 ) z  1 % z  

The z-Transform and Its Application to the Analysis of LTI Systems Page 26
Signals, Spectra and Signal Processing

X z 3/4 1/2 1/4

Answer:
 ) )
z z % 1 z % 1 z ) 1

Inversion of the partial fraction expansion. When the poles are real and distinct, the partial fraction

X z A A A,
 ) ) l)
z z % p z % p z % p,
(3.31)

becomes

A A A,
X z  ) ) l)
1 % p z  1 % p z  1 % p, z  (3.34)

Thus, by table look-up, each term of 3.34 has an inverse xn  p,  un when the ROC of Xz is the
exterior of the circle whose radius is p, (causal signal) and xn  %p,  u%n % 1when the ROC of
Xz is the interior of the circle whose radius is p, (anticausal signal). Thus, a real, single pole generates
an exponential signal.

conjugates. Thus if p, is a pole, then its conjugate pœ, is also a pole, and the coefficients of their partial
If the poles are distinct but complex conjugates, the coefficients of their partial fraction expansion are also

fractions are A, and Aœ, , which are also complex conjugates. Letting

A,  |A, |e™

and

p,  r, e_™

then the inverse of

A, Aœ,
X z  )
1 % p, z   1 % pœ, z  

will be

xn  2|A, |r, cosω, n )  un (3.35)

exponential envelope (a damped sinusoid). The distance r, of the pole from the origin determines the
Thus, each pair of complex conjugate poles results in a causal sinusoidal signal component with an

The z-Transform and Its Application to the Analysis of LTI Systems Page 27
Signals, Spectra and Signal Processing

exponential weighting (growing when r, ' 1, decaying when r, ; 1, and constant when r,  1). The
angle of the poles with respect to the positive real axis provides the frequency of the sinusoidal signal. The

phase of xn thru A, .

zeros, or equivalently, the numerator of the rational transform, affect only indirectly the amplitude and the

A/z % p,  is required. In the case of the double pole, the transform pair
In the case of multiple poles, either real or complex, the inverse transform of terms of the form

pz 
Z 
ž Ÿ  np un
1 % pz   (3.36)

can be used, provided that the ROC is |z| ' |p|.

Example 3.24

1
Determine the inverse transform of
X z 
1 % 1.5z  ) 0.5z 

when (a) ROC: |z| ' 1; (b) ROC: |z| ; 0.5; (c) ROC: 0.5 ; |z| ; 1
Answers:

xn  2 % 0.5  un

a.

xn  %2 ) 0.5 u%n % 1

b.

xn  %21 u%n % 1 % 0.5 un

c.

Determine the causal signal whose z-transform is

Example 3.25

1)z 
X z 
1 % z  ) 0.5z 

1 πn
Answer:
xn  √10 - . cos p % 1.249r un
√2 4

Determine the causal signal xn having the z-transform

Example 3.26

1
X z 
1 ) z  1 % z  

The z-Transform and Its Application to the Analysis of LTI Systems Page 28
Signals, Spectra and Signal Processing

1 3 n
Answer:
xn  ‘ %1 ) ) “ un
4 4 2

Determine the causal signal xn if its z-transform Xz is given by:
Drill Problems 3.3

1. Xz  N!20 N21

N!20

2. Xz 

0
 20 N 21
1

3. Xz 
2¢N2£
 20

4. Xz 
N21
N21

5. Xz  " 
 N(20N21
20 N21  A.$20

3.5. The One-Sided z-Transform

The two-sided z-transform requires that the corresponding signals be specified for the entire time range
%∞ ; n ; ∞. This requirement prevents its use for the evaluation of nonrelaxed systems. Recall that
these systems are described by difference equations with nonzero initial conditions. Thus, the one-sided z-
transform allows the solution of difference equations with initial conditions.

Definition. The one-sided or unilateral z-transform of the signal xn is defined by

X N z  x  n z
A
(3.37)

The one-sided z-transform differs from the two-sided transform in the lower limit of the summation, which is
always zero, whether or not the signal xn is zero for n ; 0 (is causal). Due to the choice of lower limit,
the one-sided z-transform has the following characteristics:

It does not contain information about the signal xn for negative values of n.
It is unique only for causal signals, because only these signals are zero for n ; 0.
•

The one-sided z-transform X N z of xn is identical to the two-sided z-transform of the signal
•

xnun. Since xnun is causal, the ROC of its transform, and hence the ROC of X N z is
•

always the exterior of a circle. Thus, with the one-sided z-transform, it is not necessary to refer to
their ROC.

The z-Transform and Its Application to the Analysis of LTI Systems Page 29
Signals, Spectra and Signal Processing

Determine the one-sided z-transform of the following signals (note that these signals are also given in
Example 3.27

(a) x n  1 2 5 7 0 1
Example 3.1)

(b) x n  1 2 5

7 0 1

(c) x! n  0 0 1 2 5 7 0 1

(d) x" n  δn

(e) x$ n  δn % k, k ' 0

(f) x( n  δn ) k, k ' 0

(a) XN z  1 ) 2z ) 5z ) 7z )z

Answers:
  ! $

(b) XN z  5 ) 7z 

)z !

(c) X!N z  z  ) 2z !

) 5z "
) 7z $
)z +

(d) X"N z  1;

(e) X$N z  z ,

(f) X(N z  0

Note that for noncausal signals, X N is not unique. Also for anticausal signals, X N is always zero.

Almost all properties of two-sided z-transform apply to one-sided z-transform with the exception of shifting
property.

Shifting Property of the One-Sided
-transform: Case 1 – Time Delay. If xn  X N z, then
,
¤
xn % k  z ,
¥X N z ) x%nz ¦

(3.38)

with k ' 0. In case xn is causal, then

¤
xn % k  z , N
X z (3.39)

The z-Transform and Its Application to the Analysis of LTI Systems Page 30
Signals, Spectra and Signal Processing

Determine the one-sided z-transform of the signals

Example 3.28

(a) xn  a un

(b) x n  xn % 2 where xn  a .

(a) X N z 
Answers:

 G20

(b) X N z   ) a z )a
21  
G20


Shifting Property of the One-Sided
-transform: Case 2 – Time Advance. If xn  X N z, then
, 
¤
xn ) k  z ¥X , N
z % x  n z ¦ (3.40)
A
for k ' 0.

With xn  a , find the one-sided z-transform of x n  xn ) k.

Example 3.29

z
Answer:
XN z  % z  % az
1 % az 

¤
Final Value Theorem. If xn  X N z, then

lim xn  limz % 1X N z

h h
(3.41)

The limit of 3.41 exists if the ROC of z % 1X N z includes the unit circle.

Solution of Difference Equations. The one-sided z-transform is a very efficient tool for the solution of

relating the two time-domain signals to an equivalent algebraic equation relating their one-sided z -
difference equations with nonzero initial conditions. It achieves that by reducing the difference equation

the time-domain is obtained by inverting the resulting z-transform.

transforms. This equation can be easily solved to obtain the transform of the desired signal. The signal in

The z-Transform and Its Application to the Analysis of LTI Systems Page 31
Signals, Spectra and Signal Processing

Determine a closed-form expression for the nth term of the Fibonacci sequence.
Example 3.30

1 1
Answer:
N
N N
yn  - . ”§1 ) √5¨ % §1 % √5¨ • u  n
√5 2

Example 3.31

yn  ayn % 1 ) xn

Determine the step response of the system

with %1 ; © ; 1 when the initial condition is y%1  1.

1
Answer:
yn  1 % a N 
u  n
1%a

Use the one-sided z-transform to determine yn, n – 0 in the following cases:

Drill Problems 3.4

1. yn )  yn % 1 % " yn % 2  0; y%1  y%2  1

 

2. yn % 1.5yn % 1 ) 0.5yn % 2  0; y%1  1, y%2  0

3. yn   yn % 1 ) xn; xn  p!r un; y%1  1

 

4. yn  " yn % 2 ) xn; xn  un; y%1  0, y%2  1



3.6. Analysis of LTI Systems in the z-Domain

In this section, the method by which linear, time-invariant systems are analyzed using z-transform is
discussed. In particular, the focus is given on the important class of pole-zero systems represented by
linear constant coefficient difference equations with arbitrary initial conditions.

Response of Systems with Rational System Function. If a system is described by the difference
equation
N M

yn  % a, yn % k ) b, xn % k (3.21)

,  , A

The z-Transform and Its Application to the Analysis of LTI Systems Page 32
Signals, Spectra and Signal Processing

whose system function is defined as

Y z ∑M
k0 bk z
,
B z
Hz   
Xz 1 ) ∑Nk1 ak z , Az
(3.22)

where the system functionHz will be represented by the rational function Bz/Az . Thus, Bz
contains the zeros of Hz and Az contains the poles of Hz. If there is an input signal xn to the
system, the transform of the output can be computed as the product

Yz  HzXz (3.20)

where Yz is the transform of the output yn , Hz is the system function and the transform of the
impulse response hn and Xz is the transform of the input xn. Since the input Xz will also be a
rational function, it can be expressed in the form

Nz
« z 
Q z
(3.42)

If the system is initially relaxed, the z-transform of the output will have the form

BzNz
Yz  HzXz 
AzQz
(3.43)

This system will have system poles due to Az as p , p , …, pN and input poles due to Qz as q , q ,
…, qL . We assume p, c q¯ for all k  1, 2, … , N and m  1, 2, … , L and that there are no poles
coincident with the zeros of Bz and Nz. The partial fraction expansion of 3.43 yields
N L
A, Q,
Y z  )
1 % p, z  1 % q, z 
(3.44)
,  , 

The output yn which is the inverse of Yz will be

N L

yn  A ,  p ,  u  n ) Q ,  q ,  u  n (3.45)
,  , 

It can be observed that the output response has two parts:

The response due to the system poles p , p , …, pN , which is called the natural response.
The response due to the input poles q , q , …, qL , which is called the forced response.
•
•

The z-Transform and Its Application to the Analysis of LTI Systems Page 33
Signals, Spectra and Signal Processing

This is in accordance with the solutions of the difference equation as demonstrated before.

When Xz and Hz have one or more poles in common or when Hz and/or Xz contain multiple-
order poles, then Yz will have multiple-order poles. Consequently, the partial fraction expansion of Yz
will contain factors of the form 1/1 % p± z  ,, k  1, 2, … , m where m is the pole order. The inversion
of these factors will produce the terms of the form n,  p± in the output yn of the system, as shown in
the previous section.

Response of Pole-Zero System with Nonzero Initial Conditions. Suppose that a signal xn is applied
to the pole-zero system at n  0. Thus, the signal xn is assumed to be causal. The effect of all previous
input signals to the system are reflected in the initial conditions y%1, y%2, …, y%N. The one-sided
z-transform can then be used to determine the output yn for n – 0 so that the initial conditions can be
dealt with. The z-transform of the difference equation 3.21 becomes
N , M
N
Y z  % a, z ,
¥Y N
z ) y%nz ¦ ) b, z ,
X z (3.46)
,   , A

Thus, 3.46 can be expressed as

∑M
, A b, z
, ∑N
,  a, z
, ∑,
 y%nz
Y N z  X z %
1 ) ∑N
,  a, z
, 1 ) ∑,  a, z ,
N
(3.47a)

NA z
Y N z  HzXz )
Az
(3.47b)

where NA z  % ∑N ,  a, z
, ∑,
 y%nz . From 3.47a and 3.47b, it can be observed that the
transform of the response with nonzero initial conditions has two parts:

responseY^ z.
• The output of the system due to the input without initial condition, or the zero-state

responseYaN n.
• The output of the system due to the initial conditions without input, or the zero-input

domain by determining the inverse z-transforms of Y^ z and YaN z separately and adding the results.
Hence the total response is the sum of these two output components, which can be expressed in the time

It can be separately shown that the effect of initial conditions is to slightly alter A, in

The z-Transform and Its Application to the Analysis of LTI Systems Page 34
Signals, Spectra and Signal Processing
N L
A, Q,
Y z  )
1 % p, z  1 % q, z 
(3.44)
,  , 

or the natural response of the system.

Example 3.32

yn  0.9yn % 1 % 0.81yn % 2 ) xn

Determine the unit step response of the system described by the difference equation

(a) y%1  y%2  0

under the following initial conditions:

(b) y%1  y%2  1

Answers:
(a) yn  ”1.099 ) 1.0880.9 cos p ! % 1.66r• un
²

(b) yn  ”1.099 % 0.10680.9 cos p ! r• un

²

Transient and Steady-State Response. The natural response of a causal system has the form
N

y 7  n  A ,  p ,  u  n (3.48)
, 

If all the system poles are less than one, then the natural response decays to zero as n approaches infinity.
In such a case, the natural response is referred to as the transient response. The rate at which the natural
response decays to zero depends on the magnitude of the poles. The farther the poles away from the
origin, the slower the decay of the response.

The forced response of the system has the form

L

y³7 n  Q ,  q ,  u  n (3.49)

, 

If all of the input poles are inside the unit circle, the forced response will decay to zero as n approaches

forced response persists for n – 0. In this case, the forced response is called the steady-state response of
infinity. But when the input poles are on the unit circle (in the case of step input or sinusoidal input) the

the system.

The z-Transform and Its Application to the Analysis of LTI Systems Page 35
Signals, Spectra and Signal Processing

Example 3.33

yn  0.5yn % 1 ) xn

Determine the transient and steady-state responses of the system characterized by the difference equation

when the input signal is xn  10 cos p r un. The system is initially at rest.
²
"

y 7 n  6.30.5 un

Answer:

πn
y³7 n  13.56 cos p % 0.5r un
4

impulse response hn satisfies the condition hn  0 for n ; 0. In terms of its system function, a
Causality and Stability. As defined previously, a causal linear time-invariant system is one whose unit

of radius r ; ∞, including the point z  ∞.

linear, time-invariant system is causal if and only if the ROC of the system function is the exterior of a circle

Also, it is stated before that a system is BIBO stable if its impulse response is absolutely summable, which
implies that its impulse response decays to zero as time approaches infinity. In terms of the system
function, a linear time-invariant system is BIBO stable if and only if the ROC of the system includes the unit
circle.

The z-Transform and Its Application to the Analysis of LTI Systems Page 36