You are on page 1of 14

184 The z-T ransform and Its Application to the Analysis of LTI Systems Chap.

W e h av e n o w d e m o n s tra te d th a t ra tio n a l z -tra n s fo rm s a re e n c o u n te re d in


c o m m o n ly u se d sy stem s a n d in th e c h a ra c te riz a tio n o f lin e a r tim e -in v a ria n t sys­
tem s. In S ectio n 3.4 w e d esc rib e se v e ra l m e th o d s fo r d e te rm in in g th e inverse
z -tra n sfo rm o f ra tio n a l fu n ctio n s.

3.4 INVERSION OF THE Z-TRANSFORM

A s w e saw in S ectio n 3.1.2, th e in v erse z -tra n s fo rm is fo rm ally given by

x ( n ) = - — < £ x ( z ) z n~ 1d z (3.4.1)
2 njjt
w h ere th e in te g ra l is a c o n to u r in te g ra l o v e r a clo sed p a th C th a t en clo ses the
o rig in an d lies w ith in th e reg io n o f c o n v e rg e n c e o f ^ ( z ) . F o r sim plicity, C can be
ta k e n as a circle in th e R O C o f X (z) in th e z-p lan e.
T h e re a re th re e m e th o d s th a t a re o fte n u se d fo r th e e v a lu a tio n o f th e inverse
z-tran sfo rm in practice:

1. D ire c t e v a lu a tio n o f (3.4.1), by c o n to u r in te g ra tio n .


2. E x p a n sio n in to a se rie s o f te rm s, in th e v a ria b le s z, an d z _1.
3. P a rtia l-fra c tio n ex p an sio n a n d ta b le lo o k u p .

3.4.1 The Inverse z-Transform by Contour Integration

In th is se ctio n w e d e m o n s tra te th e use o f th e C au ch y re sid u e th e o r e m to d e te rm in e


th e in v erse z -tra n sfo rm d irectly fro m th e c o n to u r in teg ral.

Cauchy residue theorem. L e t / ( z ) b e a fu n c tio n o f th e c o m p lex v ariab le


z an d C b e a clo sed p a th in th e z -p lan e. If th e d e riv a tiv e d f ( z ) / d z exists o n and
inside th e c o n to u r C a n d if / ( z ) has no p o le s a t z = zo, th e n

- L (f) J ! ± d z
2njjcz-zo
= |^
10,
<Zl,)' if zo is ou tsid e C
(3.4.2)

M o re g en erally , if th e (k + l ) - o r d e r d e riv a tiv e o f / ( z) exists a n d / ( z ) h a s n o p o les


a t z = zo, th e n

1 d k- ' f ( z )
»-!>' C 04.3)
0, if zo is o u ts id e C
T h e v alu es o n th e rig h t-h a n d sid e o f (3.4.2) a n d (3.4.3) a re c a lle d th e re sid u e s of
th e p o le a t z = zo- T h e re su lts in (3.4.2) a n d (3.4.3) a re tw o fo rm s o f th e Cauc hy
residue t heorem.
W e can ap p ly (3.4.2) a n d (3.4.3) to o b ta in th e v a lu e s o f m o re g en eral c o n to u r
in teg rals. T o b e specific, su p p o s e th a t th e in te g ra n d o f th e c o n to u r in te g ra l is
Sec. 3.4 Inversion of th e 2 - T ra n s fo rm 185

P(z) = f ( z ) / g ( z ) ~ w h e re f ( z ) h a s no p o les inside th e c o n to u r C an d g (z) is a


p o ly n o m ial w ith d istin ct (sim p le) ro o ts c i, ^ 2 . ___-n inside C. T h e n

(3.4.4)

1=1
w h ere
f(z)
A l (z) = ( z - z i ) P{ z ) = ( z - z l) -J - 1x (3.4.5)
g( z)
T h e v alu es (A, (-;,)} a re re sid u e s o f th e c o rre sp o n d in g p o le s at z = / = 1, 2 , . . . . n.
H e n c e th e v alu e o f th e c o n to u r in te g ra l is e q u a l to th e sum o f th e resid u es o f all
th e p o le s in sid e th e c o n to u r C.
W e o b se rv e th a t (3.4.4) w as o b ta in e d by p e rfo rm in g a p a rtia l-fra c tio n e x p a n ­
sion o f th e in te g ra n d an d ap p ly in g (3.4.2). W h en g(z) has m u ltip le -o rd e r ro o ts
as w ell as sim p le ro o ts inside th e c o n to u r, th e p a rtia l-fra c tio n e x p a n sio n , w ith a p ­
p ro p ria te m o d ifica tio n s, an d (3.4.3) can b e used to e v a lu a te th e resid u es at th e
c o rre sp o n d in g p o les.
In th e case o f th e in v erse z -tra n sfo rm , w e h ave

[resid u e of X (z )z n 1 a t z (3.4.6)
a ll p o l e s U t) in s id e C

p ro v id e d th a t th e p o les {z,} a re sim ple. If X (z )r "_1 has no p o le s inside th e c o n to u r


C fo r o n e o r m o re v alu es o f n, th e n x (n) = 0 fo r th e s e values.
T h e fo llo w in g ex am p le illu stra te s th e e v a lu a tio n o f th e in v erse z-tra n sfo rm
by u se o f th e C a u ch y re sid u e th e o re m .
Exam ple 3.4.1
Evaluate the inverse z-transform of

X(z) = ---------- r kl > kil


1 —az~
using the complex inversion integral.
Solution We have

where C is a circle at radius greater than |a|. We shall evaluate this integral using
(3.4.2) with f ( z ) = z". We distinguish two cases.
186 The ^-Transform and Its Application to the Analysis of LTI Systems Chap. 3

L If n > 0, f ( z ) has only zeros and hence no poles inside C. T he only pole inside
C is z = a. Hence

*(*) =• /(zo) = a n n > 0

2. If n < 0, / ( z ) = z" has an nth-order pole at z = 0, which is also inside C. Thus


there are contributions from both poles. For n = - 1 we have
1
x ( - l ) -= 1 S) 1 dz = 1 + - = 0
2 nj j c z ( z — a) z —a

If n = —2, we have

= 0
2) 2 n j § z H z - a ) dZ dz ( z - f l )
By continuing in the same way we can show that *(n) = 0 for n < 0. Thus
x (n) = a"u(n)

3.4.2 The Inverse z-Transform by Power Series


Expansion

The basic idea in this m eth od is the follow ing: G iven a z-transform X ( z ) with its
corresponding R O C , w e can expand X (z) into a p ow er series o f the form
OO

X(z) = £ c»z~n (3.4.7)


co

w hich con verges in the given R O C . T h en , by the u n iqu en ess o f the z-transform,
x ( n ) = c„ for all n. W hen X ( z ) is rational, the exp an sion can b e perform ed by
long division.
T o illustrate this tech n iqu e, w e w ill invert som e z-transform s involving the
sam e expression for X ( z ) , but different R O C . T his w ill also serve to em phasize
again the im portance o f the R O C in d ealing with z-transform s.
Exam ple 3A 2
D eterm ine the inverse z-transform of

1 —1.5z_1 + 0.5z “2
when

(a) ROC: |z| > 1


(b) ROC: |z| < 0.5

Solution
(a) Since the R O C is the exterior of a circle, we expect x(n) to be a causal signal.
Thus we seek a power series expansion in negative powers of z. By dividing
Sec. 3.4 Inversion of the z-Transform 187

the num erator of X{z) by its denom inator, we obtain the power series

A’UI = ! _ 3 _ -; + = 1+ + ^ + TE; "4 + " '

By com paring this relation with (3.1.1), we conclude that

Note that in each step of the long-division process, we eliminate the lowest-
power term of c~*.
(b) In this case the ROC is the interior of a circle. Consequently, the signal x(n)
is anticausal. To obtain a power series expansion in positive powers of c. we
perform the long division in the following way:
2: 2 + 6c3 + 14c4 + 30cs + 62c* + ■• ■
+ ill
1 - 3: + 2c:
3c - 2z 2
3c - 9c: + 6c3
l z 2 - 6c3
7 r - 21 c3 + 14c4
15c3 - 14c4
15c3 - 45c4 + 30cs
31c4 - 30c5
Thus

X (c) = , 1------:------= 2c: + 6c3 + 14c4 + 30c5 + 62cfi + ■• •


1_
In this case x(n) = 0 for n > 0. By comparing this result to (3.1.1), we conclude
that
Jt(n) = { 62. 30. 14.6,2, 0. 0}
t

We observe that in each step of the long-division process, the lowest-power


term of c is eliminated. We emphasize that in the case of anticausal sig­
nals we simply carry out the long division by writing down the two poly­
nomials in “reverse” order (i.e., starting with the most negative term on the
left).

F ro m th is e x a m p le w e n o te th a t, in g e n eral, th e m e th o d o f long d ivision will


n o t p ro v id e a n sw ers fo r x( n) w h en n is larg e b e c a u se th e lo n g division b eco m es
ted io u s. A lth o u g h , th e m e th o d p ro v id es a d irect e v a lu a tio n o f x( n ) , a clo sed -fo rm
so lu tio n is n o t p o ssib le , ex cep t if th e resu ltin g p a tte r n is sim p le e n o u g h to infer
th e g e n e ra l te rm x ( n ) . H e n c e th is m e th o d is used only if o n e w ish e d to d e te rm in e
th e v a lu e s o f th e first few sa m p le s o f th e signal.
188 The z-Transform and Its Application to the Analysis of LTI Systems Chap. 3

Example 3.4.3
Determ ine the inverse z-transform of
X(z) = log(l + az_1) |zj > |a|
Solution Using the power series expansion for log(l + x ), with |jc| < 1, we have
^ ( - r v r "

Thus

0, n < 0
Expansion of irrational functions into power series can be obtained from tables.

3.4.3 The Inverse z-Transform by Partial-Fraction


Expansion

In th e tab le lo o k u p m e th o d , w e a tte m p t to e x p ress th e fu n c tio n X (z) as a linear


c o m b in a tio n
X ( z ) = a j X] (;) + 02 X 2 ( 1 ) + • - - + c/Kx fc(z) (3.4.8)
w h ere X] ( ; ) , . . . . X k (z ) a re ex p ressio n s w ith in v erse tra n sfo rm s x \ ( n ) , . . . , x k W
av ailab le in a ta b le o f z-tra n sfo rm p airs. If such a d e c o m p o s itio n is possible,
th e n x( n) , th e in v erse z -tra n sfo rm o f X ( z ) , can easily b e fo u n d using th e lin earity
p ro p e rty as
x ( n) = at]Xi(n) + a 2x 2 (n) H--------\ - a Kx K (n) (3.4.9)
T h is a p p ro a c h is p a rtic u la rly useful if X ( z ) is a ra tio n a l fu n ctio n , a s in (3.3.1). W ith ­
o u t loss o f g e n erality , w e assu m e th a t ao = 1, so th a t (3.3.1) can b e e x p ressed as
_ t o _ b +b,r' + - + t„ r«
D(z) l + f l i Z - 1 H----- + a u z ~ N
N o te th a t if a 0 ^ 1. w e can o b ta in (3.4.10) fro m (3.3.1) by div id in g b o th n u m e ra to r
an d d e n o m in a to r by ao-
A ra tio n a l fu n ctio n o f th e form (3.4.10) is called p r o p e r if a N ^ 0 an d M < N.
F ro m (3.3.2) it fo llo w s th a t th is is e q u iv a le n t to saying th a t th e n u m b e r o f finite
zero s is less th a n th e n u m b e r o f fin ite p o les.
A n im p ro p e r ra tio n a l fu n ctio n ( M > N ) can alw ays b e w ritte n as th e sum of
a p o ly n o m ial an d a p r o p e r ra tio n a l fu n ctio n . T h is p ro c e d u re is illu stra te d by the
fo llow ing ex am p le.
Example 3.4.4
Express the im proper rational transform
l + 3 z - ' + n z - 2 + l 2-3
1 5
+ 1 + 6Z
in terms of a polynomial and a proper function.
Sec. 3.4 Inversion of the z-Transform 189

Solution First, we note that we should reduce the num erator so that the term s ; -2
and c- *' are eliminated. Thus we should carry out the long division with these two
polynomials written in reverse order. We stop the division when the order of the
rem ainder becomes Then we obtain

^ = 1 + 2: - i +

In g e n e ra l, an y im p ro p e r ra tio n a l fu n ctio n (M > N ) can b e e x p ressed as

X ( ;) = W i = Co + C i:" 1 + ' ' ' + Cu~n Z ~im~N) + (3 A ll)


T h e in v erse z -tra n s fo rm o f the p o ly n o m ial can easily b e fo u n d by in sp ectio n .
W e fo cu s o u r a tte n tio n on th e inversion o f p r o p e r ra tio n a l tra n sfo rm s, since any
im p ro p e r fu n c tio n can b e tra n sfo rm e d in to a p ro p e r fu n c tio n by using (3.4.11).
W e carry o u t th e d e v e lo p m e n t in tw o steps. F irst, we p e rfo rm a p a rtia l fra c ­
tio n e x p a n sio n o f th e p r o p e r ra tio n a l fu n ctio n a n d th e n w e in v ert each o f th e
term s.
L e t A"(c) b e a p ro p e r ra tio n a l fu n ctio n , th a t is,

* (.-) = — = - " - - ' " T ,1 +— +buZ^ (3.4.12)


D( z) 1 + ^ iC + • *• -t~
w h ere
aN ^ 0 an d M < N
T o sim p lify o u r discu ssion w e e lim in ate n eg ativ e p o w ers of c by m ultip ly in g b o th
th e n u m e r a to r a n d d e n o m in a to r of (3,4,12) by z N. T h is resu lts in
L „N I L _ y v -l I I L „ N -M
X{z) = - (3.4.13,
CA 4- a \ z N + ------(- <a/v
w hich c o n ta in s only p o sitiv e p o w e rs o f Since N > M , th e fu n ctio n
,N -2 _i_____ l
(3,4.14)
; : w + tiic w- 1 + - - - + flW
is also alw ays p ro p e r.
O u r ta sk in p e rfo rm in g a p a rtia l-fra c tio n ex p an sio n is to ex p ress (3.4.14)
o r, e q u iv a le n tly , (3.4.12) as a sum o f sim ple fractio n s. F o r th is p u rp o se w e first
fa c to r th e d e n o m in a to r p o ly n o m ial in (3.4.14) in to facto rs th a t c o n tain th e poles
P i, p 2, . . . , p n o f X (z). W e d istinguish tw o cases.

Distinct poles. S u p p o se th a t th e p o le s p \ , p 2 ........ p/v a re all d iffe re n t (dis­


tin ct). T h e n w e se e k an ex p an sio n of th e fo rm

+ -. - + ( 3 . 4 . 1 5 )
z z — p\ z — P2 z — Pn
T h e p ro b le m is to d e te rm in e th e coefficients A i , A 2 , . - . , A s - T h e re a re tw o w ays
to so lv e th is p ro b le m , as illu stra te d in th e follow ing exam p le.
190 The z-Transform and Its Application to the Analysis of LTI Systems Chap. 3

Example 3.4 .5
D eterm ine the partial-fraction expansion of the proper function

(3'‘U 6 ,
Solution First we elim inate the negative powers, by multiplying both num erator and
denom inator by z2. Thus
.2
X i z ) = zV2 - 11.5z
< +i 0n.5
The poles of X(z) are p\ = 1 and P2 = 0.5. Consequently, the expansion of the form
(3.4,15) is
X(z) z A\ A2
(3.4.17)
Cz —l)(z —0.5) z 1 z —0.5
A very simple method to determ ine A[ and A 2 is to multiply the equation by the
denom inator term (z - l)(z - 0.5). Thus we obtain
z = ( z - 0 .5 M i + ( z - 1 ) A 2 (3.4.18)
Now if we set z = p\ = 1 in (3.4.18), we eliminate the term involving A2. Hence
1 = (1 -0 .5 )A ,
Thus we obtain the result A i = 2. Next we return to (3.4.18) and set z = p 2 = 0.5,
thus eliminating the term involving Ai, so we have

0.5 = ( 0 .5 - 1)A2
and hence Ai = —1. Therefore, the result of the partial-fraction expansion is
X(z) 2 1
(3.4.19)
z- 1 z - 0.5

T h e exam ple given ab ove suggests that w e can determ ine the coefficients A \,
A i , . . . , Afj , by m ultiplying b oth sides o f (3.4.15) by each o f the term s (z - Pk).
k = 1 , 2 , , . . , N , and evaluating the resulting exp ression s at the corresp on d ing pole
p osition s, p \ , p i .........P n ■ T h u s w e have, in general,

(Z- Pl)X(;) = (z-wM.i+ ... + /lt+,..+ fa-P»)^ (3420)


z z - PI z - Pn
C onsequently, w ith z = Pk, (3.4.20) yield s the Jtth coefficient as

Ak = ( z ~ ^ )X (; )i k = 1, 2. . N (3.4.21)
z \z-pl
Exam ple 3.4.6
Determ ine the partial-fraction expansion of

1 17- ^ 0 .5 ,- ,3A22)
Sec. 3.4 inversion of the z-Transform 191

Solution To eliminate negative powers o f ; in (3.4.22), we multiply both num erator


and denom inator by Thus
X(z) z+ 1
- z2 - ; + 0.5
The poles of X(z) are complex conjugates
Px = \ + J i

and
Pi = \ ~ j \

Since p\ ^ p 2- we seek an expansion of the form (3.4.15). Thus


* (:) ; + l A] Ai
z (z-p i)(:-p 2) z-P\ z-pi
To obtain A , and A2, we use the formula (3,4.21), Thus we obtain
(z-pi)X(z) ; + 1 I ?+ !
At = -----------
- P 2 \ ^ P1 3+ J5-5+./5
( z - p 2)X(z) ; + l 1 1
A- = ------------
- Pi :=P2
T h e ex p a n sio n (3,4.15) a n d the fo rm u la (3.4.21) h o ld fo r b o th real a n d c o m ­
p lex p o les. T h e o n ly c o n s tra in t is th a t all p o les be d istin ct. W e also n o te th at
A; = A*. It can b e easily seen th a t th is is a c o n s e q u e n c e o f th e fact th a t p 2 = p ' .
In o th e r w o rd s, comp l ex- conj ugat e pol es result in c o mpl ex- con j ugat e coefficients in
the part ial-fraction expansi on. T h is sim ple resu lt will p ro v e v ery u se fu l la te r in o u r
d iscu ssio n .

M u ltip le - o rd e r p o l e s . If X U) h a s a p o le o f m u ltip lic ity /, th a t is, it co n ta in s


in its d e n o m in a to r th e fa c to r (z - pk)1, th e n th e e x p a n s io n (3.4.15) is n o lo n g er
tru e. In th is case a d iffe re n t ex p an sio n is n e e d e d . F irst, w e in v e stig a te th e case of
a d o u b le p o le (i.e., 1 = 2).
Exam ple 3.4.7
D eterm ine the partial-fraction expansion of

Solution First, we express (3.4,23) in terms of positive powers of in the form


* ( ;) = z2
: (z + 1)(; - l ) 2
X (z) has a simple pole at p\ = - 1 and a double pole pi = p$ = 1. In such a case the
appropriate partial-fraction expansion is

™ = ______ t ______ = + (3424)


z (z + l)(z - 1)2 z+ 1 z - l (z -1 )2
The problem is to determ ine the coefficients A 1, A2, and A3.
192 The z-Transform and Its Application to the Analysis of LTI Systems Chap. 3

We proceed as in the case of distinct poles. T o determ ine Aj, we multiply both
sides of (3.4.24) by (z + 1) and evaluate the result at z = - 1 . Thus (3.4.24) becomes
(z + l)X (z) z+ 1 z+ 1
------------------ = Ai + ------ - A 2 + ------- r - z A i
z z-1 ( z - 1)2
which, when evaluated at z = - 1 , yields
(z + 1)X(z) 1
A i = ------------------
z 4

Next, if we multiply both sides of (3.4.24) by (z - l ) 2, we obtain

= + + <3.4.25,
Z z+ 1
Now, if we evaluate (3.4.25) at z = 1, we obtain A 3. Thus
(z — l) 2X(z) I = 1
* Li 2
The remaining coefficient Az can be obtained by differentiating both sides of
(3.4.25) with respect to z and evaluating the result at z = 1. Note that it is not
necessary formally to carry out the differentiation of the right-hand side of (3.4.25),
since all term s except A2 vanish when we set z = 1. Thus
d (z — l ) 2X(z)
A2 = —
dz

T h e gen eralization o f the p rocedure in the exam p le ab ove to the case o f an


/th-order p ole ( z — p k)! is straightforward. T h e partial-fraction expansion must
contain the terms
■Alt Mk
Z - Pk ( z - Pk)2 (z - Pk)1
T h e coefficients {A,*} can b e evalu ated through d ifferen tiation as illustrated in
E xam p le 3.4.7 for / = 2.
N ow that w e have perform ed the partial-fraction exp an sion , w e are ready to
take the final step in the inversion o f X (z). First, let us con sid er th e case in which
X ( z ) contains distinct poles. F rom the partial-fraction exp an sion (3.4.15), it easily
follow s that

X ( z ) — A \ -------------r + A i -------------- + ■■•-)- A n --------------- (3.4.27)


1 - P\Z~1 1 - PlZ 1 - P n Z~ 1

T he inverse z-transform , x ( n ) = Z ~ i { X( z ) ) , can b e ob tain ed by inverting each


term in (3.4.27) and taking the corresponding linear com b ination . From T ab le 3.3
it fo llo w s that these term s can be inverted using th e form ula
( p k)nu{n), if RO C : |*| > Ip*J
(causal signals) „
- ( P k ) nu ( - n - 1), if RO C: |z| < \P k \
(anticausal sign als)
Sec. 3.4 Inversion of the ^-Transform 193

If th e signal jc(n) is cau sal, the R O C is |r | > p max- w h e re p max = m ax{|/>i|,


\p2\.........IpnII- In th is case all te rm s in (3.4.27) re su lt in causal signal c o m p o n e n ts
an d th e signal x ( n ) is given by

jr(/i) = (A i + A -Pz --------f" A Np nN )u(n) (3.4.29)


If all p o le s a re real. (3.4.29) is th e d e sire d e x p ressio n fo r th e signal j («). T h u s a
causal sig n al, h av in g a ; -tra n s fo rm th a t co n tain s real a n d d istin ct p o les, is a linear
c o m b in a tio n o f re a l e x p o n e n tia l signals.
S u p p o se n o w th a t all p o les a re d istin ct b u t som e o f th e m a re co m p lex . In
this case so m e o f th e te rm s in (3.4.27) re su lt in com plex e x p o n e n tia l c o m p o n en ts.
H o w e v e r, if th e signal x ( n) is real, w e sh o u ld b e ab le to re d u c e th e se te rm s in to
real c o m p o n e n ts. If x ( n ) is real, th e p o ly n o m ials a p p e a rin g in X (z) have real co ­
efficients. In th is case, as w e h av e seen in S ectio n 3.3, if pj is a p o le , its com plex
c o n ju g a te p j is also a p o le. A s w as d e m o n s tra te d in E x am p le 3.4.6, th e c o rre s p o n d ­
ing coefficien ts in th e p a rtia l-fra c tio n e x p an sio n a re also co m p lex co n ju g ates. T h u s
th e c o n trib u tio n o f tw o c o m p lex -co n ju g ate p o les is of th e form

x k (n) = \ A k {pt )n + A H p l )"]«(«) (3.4.30)

T h e se tw o te rm s can be c o m b in ed to form a real signal c o m p o n e n t. F irst,


we ex p re ss Aj an d Pj in p o la r form (i.e., a m p litu d e an d p h a se ) as

A* = \ Ak \eja' (3.4.31)

Pi: = (3.4.32)

w h ere a k an d fik a re th e p h a s e c o m p o n e n ts o f A k a n d p k. S u b stitu tio n o f th ese


re la tio n s in to (3.4.30) gives

x k{n) = IAi ’ + e - -'(A"+“‘ l]u(n)

or, eq u iv alen tly ,

x k(n) = 2|A *|r" c o s($ tn + a k)u(n) (3.4.33)

T h u s w e co n clu d e th a t

Z - 1 ( - — — — r + -— ~ — r ) = 2 \ A k \rR
k c o s ( f tn + a k) u(n) (3.4.34)
\i - p kz ~ l i - p;z~ v
if th e R O C is |zj > \ pk \ = rk .
F ro m (3.4.34) we o b se rv e th a t ea c h p a ir o f c o m p le x -c o n ju g a te p o le s in th e
z -d o m ain resu lts in a causal sin u so id al signal c o m p o n e n t w ith an e x p o n e n tia l e n ­
v elo p e. T h e d ista n c e rk o f th e p o le fro m th e o rig in d e te rm in e s th e e x p o n e n tia l
w eig h tin g (g ro w in g if r k > 1, d ecay in g if r k < 1, c o n s ta n t if rk = 1). T h e angle of
th e p o le s w ith re sp e c t to th e p o sitiv e re a l axis p ro v id e s th e fre q u e n c y o f th e sin u ­
so id a l signal. T h e zero s, o r e q u iv alen tly th e n u m e ra to r o f th e ra tio n a l tran sfo rm ,
affect o n ly in d ire c tly th e a m p litu d e an d th e p h a se o f x k (n) th ro u g h A k.
In th e case o f mul t i pl e p o les, e ith e r re a l o r co m p lex , th e in v erse tra n sfo rm
o f te rm s o f th e fo rm A j ( z — p k)n is re q u ire d . In th e case o f a d o u b le p o le the
194 The 2 -Transform and Its Application to the Analysis of LTI Systems Chap. 3

fo llow ing tra n sfo rm p a ir (see T a b le 3.3) is q u ite useful:


pz~x
= n p nu( n) (3.4.35)
(1 - p z ] )2
p ro v id e d th a t th e R O C is |z| > \p\. T h e g e n e ra liz a tio n to th e case o f p o le s w ith
h ig h e r m u ltip licity is left as an exercise fo r th e re a d e r.
Example 3.4.8
Determ ine the inverse z-transform of
1
X U) =
1 - 1 .5 ;-' +0.5z~ 2

(a) RO C III > 1


(b) RO C Izl < 0.5
(c) RO C 0.5 < |z| < 1

Solution This is the same problem that we treated in Exam ple 3.4.2. The partial-
fraction expansion for X(z) was determined in Example 3.4.5. The partial-fraction
expansion of X(z) yields

= <3-4-36>
To invert X(z) we should apply (3.4,28) for pi — 1 and p 2 = 0.5. However, this
requires the specification of the corresponding ROC.

(a) In case when the R O C is |z| > 1, the signal x(n) is causal and both term s in
(3.4.36) are causal terms. According to (3.4.28), we obtain
x(n) = 2 (l)n«(n) —(0.5)"u(n) = (2 — 0.5 ”)u(n) (3.4.37)
which agrees with the result in Example 3.4.2(a).
(b) When the R O C is |z| < 0.5, the signal x(n) is anticausal. Thus both term s in
(3.4.36) result in anticausal components. From (3.4.28) we obtain
x(n) = [—2 + (0.5)'I]u(—n — 1) (3.4.38)
(c) In this case the ROC 0.5 < |z| < 1 is a ring, which implies that the signal x(n) is
two-sided. Thus one of the term s corresponds to a causal signal and the other
to an anticausal signal. Obviously, the given ROC is the overlapping of the
regions (z| > 0.5 and |z| < 1. Hence the pole p 2 = 0.5 provides the causal part
and the pole p\ = 1 the anticausal. Thus
x(n) = -2 (1 ) "u( - n - 1) - (0.5)"«(n) (3.4.39)

Example 3.4.9
D eterm ine the causal signal jc(n) whose z-transform is given by
Sec, 3.4 Inversion of the z-Transtorm 195

Solution In Exam ple 3.4.6 we have obtained the partial-fraction expansion as

where
A, = Al = j - j
and

Pi = p ’ = i
Since we have a pair of complex-conjugate poles, we should use (3.4.34). The
polar forms of Aj and p, are

Hence

Example 3.4.10
D eterm ine the causal signal x(n) having the ;-transiorm

X(z) =
(1 + ; - ’)(]
Solution From Example 3.4.7 we have
3 .-1

X(Z) 41 +
+ T
41 + 2 ( 1 - ; - 1)2
By applying the inverse transform relations in (3.4.28) and (3.4.35), we obtain
1 3 1 f 1 3 w*l
x( n ) = - ( —l)"«(n) + t« (« ) + - n u ( n ) = t(-1) + - + - u(n)
4 4 2 4 4 2

3.4.4 Decomposition of Rational z-Transforms

A t th is p o in t it is a p p ro p ria te to discuss som e a d d itio n a l issues c o n c e rn in g th e


d e c o m p o s itio n o f ra tio n a l z-tran sfo rm s, w hich will p ro v e v ery u se fu l in th e im p le­
m e n ta tio n o f d isc re te -tim e system s.
S u p p o se th a t w e h ave a ra tio n a l z-tran sfo rm X ( z ) e x p re ss e d as

X( z) = (3.4.40)
196 The z-T ransform and Its Application to the Analysis of LTI Systems Chap. 3

w h ere, fo r sim plicity, w e h av e a ssu m ed th a t ao = 1. If M > N [i.e., X ( z ) is


im p ro p e r], w e c o n v e rt X (z) to a sum o f a p o ly n o m ial a n d a p r o p e r fu n c tio n
M-H
X ( z ) = J 2 c t z ~ k + X pA z ) (3.4.41)
k=o
If th e p o le s o f X pr( z ) are d istin ct, it can b e e x p a n d e d in p a rtia l fra c tio n s as

pT(z) = A \ - ------------- + A 2 ------------- r + --- + ^ jv -------------- r (3.4.42)


l-p \z~ l l ~ P 2Z~l 1 - PnZ~1
A s w e h av e a lre a d y o b se rv e d , th e re m ay b e so m e c o m p le x -c o n ju g a te p airs of
p o le s in (3.4.42). S in ce we u su a lly d eal w ith real signals, w e sh o u ld av o id com plex
co efficien ts in o u r d eco m p o sitio n . T h is can b e ach iev ed by g ro u p in g a n d co m b in in g
te rm s co n ta in in g co m p lex -co n ju g ate p o les, in th e follow ing w ay:
A — A p * z ~ l + A* - A * p z ~ l
1 - pz~] 1 — /5*z-1 1 - p z 1 - p*z ~' + PP*z ~2
(3.4.43)
_ bo + b -iZ ~ l
1 + a i z ~ l + 02 z ~2
w h ere
£>o = 2 R e ( A ) , a \ = —2 R e ( p )
(3.4.44)
b\ = - 2 Re (Ap*), a 2 = \ p \2
a re th e d e sire d co efficients. O b v io u sly , a n y ra tio n a l tra n sfo rm o f th e fo rm (3.4.43)
w ith co efficien ts given by (3.4.44), w hich is th e case w h en a 2 — 4 a 2 < 0, can be
in v e rte d using (3.4.34). B y c o m b in in g (3.4.41), (3.4.42), a n d (3.4.43) w e o b ta in a
p a rtia l-fra c tio n e x p a n sio n fo r th e z -tra n s fo rm w ith distinct p o les th a t co n ta in s real
coefficients. T h e g e n e ra l re su lt is
M—N K\ 1 1. 1 l. —1
Xiz) = ££ 3 C Z - * + E + E , / f'l -2
t l l + a u z 1+auZ 2
<3 A 4 S >
w h ere K\ + 2 AS = N . O b v io u sly , if Af = TV, th e first te rm is ju s t a c o n stan t,
a n d w h e n M < N , th is te rm v an ish es. W h e n th e r e a re also m u ltip le p oles, som e
a d d itio n a l h ig h e r-o rd e r te rm s sh o u ld b e in clu d ed in (3.4.45).
A n a lte rn a tiv e fo rm is o b ta in e d by ex p ressin g X ( z ) as a p r o d u c t o f sim ple
te rm s as in (3.4.40). H o w e v e r, th e co m p le x -c o n ju g a te p o le s a n d z e ro s sh o u ld be
co m b in e d to av o id co m p lex co efficien ts in th e d e c o m p o sitio n . S u ch c o m b in atio n s
re su lt in se c o n d -o rd e r ra tio n a l te rm s o f th e follow ing form :

(1 - Z*Z-1 )(1 - z*kz ~ ' ) 1 + b\kZ~l + b u z ~2


(3.4.46)
(1 - />*z- 1 ) ( l - p*kz ~ l ) 1 -I- a u z -1 + a u z ~2
w h ere
b\ k = - 2 R t ( z k ) , flu = —2 R e (p * )
(3.4.47)
b u = jz i l , 02* = \Pk\
Sec. 3.5 The One-sided 2 -Transform 197

A ssu m in g fo r sim p licity th a t M = N, w e se e th a t X (z) can be d e c o m p o s e d in the


follow ing way:

1 + a kz 1 1 + a u z 1+ axz 2 <3 -4 -48>

w h ere N = K\ + 2 / ^ * W e will r e tu rn to th e s e im p o rta n t fo rm s in C h a p te rs 7 a n d 8.

3.5 THE ONE-SIDED Z-TRANSFORM

T h e tw o -sid ed z -tra n sfo rm re q u ire s th a t th e c o rre sp o n d in g signals be specified


fo r th e e n tire tim e ran g e —oo < n < oo. T his re q u ire m e n t p re v e n ts its u se for
a v ery u se fu l fam ily o f p ra c tic a l p ro b lem s, n am ely th e e v a lu a tio n o f th e o u tp u t
o f n o n re la x e d system s. A s we recall, th e s e system s a re d e sc rib e d by d ifferen ce
e q u a tio n s w ith n o n z e ro initial c o n d itio n s. Since th e in p u t is a p p lie d a t a finite
tim e, say n ()l b o th in p u t a n d o u tp u t signals are specified fo r n > no, b u t by no
m e a n s a re z e ro fo r n < no- T h u s th e tw o -sid ed z-tran sfo rm c a n n o t b e used. In this
se c tio n w e d e v e lo p th e o n e -sid e d z-tra n sfo rm w hich can be u se d to solve differen ce
e q u a tio n s w ith in itial c o n d itio n s.

3.5.1 Definition and Properties

T h e one- si ded o r unilateral z -tra n sfo rm o f a signal x ( n ) is d efin ed by


CC

* + (;) -
n=0

W e also u se th e n o ta tio n s Z +{x(n)} a n d

*{n ) X + (z)

T h e o n e -sid e d z -tra n sfo rm d iffers fro m th e tw o -sid ed tra n sfo rm in th e low er


lim it o f th e su m m a tio n , w hich is aiw ays z e ro , w h e th e r or n o t th e signal x ( n ) is zero
fo r n < 0 (i.e., cau sal). D u e to th is choice o f lo w er lim it, th e o n e -sid e d z-tran sfo rm
h as th e fo llow ing ch aracteristics:

1. It d o e s n o t c o n ta in in fo rm a tio n a b o u t th e signal jc(n) fo r n e g a tiv e v alu es o f


tim e (i.e., fo r n < 0).
2. It is un i q u e o n ly fo r ca u sa l signals, becau se only th ese signals a re z e ro for
n < 0.
3. T h e o n e -sid e d z -tra n s fo rm A,+(z) o f x( n ) is id en tical to th e tw o -sid ed z-
tra n s fo rm o f th e signal x( n) u( n) . S ince x ( n ) u ( n ) is causal, th e R O C o f its
tra n sfo rm , a n d h en c e th e R O C of X + {z), is alw ays th e e x te rio r o f a circle.
T h u s w h en w e d e a l w ith o n e-sid ed z-tran sfo rm s, it is n o t n ecessa ry to re fe r
to th e ir R O C .

You might also like