INTRODUCTION TO DIGITAL SIGNAL PROCESSING

Dr. Dominic Ho
Department of Electrical Engineering & Computer Science
University of Missouri
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

Ch.4 Z-Transform and its Application to the Analysis of LTI System

4.1 The Z-Transform

The z-transform of a discrete-time signal x(n) is defined as
∞
Z{x(n)} = x(n) z −n = X(z) , ROC (1)
X

n=−∞
where z is a complex variable.
The set of z-values for which the summation converges is called the region of convergence
(ROC) of the transform. One must specify ROC in a z-transform.

4.1.1 Region of Convergence (ROC)

ROC is dependent on the duration of a discrete-time signal.
1. Finite duration signal
• Causal:
X(z) = aM z −M + aM +1z −(M +1) + · · · + aN z −N , N >M ≥0 (2)
ROC is the entire z-plane except at z = 0.
• Anticausal:
X(z) = a−N z N + a−N +1z N −1 + · · · + a−M z M , N >M ≥0 (3)
Dr. Dominic Ho 1
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

ROC is the entire z-plane except z = ∞.

• Two sided:
X(z) = a−M z M + a−M +1z M −1 + · · · + a0 + · · · + aN z −N , N, M ≥ 0 (4)
ROC is the entire z-plane except z = 0 and z = ∞.
2. Infinite duration sequence
• Causal:
ROC is |z| > r2.

Eg. Find the ROC of Z{x(n) = an u(n)}

Sol.
∞ 1
X(z) = anz −n = , if |az −1
| < 1. (5)
X

n=0 1 − az −1
Hence the ROC is |z| > |a|.

• Anticausal:
ROC is |z| < r1.

Eg. Find the ROC of Z{ x(n) = −bnu(−n − 1) }.

Dr. Dominic Ho 2
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

Sol.
−1 ∞
n −n
X(z) = −b z =− (b−1z)n
X X

n=−∞ n=1
−1
−b z 1
= = , if |b−1z| < 1. (6)
1 − b−1z 1 − bz −1
Hence the ROC is |z| < |b|.

• Two Sided:
ROC is r2 < |z| < r1.

Eg. Find the ROC of Z{ x(n) = anu(n) − bnu(−n − 1) }.

Sol.

Dr. Dominic Ho 3
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

4.1.2 Z-Transform of Some Common Signals

See text book, page 174 table 3.3.
4.1.3 Properties of Z-Transform
Z{x(n)} = X(z) , ROC: rl < |z| < ru (7)
Z{x1(n)} = X1(z) , ROC: r1l < |z| < r1u (8)
Z{x2(n)} = X2(z) , ROC: r1l < |z| < r2u (9)

1. Linearity
Z{ a1x1(n) + a2x2(n) } = a1X1(z) + a2X2(z) , (10)
ROC: max{ r1l , r2l } < |z| < min{ r1u, r2u }. (11)
2. Translation
Z{ x(n − n0) } = z −n0 X(z) , ROC: rl < |z| < ru. (12)
3. Multiplication by an Exponential
z!
n
Z{ a x(n) } = X , ROC: |a|rl < |z| < |a|ru. (13)
a
4. Multiplication by a ramp
dX(z)
Z{ nx(n) } = −z , ROC: rl < |z| < ru. (14)
dz
Dr. Dominic Ho 4
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

5. Convolution
Z{ x1(n) ∗ x2(n) } = X1(z) X2(z) , (15)
ROC: at least max{ r1l , r2l } < |z| < min{ r1u, r2u }. (16)
6. Multiplication
1 I z ! −1
Z{ x1(n)x2(n) } = X1(ν)X2 ν dν (17)
2πj c ν
ROC: at least r1l r2l < |z| < r1u r2u. (18)
where c is a complex contour integral and c is any closed contour in the intersection of
H

the ROC’s of X1(ν) and X2(z/ν).

7. Initial value theorem
x(0) = z→∞
lim X(z). (19)
8. Final value theorem
−1
lim
n→∞
x(n) = lim { (1 − z )X(z) }. (20)
z→1
9. Parseval’s theorem
∞ 1 I ∗  1  −1
 
∗
x1(n) x2 (n) = X (ν)X ν dν (21)
X
1 2
n=−∞ 2πj c ν∗
provided that r1l r2l < 1 < r1u r2u, and c is any close contour inside
r1l r2l < |z| < r1u r2u.
Dr. Dominic Ho 5
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

Eg. Find the z-transform of x(n) = (n − 2) cos( ωo(n − 2) ) u(n − 2).

Sol.

Dr. Dominic Ho 6
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

4.1.4 Poles and Zeros

In general,
N (z) b0 + b1z −1 + · · · + bM z −M −k
PM
k=0 bk z
X(z) = = = PN (22)
D(z) a0 + a1z −1 + · · · + aN z −N k=0 ak z
−k

6 0 and b0 6= 0,
If a0 =
b b
b0z −M z M + b10 z M −1 + · · · + bM0
X(z) =
a0z −N z N + aa01 z N −1 + · · · + aaN0
b0 N −M (z − z1) (z − z2) · · · (z − zM )
= z . (23)
a0 (z − p1) (z − p2) · · · (z − pN )
• z1, z2, · · · , zM are called zeros, since X(zi) = 0 for i = 1, 2, · · · , M .
• p1, p2, · · · , pN are called poles, since X(pi) = ∞ for i = 1, 2, · · · , N .
• if N > M , there are extra N − M zeros at zero.
• total number of zeros = total number of poles.
• poles are denoted by × in the z-plane while zeros by o.

Dr. Dominic Ho 7
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

Eg. Find the pole(s) and zero(s) of x(n) = anu(n), 0 < a < 1.
Sol.

4.1.5 Pole Location and Time Domain Behavior for Causal Signals
Consider x(n) = anu(n),
z
anu(n) ↔, ROC: |z| > a. (24)
z−a
• pole inside the unit circle, signal is bounded.
• pole closer to the origin, signal decays faster.
• pole on the unit circle, signal is oscillating.
• pole outside the unit circle, signal is unbounded.

Dr. Dominic Ho 8
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

4.2 Inverse Z-Transform

Three methods: (1) Complex inversion, (2) polynomial division, (3) partial fraction
expansion.
4.2.1 Power Series Expansion

Z(z) = · · · + x(−2)z 2 + x(−1)z + x(0) + x(1)z −1 + · · · (25)

Given a z-transform, we separate it into causal and anticausal parts according to the ROC.
The causal part is expanded in negative power of z while the anticausal part in positive
power of z.
Eg. Find Z −1{ z/(3z 2 + 4z + 1) } for ROC: |z| > 1 and for ROC: |z| < 1/3.
Sol.

Dr. Dominic Ho 9
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

4.2.2 Partial Fraction Expansion

The method factorizes the z-transform into first order section and uses table lookup to
obtain the inverse transform.
Useful z-transform pairs:
z
anu(n) ↔ , ROC: |z| > |a| (26)
z−a
z
−bnu(−n − 1) ↔ , ROC: |z| < |b| (27)
z−b
Eg. Find Z −1{ z/(3z 2 − 4z + 1) } for the ROC 1/3 < |z| < 1.
Sol.

Dr. Dominic Ho 10
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

4.3 Analysis of LTI System by Z-Transform

4.3.1 System Function

LTI system
x(n) y(n)
h(n)

The system function is defined as

∞
H(z) = Z{ h(n) } = h(n)z −n , ROC. (28)
X

n=−∞

Given an LTI system described by

N M
y(n) = − ak y(n − k) + bk x(n − k) (29)
X X

k=1 k=0
we do not need to solve for h(n) first to determine H(z).
If we put x(n) = δ(n) in the above equation, then y(n) = h(n). An impulse function is
zero from n = −∞ to -1, i.e. x(n) = 0 for n < 0. Hence y(n) = 0 for n < 0. Taking
z-transform of (29),
N M
−k
H(z) = − ak z H(z) + bk z −k since Z{δ(n)} = 1. (30)
X X

k=1 k=0

Dr. Dominic Ho 11
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

Hence PM −k
k=0 bk z
H(z) = . (31)
1 + N a z −k
P
k=1 k

Suppose the system is causal, h(n) = Z −1{ H(z) }. Furthermore, Assuming no multiple
poles, after partial fraction expansion,
N qk
H(z) = (32)
X
−1
k=0 (1 − pk z )

and hence
N
h(n) = qk (pk )n u(n). (33)
X

k=0
For multiple poles pl with multiplicity L, we have the term ql (pl )nnL u(n).
A BIBO system requires |h(n)| < ∞. This is satisfied if and only if |pk | < 1.
P

Dr. Dominic Ho 12
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

*Hence a LTI sytem is BIBO stable if and only if the ROC for H(z) contains the unit
circle (the poles of H(z) are all lie inside the unit circle).
Assuming all zero initial conditions, taking z-transform of
N M
y(n) = − ak y(n − k) + bk x(n − k) (34)
X X

k=1 k=0
gives
N M
−k
Y (z) = − ak z Y (z) + bk z −k X(z). (35)
X X

k=1 k=0

Hence PM −k
k=0 bk z
Y (z) = H(z)X(z) = X(z). (36)
1 + Nk=1 ak z
−k
P

When initial conditions are zero (i.e. y(n) = x(n) = 0 for n < 0), the system output with
arbitrary input x(n) can be found by
y(n) = Z −1{ H(z)X(z) }. (37)

Dr. Dominic Ho 13
Ch.4 Z-Transform and its Application to the Analysis of LTI System ECE4830/7830

Eg. Find the output of

y(n) = 0.9y(n − 1) − 0.81y(n − 2) + x(n) (38)
when x(n) = u(n) and y(−1) = y(−2) = 0.
Sol.

Dr. Dominic Ho 14