Lecture 6: Inverse z-Transform

ECE 153/250: DIGITAL SIGNAL PROCESSING

REF: 3.3;3.5
Review
❑ z-transform
➢ Definition
◦ 𝑋 𝑧 = σ∞
𝑛=−∞ 𝑥 𝑛 𝑧
−𝑛

➢ Connection to DTFT
◦ 𝑧 = 𝑒𝑗
➢ ROC
◦ Set of values of 𝑧 such that z-transform series converges
◦ σ∞𝑛=−∞ 𝑥 𝑛 𝑧
−𝑛 < ∞

➢ Left and right-sided sequences

◦ Need specifying both algebraic expression and ROC
➢ Properties of ROC of z-Transform
◦ Right-sided sequence: ROC is outside the outermost pole
◦ Left-sided sequence: ROC is inside the innermost pole
◦ Two-sided sequence: ROC is a ring
◦ Finite-length sequence: ROC is entire z-plane, except possibly 𝑧 = 0 or 𝑧 = ∞
➢ Stability and Causality

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 2

Properties of z-Transform
❑ Linearity
➢ If 𝑥1 𝑛 𝑋1 𝑧 with ROC 𝑅1 and 𝑥2 𝑛 𝑋2 𝑧 with ROC 𝑅2
➢ Then 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 𝑎𝑋1 𝑧 + 𝑏𝑋2 𝑧 with ROC 𝑅 ⊃ 𝑅1 ∩ 𝑅2
❑ Time shifting
➢ If 𝑥 𝑛 𝑋 𝑧 with ROC 𝑅
➢ Then 𝑥 𝑛 − 𝑑 𝑧 −𝑑 𝑋 𝑧 with ROC 𝑅 (except with possible addition or
deletion of 0 and ∞)

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 3

Properties of z-Transform
❑ Multiplication by an Exponential Sequence
➢ If 𝑥 𝑛 𝑋 𝑧 with ROC 𝑅
➢ Then 𝑧0𝑛 𝑥 𝑛 𝑋 𝑧/𝑧0 , ROC = |𝑧0 |𝑅
❑ Time reversal
➢ If 𝑥 𝑛 𝑋 𝑧 with ROC 𝑅
➢ 𝑥 −𝑛 𝑋 1/𝑧 with ROC 1/𝑅

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 4

Properties of z-Transform
❑ Differentiation of 𝑋 𝑧
➢ If 𝑥 𝑛 𝑋 𝑧 with ROC 𝑅
𝑑𝑋 𝑧
➢ Then 𝑛𝑥 𝑛 −𝑧 , ROC = 𝑅
𝑑𝑧

❑ Conjugation of a complex sequence

➢ If 𝑥 𝑛 𝑋 𝑧 with ROC 𝑅
➢ Then 𝑥 ∗ 𝑛 𝑋 ∗ (𝑧 ∗ ), ROC = 𝑅

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 5

Properties of z-Transform
❑ Convolution of Sequences
➢ If 𝑥1 𝑛 𝑋1 𝑧 with ROC 𝑅1 and 𝑥2 𝑛 𝑋2 𝑧 with ROC 𝑅2
➢ Then 𝑥1 𝑛 ∗ 𝑥2 [𝑛] 𝑋1 𝑧 𝑋2 (𝑧) with ROC 𝑅 ⊃ 𝑅1 ∩ 𝑅2

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 6

Examples
❑ 𝑥 𝑛 =𝛿 𝑛−5
➢ 𝛿𝑛 1⇒𝛿 𝑛−5 𝑧 −5 , 𝑧 ≠ 0
❑ 𝑥 𝑛 = 3𝑛 𝑢 𝑛 − 1
𝑧 1 3
➢ 3𝑛 𝑢 𝑛 ⇒ 3𝑛−1 𝑢 𝑛 − 1 ⇒ 3𝑛 𝑢 𝑛 − 1 , 𝑧 >3
𝑧−3 𝑧−3 𝑧−3

❑ 𝑥 𝑛 = (𝑛 − 1)𝑢 𝑛 − 2
𝑧−1
𝑧 𝑧 −1 𝑑 2𝑧−1
𝑧−1
➢ 𝑢𝑛 ⇒ 𝑢𝑛 −2 ⇒ 𝑛𝑢 𝑛 − 2 −𝑧 =
𝑧−1 𝑧−1 𝑑𝑧 𝑧(𝑧−1)
2𝑧−1 𝑧 −1 1
➢ 𝑛−1 𝑢 𝑛−2 − = , 𝑧 >1
𝑧 𝑧−1 𝑧−1 𝑧−1 2

❑ 𝑥 𝑛 = 𝑛 0.5 𝑛 𝑢 𝑛
𝑧
𝑧 𝑑 𝑧−0.5 0.5𝑧
➢ 0.5 𝑛 𝑢 𝑛 ⇒ 𝑛 0.5 𝑛 𝑢 𝑛 −𝑧 = , 𝑧 > 0.5
𝑧−0.5 𝑑𝑧 𝑧−0.5 2

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 7

z-transforms for common sequences

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 8

Inverse z-Transform
❑ Inverse z transform corresponds to finding 𝑥[𝑛] given 𝑋(𝑧) and its ROC
❑ The z transform and inverse z transform are one-to-one mapping
provided that the ROC is given
𝑥𝑛 𝑋(𝑧)
❑ Four commonly used methods
➢ Inspection
➢ Partial Fraction Expansion
➢ Power Series Expansion
➢ Cauchy Integral Theorem
1
◦ 𝑥 𝑛 = 2𝜋𝑗 ‫𝑛 𝑧 𝑧 𝑋 ׯ‬−1 𝑑𝑧
◦ Integration over a closed contour within the ROC of the z-transform

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 9

Inspection
❑ When we are familiar with certain transform pairs, we can do the
inverse transform by inspection.
❑ Example
➢ Determine 𝑥[𝑛], which has the z-transform given by
𝑧
𝑋 𝑧 = , |𝑧| > 0.5
2𝑧−1
➢ We can rewrite 𝑋 𝑧 as
0.5
𝑋 𝑧 =
1−0.5𝑧 −1
1
➢ Recall that 𝑎𝑛 𝑢 𝑛 , |𝑧| > 𝑎
1−𝑎𝑧 −1
➢ 𝑥 𝑛 = 0.5 0.5 𝑛 𝑢 𝑛 = 0.5 𝑛+1 𝑢 𝑛

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 10

Partial Fraction Expansion
❑ 𝑋(𝑧) is a rational function in 𝑧 −1
σ𝑀
𝑘=0 𝑏𝑘 𝑧
−𝑘
➢ 𝑋 𝑧 = σ𝑁 −𝑘
𝑘=0 𝑎𝑘 𝑧

❑ Poles and zeros

➢ multiply 𝑧 𝑀+𝑁 to both numerator and denominator
𝑧 𝑁 σ𝑀
𝑘=0 𝑏𝑘 𝑧
𝑀−𝑘
◦ 𝑋 𝑧 =
𝑧 𝑀 σ𝑁
𝑘=0 𝑎𝑘 𝑧
𝑁−𝑘

➢ 𝑀 zeros and 𝑁 poles at nonzero locations

❑Toobtain the partial fraction expansion, we need to determine the
nonzero poles: 𝑑1 , 𝑑2 , ⋯ 𝑑𝑁

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 11

Case 1: 𝑀 < 𝑁 and all poles are of first order
❑ For first order poles,𝑑1 , 𝑑2 , ⋯ 𝑑𝑁 are distinct.
𝐴𝑘
❑ 𝑋 𝑧 = σ𝑁
𝑘=1 1−𝑑𝑘 𝑧 −1

❑ 𝐴𝑘 = 1 − 𝑑𝑘 𝑧 −1 𝑋 𝑧 |𝑧=𝑑𝑘

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 12

Example
❑ Determine 𝑥[𝑛], which has the z-transform given by:
1+0.1𝑧 −1
𝑋 𝑧 = − , 0.8 < |𝑧| < 1.25
1−2.05𝑧 −1 +𝑧 −2
❑ We first multiply 𝑧 2 to both numerator and denominator:
𝑧(𝑧+0.1)
𝑋 𝑧 = − 2
𝑧 −2.05𝑧+1
➢ There are two zeros at 𝑧 = 0 and 𝑧 = −0.1
➢ There are two poles at 𝑧 = 0.8 and 𝑧 = 1.25

❑ Rewrite 𝑋 𝑧 as:
𝐴1 𝐴2
𝑋 𝑧 = +
1−0.8𝑧 −1 1−1.25𝑧 −1

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 13

Example
❑ 𝐴1 = 1 − 0.8𝑧 −1 𝑋 𝑧 |𝑧=0.8 = 2
❑ 𝐴2 = 1 − 1.25𝑧 −1 𝑋 𝑧 |𝑧=1.25 = −3
2 3
❑ Thus, 𝑋 𝑧 = −
1−0.8𝑧 −1 1−1.25𝑧 −1
❑ 𝑥 𝑛 =2 0.8 𝑛 𝑢 𝑛 + 3 1.25 𝑛 𝑢[−𝑛 − 1]

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 14

Case 2: 𝑀 ≥ 𝑁 and all poles are of first order
❑ In this case, can be expressed as:
𝐴𝑘
𝑋 𝑧 = σ𝑀−𝑁
𝑙=0 𝐵𝑙 𝑧 −𝑙
+ σ𝑁
𝑘=1 1−𝑑𝑘 𝑧 −1

❑𝐵𝑙 are obtained by long division of the numerator by the

denominator

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 15

Example
❑ Determine which has transform of the form
2
1+2𝑧 −1 +𝑧 −2 1+𝑧 −1
𝑋 𝑧 = = , |𝑧| > 1
1−1.5𝑧 −1 +0.5𝑧 −2 (1−0.5𝑧 −1 )(1−𝑧 −1 )
❑ 𝑋 𝑧 can be expressed as
𝐴1 𝐴2
𝑋 𝑧 = 𝐵0 + +
1−0.5𝑧 −1 1−𝑧 −1
➢ 𝐵0 = 2
−1+5𝑧 −1
➢ 𝑋 𝑧 =2+
(1−0.5𝑧 −1 )(1−𝑧 −1 )
➢ 𝐴1 = −9; 𝐴2 = 8
9 8
➢𝑋 𝑧 =2− +
1−0.5𝑧 −1 1−𝑧 −1
1 𝑛
❑ Thus 𝑥 𝑛 = 2𝛿 𝑛 − 9 𝑢 𝑛 + 8𝑢[𝑛]
2

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 16

Case 3: 𝑀 < 𝑁 with multiple-order pole(s)
❑ In this case
𝐴𝑘 𝐶𝑚
𝑋 𝑧 = σ𝑁
𝑘=1,𝑘≠𝑖 + σ𝑠𝑚=1
1−𝑑𝑘 𝑧 −1 1−𝑑𝑖 𝑧 −1 𝑚

❑ 𝐶𝑚 can be determined by
1 𝑑 𝑠−𝑚
➢ 𝐶𝑚 = 1 − 𝑑𝑖 𝑤 𝑠 𝑋 𝑤 −1 |𝑤=𝑑−1
𝑠−𝑚 ! −𝑑𝑖 𝑠−𝑚 𝑑𝑤 𝑠−𝑚 𝑖

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 17

Example
❑ Determine 𝑥[𝑛], which has the z-transform given by:
4
𝑋 𝑧 =
1+𝑧 −1 1−𝑧 −1 2
❑ a first order pole at 𝑧 = −1 and a second order pole at 𝑧 = 1
𝐴 𝐶1 𝐶2
𝑋 𝑧 = −1
+ −1
+
1+𝑧 1−𝑧 1 − 𝑧 −1 2
4
➢𝐴 = | =1
1−𝑧 −1 2 𝑧=−1
1 𝑑 2 4
➢ 𝐶1 = 1 − 1𝑤 |
2−1 ! −1 2−1 𝑑𝑤 1+𝑤 1−𝑤 2 𝑤=1
𝑑 4 4
=− |𝑤=1 = | =1
𝑑𝑤 1+𝑤 1+w 2 𝑤=1
1 2 4 4
➢ 𝐶2 = 1 − 1𝑤 | = | =2
2−2 ! −1 2−2 1+𝑤 1−𝑤 2 𝑤=1 1+𝑤 𝑤=1
1 1 2
❑𝑋 𝑧 = + +
1+𝑧 −1 1−𝑧 −1 1−𝑧 −1 2

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 18

Case 4: 𝑀 ≥ 𝑁 with multiple-order pole(s)
❑ In this case, 𝑋 𝑧 can be represented as
𝑁 𝐴𝑘 𝐶𝑚
𝑋 𝑧 = σ𝑀−𝑁
𝑙=0 𝐵𝑙 𝑧 −𝑙
+ σ𝑘=1,𝑘≠𝑖 + σ𝑠𝑚=1
1−𝑑𝑘 𝑧 −1 1−𝑑𝑖 𝑧 −1 𝑚
➢ 𝐴𝑘 can be calculated based on Case I
➢ 𝐵𝑙 can be calculated based on Case 2
➢ 𝐶𝑚 can be calculated based on Case 3

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 19

Power Series Expansion
❑ When 𝑋 𝑧 is expanded as power series according to
𝑋 𝑧 = σ∞
𝑛=−∞ 𝑥 𝑛 𝑧 −𝑛

❑𝑥 𝑛 can be determined by finding the coefficients.

❑ Example:
➢𝑋 𝑧 = 2𝑧 2 1 − 0.5𝑧 −1 1 + 𝑧 −1 1 − 𝑧 −1
➢ Expanding 𝑋 𝑧 yields 𝑋 𝑧 = 2𝑧 2 − 𝑧 − 2 − 𝑧 −1
➢ Thus 𝑥 𝑛 = 2𝛿 𝑛 + 2 − 𝛿 𝑛 + 1 − 2𝛿 𝑛 + 𝛿[𝑛 − 1]

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 20

Example: Inverse Z-Transform via Power Series Expansion

❑Given:
➢𝑋 𝑧 = log(1 + 𝑎𝑧 −1 ) , 𝑧 > 𝑎
❑ Use power series expansion for log(1 + 𝜆)
−1 𝑛+1 𝜆𝑛
➢ log(1 + 𝜆) = σ∞𝑛=1 , 𝜆 <1
𝑛
➢ Apply with 𝜆 = 𝑎𝑧 −1
∞ −1 𝑛+1 𝑎𝑛 𝑧 −𝑛
◦ Thus log(1 + 𝑎𝑧 −1 ) = σ𝑛=1
𝑛

−1 𝑛+1 𝑎𝑛
❑ 𝑥[𝑛] is deduced as 𝑥 𝑛 = 𝑢[𝑛 − 1]
𝑛

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 21

Transfer Function of Linear Time-Invariant System
❑ An LTI system can be characterized by the difference equation
➢ σ𝑁
𝑘=0 𝑎𝑘 𝑦 𝑛 − 𝑘 = σ 𝑀
𝑚=0 𝑏𝑚 𝑥 𝑛 − 𝑚

❑ Apply z transform:
➢ 𝑌(𝑧) σ𝑁 𝑎
𝑘=0 𝑘 𝑧 −𝑘 = 𝑋(𝑧) σ𝑀 𝑏
𝑚=0 𝑚 𝑧 −𝑚

❑ Define the transfer function

𝑌 𝑧 σ𝑀
𝑚=0 𝑏𝑚 𝑧
−𝑚 𝑏0 ς𝑀
𝑘=1 1−𝑐𝑘 𝑧
−1
➢ 𝐻 𝑧 = = σ𝑁
=
𝑋 𝑧 𝑘=0 𝑎𝑘 𝑧
−𝑘 𝑎0 ς𝑁
𝑘=1 1−𝑑𝑘 𝑧
−1

➢ Zeros: 𝑐𝑘 and Poles: 𝑑𝑘

➢ 𝐻 𝑧 describes the input-output relationship of the system in the z-domain.
➢ Frequency response: 𝐻 𝑒 𝑗𝜔 = 𝐻 𝑧 |𝑧=𝑒 𝑗𝜔

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 22

Example
❑Question: Consider an LTI system that is stable and for which
𝐻(𝑧), the z-transform of the impulse response, is given by 𝐻 𝑧 =
3
1 −1 , Determine the output for unit step sequence.
1+3𝑧

❑ Solution:
3 9
3 1 4 4
➢𝑌 𝑧 =𝐻 𝑧 𝑋 𝑧 = 1 = 1 −1 +
1+ 𝑧 −1 1−z−1
3
1+3𝑧 1−z−1
3 1 𝑛 9
➢ 𝑦𝑛 = − 𝑢 𝑛 + 𝑢[𝑛]
4 3 4

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 23

FIR and IIR
❑ Rational system function
𝑏0 ς𝑀
𝑘=1 1−𝑐𝑘 𝑧
−1
𝑁 𝐴𝑘
➢ 𝐻 𝑧 = = σ𝑀−𝑁
𝑟=0 𝐵𝑟 𝑧
−𝑟
+ σ𝑘=0
𝑎0 ς𝑁
𝑘=1 1−𝑑𝑘 𝑧
−1 1−𝑑 𝑧 −1 𝑘

❑ Impulse response
𝑁 𝑛
➢ ℎ 𝑛 = σ𝑀−𝑁
𝑟=0 𝐵𝑟 𝛿[𝑛 − 𝑟] + σ 𝑘=0 𝑘 𝑘 𝑢[𝑛]
𝐴 𝑑
❑ IIR: at least one pole does not cancel with a zero
➢ impulse response will be infinite
❑ FIR: transfer function does not have any poles except at z=0
➢ impulse response ℎ[𝑛] = σ𝑀 𝑘=0 𝑏𝑘 𝛿[𝑛 − 𝑘]
➢ Impulse response is of finite length

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 24

Example of IIR System
❑A causal system with following equation
𝑦 𝑛 − 𝑎𝑦 𝑛 − 1 = 𝑥[𝑛]
1
➢ Transfer function: 𝐻 𝑧 = ROC 𝑧 >𝑎
1−𝑎𝑧 −1
➢ Impulse response: ℎ 𝑛 = 𝑎𝑛 𝑢[𝑛]

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 25

Example of FIR System
❑ Consider the following impulse response
𝑎𝑛 , 0 ≤ 𝑛 ≤ 𝑀
ℎ 𝑛 =ቊ
0, otherwise
1−𝑎𝑀+1 𝑧 −𝑀−1
➢ Transfer function 𝐻 𝑧 = σ𝑀
𝑛=0 ℎ 𝑛 𝑧 −𝑛 = σ𝑀 𝑛 −𝑛
𝑛=0 𝑎 𝑧 =
1−𝑎𝑧 −1
𝑗2𝜋𝑘
➢ Zeros: 𝑧𝑘 = 𝑎𝑒 𝑀+1

➢ Difference equation:
𝑦 𝑛 = σ𝑀 𝑘
𝑛=0 𝑎 𝑥[𝑛 − 𝑘]

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 26

Stability and Causality
❑A system function does not uniquely specify a system
➢ Need to know the ROC
❑ Properties of system gives clues about the ROC
➢ Causal systems must be right sided
➢ Stability implies that the ROC includes the unit circle

❑ Causal AND stable systems

➢ Causal: the ROC is outside outermost pole
➢ Stable: unit circle included in ROC
➢ This means outermost pole is inside unit circle
➢ Hence all poles are inside unit circle

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 27

Summary: contrast with continuous-time systems

Lecture 6: Inverse z-Transform ECE 153/250 DIGITAL SIGNAL PROCESSING 28