Chapter 2

Z - transform
 The z-Transform
 The z-transform is a generalization of the DTFT and it is applicable to a broader class of signals and
systems.
 Definition
 Direct transform

 Inverse transform

 The z-transform has a region of convergence (ROC), which is the values of z for which the infinite
sum in the direct transform is finite. Without the ROC, the z-transform is an ambiguous
representation of a signal.
 The inverse transform is given by the contour integral over some complex path C. This is generally
laborious, so we'll obtain the inverse z-transform through indirect methods such as partial fraction
expansion.
2
The z-Transform
 x[n] = 𝑧𝑧 𝑛𝑛 , 𝑧𝑧 ∈ ℂ is another eigenfunction of LTI systems
 Proof:
 from convolution sum

 Rearranging

 By definition
 We can always write 𝑧𝑧 𝑛𝑛 = 𝑟𝑟 𝑛𝑛 𝑒𝑒 𝑗𝑗𝜔𝜔𝑛𝑛 (polar coordinates)
 The eigenfunction 𝑒𝑒 𝑗𝑗𝜔𝜔𝑛𝑛 , used in the DTFT, is just a particular case (r = 1) of the
eigenfunction used in the z-transform
 The factor 𝑟𝑟 𝑛𝑛 helps the z-transform sum converge to a broader class of signals 3
Relation between DTFT and the z-Transform
 DTFT z-Transform
 Direct transform

4
Relation between DTFT and the z-Transform
 DTFT z-Transform
 Direct transform

 The DTFT is equal to the z-transform evaluated on the unit circle DTFT is periodic with
period 2π.
 Question: Does the DTFT exist if the ROC of the z-transform does not include the unit circle?

5
One sided or unilateral Z-transform

6
No uniform convergence
No Z - transform

7
Right – sided exp.

8
Left –sided exp.

9
Sum of two exps.

10
11
Two – sided exp.

12
Bounded sequences

 ROC?

13
Finite Length Truncated exp.

14
15
16
Figure 3.8 Examples of four z -transforms with the
same pole–zero locations, illustrating the different
possibilities for the ROC, each of which
corresponds to a different sequence: (b) to a right-
sided sequence, (c) to a left-sided sequence, (d) to
a two-sided sequence, and (e) to a two-sided 17
sequence.
Stability, Causality and ROC

18
For what will we use the z-transform?
 Representing LTI systems
 Determining stability of LTI systems
 Solving difference equations

19
 Rational z-transforms
 Rational z-transforms are a ratio of two polynomials in 𝑍𝑍 −1 (or z)

 The zeros of a z-transform are the values of z for which 𝑋𝑋 𝑧𝑧 = 0

 The poles of a z-transform are the values of z for which 𝑋𝑋 𝑧𝑧 = ∞. By definition, the ROC cannot
contain any pole
 𝑋𝑋(𝑧𝑧)has M zeros 𝑧𝑧1 , … , 𝑧𝑧𝑀𝑀 , which are the roots of the numerator polynomial B(𝑧𝑧)
 𝑋𝑋(𝑧𝑧) has N poles 𝑝𝑝1 , … , 𝑝𝑝𝑁𝑁 , which are the roots of the denominator polynomial A(𝑧𝑧)
 If N − M > 0, 𝑋𝑋(𝑧𝑧) has more 𝑁𝑁 − 𝑀𝑀zeros at z = 0
 If N − M < 0, 𝑋𝑋(𝑧𝑧) has more 𝑁𝑁 − 𝑀𝑀 poles at z = 0
 If the coefficients 𝑏𝑏0 , … , 𝑏𝑏𝑀𝑀 , {𝑎𝑎0 , … , 𝑎𝑎𝑁𝑁 }are real, the poles and zeros are either real or they appear
in complex conjugate pairs 20
Examples
 Right-sided exponential x[n] = 𝑎𝑎𝑛𝑛 𝑢𝑢[𝑛𝑛]

 This sum converges only if 𝑎𝑎𝑧𝑧 −1 < 1. Hence, ROC: 𝑧𝑧 > |𝑎𝑎|
 From the sum of an infinite geometric progression:

 Pole at z = a and zero at z = 0.

 The ROC of a causal signal is the exterior of a circle
 whose radius is the magnitude of the outermost pole |a|.
21
Examples
 Left-sided exponential x[n] = −𝑎𝑎𝑛𝑛 𝑢𝑢[−𝑛𝑛 − 1]

 This sum converges only if 𝑧𝑧𝑎𝑎−1 < 1. Hence, ROC: 𝑧𝑧 < |𝑎𝑎|
 Once again from the sum of an infinite geometric progression:

 Without the ROC, the z-transform is an ambiguous representation of

a signal.
 The ROC of an anti-causal signal is the interior of a circle whose
radius is the magnitude of the innermost pole |a|. 22
Properties of the ROC
 The ROC tells a lot about a signal or system
 1. If the ROC is the exterior of a circle the system/signal is causal, where a is the outermost
pole. The ROC cannot contain any pole
 2. If a LTI system is BIBO stable the ROC must contain the unit circle.
 Proof:

 3. A causal and stable LTI system has all poles inside the unit circle. This follows from the first two properties.

23
Inverse Z-transform
 Inspection method

24
Partial fraction estimation

25
Example

26
27
If M>N

28
With multiple order poles

29
Example

30
31
Power series expansion

 Example

32
Example

33
Example

34
Properties of Z transform

35
 IMPORTANT EXAMPLES IN THE BOOK!!!

36
Linear constant coefficient difference equation
 Many practical problems appear in the form of difference equations

 This difference equation denes an LTI system only if:

 1. All coefficients ak; bk are constant
 2. The initial conditions (or rest conditions) are zero 𝑦𝑦 −𝑁𝑁 = 𝑦𝑦 −𝑁𝑁 + 1 = ⋯ = 𝑦𝑦[−1] = 0
 Applying the linearity and time-shift properties of the z-transform:

 H(z) is a rational z-transform.

37
Example: first-order system
 First-order system:

 Calculating the z-transform:

 This corresponds to the exponential:

 Questions: Why the causal exponential? For what values of a is this system stable?
 This system is autoregressive i.e., the present output depends on previous outputs
 Autoregressive systems have infinite impulse response (IIR)
 Systems with rational z-transforms with non-zero poles are IIR
38
 In Matlab:
 > freqz(1; [1,-0.5])

39
Example: moving average system
 Moving average:

 Calculating the z-transform:

 This system has impulse response

 The impulse response only depends on a finite number of previous inputs. Hence, this system has a finite
impulse response (FIR)

40
 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑀𝑀 = 4, 𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑦𝑦[𝑛𝑛] = 1/4 (𝑥𝑥[𝑛𝑛] + 𝑥𝑥[𝑛𝑛 − 1] + 𝑥𝑥[𝑛𝑛 − 2] + 𝑥𝑥[𝑛𝑛 − 3])
1 −1 −3 𝑗𝑗𝜔𝜔 𝑗𝑗𝜔𝜔 1
 𝐻𝐻(𝑧𝑧) = (1 + 𝑧𝑧 + ⋯ + 𝑧𝑧 ) → (𝑧𝑧 = 𝑒𝑒 ) → 𝐻𝐻(𝑒𝑒 ) = (1 + 𝑒𝑒 −𝑗𝑗𝜔𝜔 + … + 𝑒𝑒 −3𝑗𝑗𝜔𝜔 )
4 4
 In Matlab:
 > freqz([1; 1; 1; 1]/4,1, 1)

41
Example: output of a moving average filter
 Suppose the input signal frequency is 𝜔𝜔0 = 2𝜋𝜋

42
Example: output of a moving average filter
 Suppose the input signal frequency is 𝜔𝜔0 = 0.5𝜋𝜋
 What happened?

43
 Systems can be linear, time-invariant, memoryless, causal, and stable

 Linear and time-invariant (LTI) systems are completely characterized by their impulse response
 We can use the convolution sum to calculate the output of an LTI system to any signal
 The complex exponential 𝑒𝑒 𝑗𝑗𝜔𝜔𝑛𝑛 , and more generally 𝑧𝑧 𝑛𝑛 , are eigenfunctions of LTI systems
 Frequency-domain representation of signals and systems
 Discrete-time Fourier transform (DTFT)
 z-transform and ROC. Without specifying the ROC the z-transform is ambiguous
 The DTFT is equivalent to the z-transform evaluated on the unit circle:
 𝐻𝐻(𝑒𝑒 𝑗𝑗𝜔𝜔 ) = 𝐻𝐻(𝑧𝑧 = 𝑒𝑒 𝑗𝑗𝜔𝜔 ), provided that the unit circle is in the ROC of the z-transform

44