DIGITAL SIGNAL PROCESSING

LECTURE #13, 14, 15

13 – 15TH MARCH 2023
CHAPTER 3 – THE Z – TRANSFORM AND ITS APPLICATION
TO THE ANALYSIS OF LTI SYSTEMS

 (3.1) The z – transform

 (3.2) Properties of the z – transform

 (3.3) Rational z – transform

 (3.4) Inversion of the z – transform

 (3.5) Analysis of linear time-invariant systems in the z – domain

 (3.6) The one-sided z – transform

 3.1 THE Z – TRANSFORM

Motivation:

 Makes analysis of LTI systems easier (just as we learnt that solving difference equations for determining
output of infinite duration signals is easier than finding output through convolution).

 𝑧 – transform does not convert a time – domain signal to a frequency – domain signal. It is instead a
mathematical tool that helps analyze LTI systems.

 Unlike Fourier transform, 𝑧 – transform is applicable to unstable signals/systems too and hence helpful
for system designing i.e., controller designing for unstable systems.

 With 𝑧 – transform applied on a signal, the overall information remains intact/same, only domain (or
appearance) of the signal changes from 𝑛 – domain (real integral values) to 𝑧 – domain (complex values).
 3.1 THE Z – TRANSFORM
3.1.1 The direct z – transform
𝑧 – transform of a signal 𝑥 𝑛 is computed mathematically as:
∞

𝑋 𝑧 = ෍ 𝑥 𝑛 𝑧 −𝑛
𝑛=−∞
𝑧 – domain 𝑛 – domain
Understanding it graphically:

𝑛 – domain 𝑧 – domain
with real with
integral complex
values and values and a
a1-D signal. 2-D signal.
 3.1 THE Z – TRANSFORM

 During transformation from 𝑛 – domain (𝑥 𝑛 ) to 𝑧 – domain (𝑋 𝑧 ), individual samples are not

transformed to distinct 𝑧 – domain locations. Rather, the complete 𝑛 – domain signal, after the
mathematical process, is transformed to the 𝑧 – domain signal.

 Just like in time-domain, information of time instants and the corresponding amplitude values is needed to
fully define 𝑥 𝑛 , likewise for a 𝑧 – domain signal we need to consider 𝑋 𝑧 - the mathematical
expression, as well as the associated set of points (or region) for complete information of the transformed
signal.

1
𝑥𝑛 = 𝑎𝑛 , 𝑛 ≥0 𝑋 𝑧 = , 𝑧> 𝑎
1 − 𝑎𝑧 −1
 3.1 THE Z – TRANSFORM
Significance of 𝑧 𝑛 as an Eigen-function of an LTI system:
 Eigen functions are the signals which when given at the input of an LTI system, appear at output. There
can be, however, change in the amplitude and phase of input/output signals.
 Different Eigen-functions of LTI systems are:
 Complex exponential signals 𝑒 ±𝑗𝜔𝑛 , 𝑧 ±𝑛 .
 Sinusoidal signals cos 𝜔𝑛 , sin 𝜔𝑛 .
 Thus:
𝐿𝑇𝐼 𝑠𝑦𝑠𝑡𝑒𝑚
𝑟𝑥 𝑒 𝑗𝜃𝑥 𝑒 𝑗𝜔𝑛 = 𝑟𝑥 𝑒 𝑗 𝜔𝑛+ 𝜃𝑥
𝑟𝑦 𝑒 𝑗 𝜔𝑛+ 𝜃𝑦

𝐿𝑇𝐼 𝑠𝑦𝑠𝑡𝑒𝑚
𝑟𝑥 cos 𝜔𝑛 + 𝜃𝑥 𝑟𝑦 cos 𝜔𝑛 + 𝜃𝑦
 The change in amplitude and phase is caused by multiplication of the Eigen-function with the Eigen-value.
𝐿𝑇𝐼 𝑠𝑦𝑠𝑡𝑒𝑚
𝑥 𝑛 = EigenFunction 𝑦 𝑛 = (EigenValue) × (EigenFunction)
 3.1 THE Z – TRANSFORM
Significance of 𝑧 𝑛 as an Eigen-function of an LTI system:
 Also, the information on Eigen-value is determined/obtained from the characteristics of the underlying
system (i.e., it’s impulse response). And hence:
𝐿𝑇𝐼 𝑠𝑦𝑠𝑡𝑒𝑚 −1
𝑥 𝑛 = EigenFunction 𝑦 𝑛 = (EigenValue1 ) × (EigenFunction)
𝐿𝑇𝐼 𝑠𝑦𝑠𝑡𝑒𝑚 −2
𝑥 𝑛 = EigenFunction 𝑦 𝑛 = (EigenValue2 ) × (EigenFunction)
 In 𝑧 – domain, the Eigen-value is 𝐻 𝑧 and thus:
𝐿𝑇𝐼 𝑠𝑦𝑠𝑡𝑒𝑚
±𝑛
𝑥 𝑛 =𝑧 𝑦 𝑛 = 𝐻 𝑧 × 𝑥 𝑛 = 𝐻 𝑧 𝑧 ±𝑛
 We see that the complex procedures of convolution and difference equation solutions is transformed to a
simpler multiplication operation if signals are considered in 𝑧 – domain.
∞
 What if the input signal is not in the form of 𝑧 ±𝑛 ? 𝐻 𝑧 = ෍ ℎ 𝑛 𝑧 −𝑛
𝐿𝑇𝐼 𝑠𝑦𝑠𝑡𝑒𝑚
𝑛=−∞
𝑥 𝑛 ↔𝑋 𝑧 𝑌 𝑧 =𝐻 𝑧 𝑋 𝑧 ↔𝑦 𝑛
 3.1 THE Z – TRANSFORM
Summarizing:
➢ Similar to Fourier transform that transforms an aperiodic signal into unaccountably
infinite sinusoidal signals, 𝑧 – transform transforms a signal into unaccountably infinite
number of 𝑧 ±𝑛 signals.
➢ If 𝑥 𝑛 = 𝑧 ±𝑛 ,
𝐿𝑇𝐼 𝑠𝑦𝑠𝑡𝑒𝑚
𝑥𝑛 = 𝑧 ±𝑛 𝑦 𝑛 = 𝑧 ±𝑛 𝐻 𝑧
➢ If 𝑥 𝑛 ≠ 𝑧 ±𝑛 ,
∞
𝐿𝑇𝐼 𝑠𝑦𝑠𝑡𝑒𝑚 𝑧 −1
𝑋 𝑧 = ෍ 𝑥 𝑛 𝑧 −𝑛 𝑌 𝑧 =𝑋 𝑧 𝐻 𝑧 𝑦𝑛
𝑛=−∞
➢ The 𝑧 – transform expression 𝑋 𝑧 must be accompanied with the associated region,
called region of convergence (or RoC) to convey complete information of the transform-
ed signal.
 3.1 THE Z – TRANSFORM EXAMPLES
Example 1 (finite duration causal signal)
𝑥 𝑛 = 1, 2, 5, −1, 0, 7
∞

𝑋 𝑧 = ෍ 𝑥 𝑛 𝑧 −𝑛 = ⋯ + 𝑥 −2 𝑧 2 + 𝑥 −1 𝑧 1 + 𝑥 0 𝑧 0 + 𝑥 1 𝑧 −1 + 𝑥 2 𝑧 −2 + ⋯
𝑛=−∞

Since samples from −∞ to −1 and 6 to ∞ are zero and hence won't appear in the final expression.
𝑋 𝑧 = 1 + 2𝑧 −1 + 5𝑧 −2 − 𝑧 −3 + 7𝑧 −5
Example 2 (finite duration anti-causal signal)
0
𝑥 𝑛 = 1, 2, 5, −1, 0, 7,
↑
𝑋 𝑧 = 7𝑧 1 − 𝑧 3 + 5𝑧 4 + 2𝑧 5 + 𝑧 6
Example 3 (finite duration non-causal signal)
5
𝑥 𝑛 = 1, 2, , −1, 0, 7
↑
𝑋 𝑧 = 𝑧 2 + 2𝑧 + 5 − 𝑧 −1 + 7𝑧 −3
 3.1 THE Z – TRANSFORM EXAMPLES
The COMPLETE z – transform expressions shall be:

𝑋 𝑧 = 1 + 2𝑧 −1 + 5𝑧 −2 − 𝑧 −3 + 7𝑧 −5 , 𝑧≠0

𝑋 𝑧 = 7𝑧 1 − 𝑧 3 + 5𝑧 4 + 2𝑧 5 + 𝑧 6 , 𝑧≠∞

𝑋 𝑧 = 𝑧 2 + 2𝑧 + 5 − 𝑧 −1 + 7𝑧 −3 , 𝑧 ≠ 0, ∞
 3.1 THE Z – TRANSFORM EXAMPLES
Example 1 (infinite duration causal signal)
𝑥 𝑛 = 𝑎𝑛 𝑢 𝑛
∞ ∞ ∞
1
𝑋 𝑧 = ෍ 𝑎𝑛 𝑢 𝑛 𝑧 −𝑛 = ෍ 𝑎𝑛 𝑧 −𝑛 = ෍ 𝑎𝑧 −1 𝑛
= , 𝑎𝑧 −1 < 1 𝑧 >𝑎
1 − 𝑎𝑧 −1
𝑛=−∞ 𝑛=0 𝑛=0

Example 2 (infinite duration anti-causal signal)

𝑥 𝑛 = −𝑏𝑛 𝑢 −𝑛 − 1
∞ −1
1 𝑧 <𝑏
𝑋 𝑧 = ෍ −𝑏𝑛 𝑢 −𝑛 − 1 𝑧 −𝑛 =− ෍ 𝑏𝑛 𝑧 −𝑛 = , 𝑏 −1 𝑧 < 1
1 − 𝑏𝑧 −1
𝑛=−∞ 𝑛=−∞

Example 3 (infinite duration non-causal signal)

𝑥 𝑛 = 𝑎 𝑛 𝑢 𝑛 − 𝑏𝑛 𝑢 −𝑛 − 1
∞ −1
𝑎< 𝑧 <𝑏
𝑋 𝑧 = ෍ 𝑎𝑧 −1 𝑛
− ෍ 𝑏𝑧 −1 𝑛

𝑛=0 𝑛=−∞
 INFINITE DURATION NON-CAUSAL SIGNALS
Example 1
For the given 𝑥[𝑛], find 𝑋(𝑧).
𝑛 𝑛
1 3
𝑥𝑛 = 𝑢𝑛 − 𝑢 −𝑛 − 1
2 2
2 − 2𝑧 −1 1 3
𝑋 𝑧 = , < 𝑧 <
1 3 2 2
1 − 𝑧 −1 1 − 𝑧 −1
2 2
Example 2
For the given 𝑥[𝑛], find 𝑋(𝑧).
𝑛 𝑛
1 1
𝑥𝑛 = 𝑢𝑛 − 𝑢 −𝑛 − 1
2 4
𝑿(𝒛) does not exist!!!
 3.1 THE Z – TRANSFORM EXAMPLES
The COMPLETE z – transform expressions shall be:

1
𝑋 𝑧 = , 𝑧 >𝑎
1 − 𝑎𝑧 −1

1
𝑋 𝑧 = , 𝑧 <𝑏
1 − 𝑏𝑧 −1

1 1
𝑋 𝑧 = + , 𝑎< 𝑧 <𝑏
1 − 𝑎𝑧 −1 1 − 𝑏𝑧 −1
FURTHER FOR INFINITE DURATION SIGNALS …
For a sum of causal signals, such as For a sum of anti-causal signals, For a non-causal signal, such as
1 𝑛 1 𝑛 1 𝑛 1 𝑛
𝑥𝑛 = 𝑢 𝑛 ± 𝑢 𝑛, such as 𝑥 𝑛 = 𝑢 −𝑛 − 1 ± 𝑥𝑛 = 𝑢 −𝑛 − 1 ±
2 3 2 2
region of convergence would lie 1 𝑛 1 𝑛
𝑢 −𝑛 − 1 , region of 𝑢 𝑛 , region of convergence
OUTSIDE the BIGGEST circle. 3 3
convergence would lie INSIDE the would lie in a LOOP/RING
SMALLEST circle. shaped region.
FURTHER FOR INFINITE DURATION SIGNALS …
For a sum of causal signals, such as For a sum of anti-causal signals, For a non-causal signal, such as
1 𝑛 1 𝑛 1 𝑛 1 𝑛
𝑥𝑛 = 𝑢 𝑛 ± 𝑢 𝑛, such as 𝑥 𝑛 = 𝑢 −𝑛 − 1 ± 𝑥𝑛 = 𝑢 −𝑛 − 1 ±
2 3 2 3
region of convergence would lie 1 𝑛 1 𝑛
𝑢 −𝑛 − 1 , region of 𝑢 𝑛 , region of convergence
OUTSIDE the BIGGEST circle. 3 2
convergence would lie INSIDE the would be as follows and hence its
SMALLEST circle. 𝑿(𝒛) won't exist.
 THINGS TO INVESTIGATE:

End problems: 1, 2, 3, 4, 5, 6.