23ECE112

Signal Processing I
Latha Srinivasan
Assistant Professor
Department of ECE
ASE, Bengaluru
Z transform
• Used to represent and analyse discrete time signals and systems.

• For a given signal x[n], the Z transform is defined as 𝑋 𝑧 = σ∞

𝑛=−∞ 𝑥[𝑛]𝑧
−𝑛
where z is a complex number
represented by 𝑧 = 𝑟𝑒 𝑗Ω . r=|z| and Ω = ∠𝑧.

• One can also define a region of convergence for the Z transform

• ROC in Z transform is the range of values of z for which Z transform converges.

• We need to specify the ROC along with the X(z) to make the Z transform representation unique for a
particular signal
 Relation between Z transform and DTFT
• By definition, DTFT of x[n], 𝑋 Ω = σ∞
𝑛=−∞ 𝑥 𝑛 𝑒
−𝑗Ω𝑛

• Z-transform of x[n], 𝑋 𝑧 = σ∞
𝑛=−∞ 𝑥 𝑛 𝑧
−𝑛

• If we substitute 𝑧 = 𝑒 𝑗Ω in the Z transform equation, then it becomes equivalent to the DTFT

• But for Z transform to converge 𝑧 = 𝑒 𝑗Ω has to be in the ROC of Z-transform. Only then we can substitute and
obtain the DTFT

• 𝑋 𝑧 |𝑧=𝑒 𝑗Ω = 𝑋(Ω)

• 𝑧 = 𝑒 𝑗Ω has to be in the ROC of Z-transform implies that ROC should contain the unit circle.
 Relation between Z transform and DTFT

• | σ∞
𝑛=−∞ 𝑥 𝑛 𝑒
−𝑗Ω𝑛
| ≤ σ∞
𝑛=−∞ 𝑥 𝑛 |𝑒
−𝑗Ω𝑛
|

• | σ∞
𝑛=−∞ 𝑥 𝑛 𝑒
−𝑗Ω𝑛
| ≤ σ∞
𝑛=−∞ 𝑥 𝑛

• For the DTFT to converge | σ∞

𝑛=−∞ 𝑥 𝑛 𝑒
−𝑗Ω𝑛 | < ∞

• That implies σ∞
𝑛=−∞ 𝑥 𝑛 <∞

• If x[n] is absolutely summable Z transform evaluated at 𝑧 = 𝑒 𝑗Ω will also be finite

• That means ROC will contain 𝑧 = 𝑒 𝑗Ω (unit circle)

 Numerical Example

• Consider 𝑥 𝑛 = 𝑎𝑛 𝑢[𝑛]

• 𝑋 𝑧 = σ∞
𝑛=−∞ 𝑥[𝑛]𝑧
−𝑛
= σ∞ 𝑛
𝑛=−∞ 𝑎 𝑢[𝑛]𝑧
−𝑛

𝑎 𝑛
• 𝑋 𝑧 = σ∞ 𝑛 −𝑛
𝑛=0 𝑎 𝑧
∞
= σ𝑛=0 𝑧

1 𝑧
• 𝑋 𝑧 = 𝑎 = 𝑧−𝑎
1− 𝑧

𝑎
• This is possible only if |𝑧 | <1

• Or in other words |z|>|a|. ROC is |z|>|a|

• There is zero at z=0 and pole at z=a.

 Numerical Example

• Consider 𝑥 𝑛 = −𝑎𝑛 𝑢[−𝑛 − 1]

• 𝑋 𝑧 = σ∞
𝑛=−∞ 𝑥[𝑛]𝑧
−𝑛
= − σ∞ 𝑛
𝑛=−∞ 𝑎 𝑢[−𝑛 − 1]𝑧
−𝑛

∞ 𝑧 𝑚
• 𝑋 𝑧 = − σ−1 𝑛 −𝑛
𝑛=−∞ 𝑎 𝑧 =− σ𝑚=1
𝑎

1 𝑎 𝑧
• 𝑋 𝑧 =− 𝑧 −1 =− − 1 = (𝑧−𝑎)
1−𝑎 𝑎−𝑧

𝑧
• This is possible only if |𝑎 | <1

• Or in other words |z|<|a|. ROC is |z|<|a|

• Both signals 𝑎𝑛 𝑢[𝑛] and −𝑎𝑛 𝑢[−𝑛 − 1] have the same form of X(z) but ROC is different.
Numerical Example
• Find Z transform and ROC for 𝑥[𝑛] = 𝛿[𝑛]

• 𝑋 𝑧 = σ∞
𝑛=−∞ 𝑥[𝑛]𝑧
−𝑛
= σ∞
𝑛=−∞ 𝛿[𝑛]𝑧
−𝑛

• 𝑋 𝑧 = 1𝑧 −0 = 1

• 𝑋 𝑧 = 1 for all values of z except at |z|=0 and ∞

• Here the ROC is the 0<|z|< ∞.

Numerical Example
• Find Z transform and ROC for 𝑥[𝑛] = 𝑢[𝑛]

• 𝑋 𝑧 = σ∞
𝑛=−∞ 𝑥[𝑛]𝑧
−𝑛
= σ∞
𝑛=−∞ 𝑢[𝑛]𝑧
−𝑛

1 𝑧
• 𝑋 𝑧 = σ∞
𝑛=0 𝑧
−𝑛 = σ∞ (𝑧 −1 )𝑛 =
𝑛=0 = 𝑧−1
1−𝑧 −1

• This is possible only if |𝑧 −1 | <1

• Or in other words |z|>1. ROC is |z|>1

Numerical Example
• Consider 𝑥 𝑛 = 𝑐𝑜𝑠 Ω0 𝑛 𝑢 𝑛

• 𝑋 𝑧 = σ∞
𝑛=−∞ 𝑥[𝑛]𝑧
−𝑛
= σ∞
𝑛=−∞ 𝑐𝑜𝑠 Ω0 𝑛 𝑢 𝑛 𝑧
−𝑛

1
• 𝑋 𝑧 = σ∞
𝑛=0 𝑐𝑜𝑠 Ω0 𝑛 𝑧
−𝑛
= σ∞
𝑛=0(𝑒
𝑗Ω0 𝑛
+ 𝑒 −𝑗Ω0 𝑛 ) 𝑧 −𝑛
2

𝑛 𝑛
1 ∞ 𝑗Ω0 𝑛 −𝑛 −𝑗Ω0 𝑛 −𝑛 1 ∞ 𝑒 𝑗Ω0 ∞ 1
• 𝑋 𝑧 = σ 𝑒 𝑧 +𝑒 𝑧 = (σ𝑛=0 + σ𝑛=0 𝑗Ω0 )
2 𝑛=0 2 𝑧 𝑒 𝑧

1 1 1 1 1−𝑒 −𝑗Ω0 𝑧 −1 1−𝑒 𝑗Ω0 𝑧 −1 1 2−2cos Ω0 𝑧 −1 1−cos Ω0 𝑧 −1

• 𝑋 𝑧 = + = + = =
2 1−𝑒 𝑗Ω0 𝑧 −1 1−𝑒 −𝑗Ω0 𝑧 −1 2 (1−𝑒 𝑗Ω0 𝑧 −1 )(1−𝑒 −𝑗Ω0 𝑧 −1 ) (1−𝑒 𝑗Ω0 𝑧 −1 )(1−𝑒 −𝑗Ω0 𝑧 −1 ) 2 1−2 cos Ω0 𝑧 −1 + 𝑧 −2 1−2 cos Ω0 𝑧 −1 + 𝑧 −2

𝑧 2 −cos Ω0 𝑧
• 𝑋 𝑧 =
𝑧 2 −2 cos Ω0 𝑧+1

• There are 2 poles 𝑝1 = 𝑒 𝑗Ω0 and 𝑝2 = 𝑒 −𝑗Ω0

• ROC is of the form |z|> |𝑒 𝑗Ω0 , |𝑒 −𝑗Ω0 . i.e, |z|>1

Properties of Z transform

𝑧 𝑋(𝑧)
𝑌 𝑧 =
𝑧−1
𝑧 −1 is also called unit delay operator
since delay by one unit involves The accumulation operation adds a pole at z=1 and zero at z=0
multiplication by 𝑧 −1
Properties of Z transform and Inverse Z transform
Z transform of common signals
 Inverse Z transform
• 𝑋 𝑧 = σ∞
𝑛=−∞ 𝑥[𝑛]𝑧
−𝑛

• We can view X(z) as power series in 𝑧 −𝑛 . The coefficient value corresponding to 𝑧 −𝑛 will be x[n].
• Another way to obtain x[n] from X(z) is through partial fractions
𝑁(𝑧) 𝑧−𝑧1 𝑧−𝑧2 …(𝑧−𝑧𝑚 )
• We can X(z) as a rational function of the form 𝑋 𝑧 = 𝐷(𝑧) = 𝑘( )
𝑧−𝑝1 𝑧−𝑝2 …(𝑧−𝑝𝑛 )

• The set 𝑧𝑘 1 ≤ 𝑘 ≤ 𝑚 are known as the zeros of X(z).

• The set 𝑝𝑘 1 ≤ 𝑘 ≤ 𝑛 are known as the poles of X(z). Assume that m<n and all poles are distinct .

𝑋 𝑧 𝑐0 𝑐1 𝑐2 𝑐𝑛 𝑐0 𝑐𝑘
• Then we can write = + + + ⋯+ = + σ𝑛𝑘=1
𝑧 𝑧 𝑧−𝑝1 𝑧−𝑝2 𝑧−𝑝𝑛 𝑧 𝑧−𝑝𝑘

𝑋 𝑧
• Where 𝑐0 = 𝑋 𝑧 |𝑧=0 and 𝑐𝑘 = 𝑧 − 𝑝𝑘 |𝑧=𝑝𝑘
𝑧

𝑐1 𝑧 𝑐2 𝑧 𝑐𝑛 𝑧
• Multiplying by z we get 𝑋 𝑧 = 𝑐0 + + + ⋯+
𝑧−𝑝1 𝑧−𝑝2 𝑧−𝑝𝑛
Inverse Z transform
𝑟 𝑋 𝑧
• For the multiple pole case X(z) has a factor of the form 𝑧 − 𝑝𝑖 where pole 𝑝𝑖 has multiplicity r, then has terms of the
𝑧
𝜆1 𝜆2 𝜆𝑟
form + + ⋯+
𝑧−𝑝𝑖 𝑧−𝑝𝑖 2 𝑧−𝑝𝑖 𝑟

1 𝑑𝑘 𝑟 𝑋(𝑧)
• The coefficient 𝜆𝑟−𝑘 = 𝑧 − 𝑝𝑖 evaluated at z= 𝑝𝑖
𝑘! 𝑑𝑧 𝑘 𝑧

• Properties of ROC: ROC does not contain any poles

• If x[n] is a finite sequence (x[n] is nonzero only for particular range of time instances) and X(z) converges for some value of z,
then ROC will be the entire z plane.

• For right-sided signal(x[n]=0 for n<𝑁1),the ROC will be of the form |𝑧| > 𝑟𝑚𝑎𝑥 where 𝑟𝑚𝑎𝑥 is maximum of magnitude of
poles of X(z).

• For left-sided signal(x[n]=0 for n>𝑁2 ),the ROC will be of the form |𝑧| < 𝑟𝑚𝑖𝑛 where 𝑟𝑚𝑖𝑛 is smallest magnitude of poles of
X(z).

• For two-sided signal x[n],the ROC will be of the form 𝑟1 < |𝑧| < 𝑟2 where 𝑟1 and 𝑟2 are magnitude of two poles of X(z).
 Z transform for discrete time LTI systems
• Consider discrete time LTI system with impulse response h[n].

• Since y[n]=x[n]*h[n] we have Y(z)=X(z)H(z)

𝑌(𝑧)
• We can rewrite it as 𝐻 𝑧 = 𝑋(𝑧)

• H(z) is known as the transfer function of the system and is Z transform of h[n].

• For system to be causal h[n]=0 for n<0, i.e h[n] is right sided signal.

• Therefore, ROC of H(z) is of the form |𝑧| > 𝑟𝑚𝑎𝑥

• For system to be BIBO stable σ∞

𝑛=−∞ ℎ[𝑛] < ∞

• In other words, ROC should contain the unit circle.

Transfer Function defined by Difference Equation
• Consider a discrete time LTI system with input x[n], output y[n]

• Then difference equation representation of LTI system is given as σ𝑁 𝑀

𝑘=0 𝑎𝑘 𝑦 𝑛 − 𝑘 = σ𝑘=0 𝑏𝑘 𝑥 𝑛 − 𝑘

• Taking Z transform on both sides we get σ𝑁 −𝑘 𝑀 −𝑘

𝑘=0 𝑎𝑘 𝑧 𝑌(𝑧) = σ𝑘=0 𝑏𝑘 𝑧 𝑋(𝑧)

• 𝑌(𝑧) σ𝑁
𝑘=0 𝑎𝑘 𝑧
−𝑘 = 𝑋(𝑧) σ𝑀 𝑏 𝑧 −𝑘
𝑘=0 𝑘

• After rearranging we get

𝑌(𝑧) σ𝑀
𝑘=0 𝑏𝑘 𝑧
−𝑘
• =𝐻 𝑧 = σ𝑁 −𝑘
𝑋(𝑧) 𝑘=0 𝑎𝑘 𝑧

• It is a rational transfer function, and we can obtain h[n] from obtaining inverse Z transform of H(z)
Numerical Example
𝑎𝑛 0≤𝑛 ≤𝑁−1
• Obtain Z transform of 𝑥 𝑛 = ቊ . Also obtain its zeros and poles and sketch the pole-zero plot
0 𝑂𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

• 𝑋 𝑧 = σ∞
𝑛=−∞ 𝑥[𝑛]𝑧
−𝑛
= σ𝑁−1 𝑛 −𝑛
𝑛=0 𝑎 𝑧

𝑎 𝑁
𝑁−1 𝑎 𝑛 1− 𝑧 𝑧 𝑁 −𝑎𝑁 1
• 𝑋 𝑧 = σ𝑛=0 = =
𝑧 𝑎 𝑧−𝑎 𝑧 𝑁−1
1−
𝑧

• Poles are at z=a and at z=0 with multiplicity N-1

• Zeros are obtained by solving 𝑧 𝑁 − 𝑎𝑁 = 0 i.e, 𝑧 𝑁 = 𝑎𝑁 ⇒ 𝑧 𝑁 = 𝑎𝑁 × 1

𝑗2𝜋𝑘 𝑗2𝜋𝑘
• This can be written as 𝑧 = 𝑎𝑒 𝑁 where 𝑒 𝑁 is called as Nth root of unity

• If you set k = 0, we have zero at z = a. So, the zero and pole at z = a cancels
Numerical Example
1 𝑛 1 𝑛
• Obtain Z transform and ROC of 𝑥 𝑛 = 3
𝑢 𝑛 + 2
𝑢 −𝑛 − 1

𝑧
• We know that the Z transform for 𝑎𝑛 𝑢[𝑛] is 𝑧−𝑎 with ROC |z|>|a|

1 𝑛 𝑧 1
• So 𝑢 𝑛 will have Z transform as 1 with ROC |z|>
3 𝑧−3 3

𝑧
• For −𝑎𝑛 𝑢[−𝑛 − 1], the Z transform is 𝑧−𝑎 with ROC |z|<|a|

1 𝑛 −𝑧 1
• So 2
𝑢 −𝑛 − 1 will have Z transform 1 with ROC |z|<2
𝑧−2

𝑧 𝑧 −𝑧 1 1
• Therefore 𝑋 𝑧 = 1 − 1 = 1 1 with ROC |z|>3 ∩ |z|<2
𝑧−3 𝑧−2 6(𝑧−3)(𝑧−2)

−𝑧 1 1
• 𝑋 𝑧 = 1 1 with ROC as 3 < 𝑧 < 2
6(𝑧−3)(𝑧−2)

1 1
• There is a zero at z=0 and poles at 𝑧 = 3 and 2
 Numerical Example
• Obtain the Z transform and ROC of 𝑥 𝑛 = 𝑎(𝑛+2) 𝑢 𝑛 + 2

𝑧
• We know that the Z transform for y n = 𝑎𝑛 𝑢[𝑛] is Y(z) = 𝑧−𝑎 with ROC |z|>|a|

• x[n] is a time advanced version of 𝑦[𝑛]

𝑧3
• Therefore, 𝑋 𝑧 = 𝑧 2𝑌 𝑧 = with ROC |z|>|a|
𝑧−𝑎
 Numerical Example
3𝑧 2 −5𝑧
• Given 𝑋 𝑧 = with ROC |z|<1, Obtain x[n]
𝑧 2 −3𝑧+2

• Given ROC is |z|<1 therefore it is a left handed sequence

3𝑧 2 −5𝑧 𝑋(𝑧) 𝑐0 𝑐1 𝑐2
• We can rewrite X(z) as 𝑋 𝑧 = ⇒ = + +
(𝑧−2)(𝑧−1) 𝑧 𝑧 𝑧−2 𝑧−1

5
• The zeros are at values of z where 3𝑧 2 − 5𝑧 = 0 i.e, z=0 and 3

3×0−5×0
• 𝑐0 = 𝑋 𝑧 |𝑧=0 = 0−3×0+2 = 0

𝑋(𝑧) 3×2−5
• 𝑐1 = 𝑧 − 2 | = =1
𝑧 𝑧=2 (2−1)

𝑋(𝑧) 3×1−5×1
• 𝑐2 = 𝑧 − 1 | = =2
𝑧 𝑧=1 (1−2)
 Numerical Example
𝑋(𝑧) 1 2 𝑧 2𝑧
• Therefore = + ⇒ 𝑋(𝑧) = +
𝑧 𝑧−2 𝑧−1 𝑧−2 𝑧−1

𝑧
• The left sided sequence corresponding to is −2𝑛 𝑢 −𝑛 − 1
𝑧−2

2𝑧
• The left sided sequence corresponding to is −2𝑢 −𝑛 − 1
𝑧−1

• Therefore, 𝑥 𝑛 = −2𝑢 −𝑛 − 1 −2𝑛 𝑢 −𝑛 − 1

Numerical Example
𝑧(−𝑧 2 +5𝑧−5)
• Given 𝑋 𝑧 = (𝑧−2) (𝑧−3)2
with ROC 2 < |z|<3 Obtain x[n]

• Based on the ROC we can say that the signal is a two-sided signal with poles at p=2 and p=3 with multiplicity at 2.

𝑋(𝑧) 𝑐0 𝑐1 𝜆1 𝜆2
• Partial fraction expansion of X(z) will be 𝑧
= 𝑧
+ 𝑧−2
+ 𝑧−3
+ 𝑧−3 2

0
• 𝑐0 = 𝑋 𝑧 |𝑧=0 = −2×9 = 0

𝑋(𝑧) −4+10−5
• 𝑐1 = 𝑧 − 2 | = =1
𝑧 𝑧=2 (2−3)2

1 𝑑𝑘 𝑟 𝑋(𝑧)
• 𝜆𝑟−𝑘 = 𝑘! 𝑑𝑧 𝑘 𝑧 − 𝑝𝑖 𝑧
evaluated at z= 𝑝𝑖

2 𝑋(𝑧) | (−9+5×3−5)
• 𝜆2 = 𝑧 − 3 𝑧 𝑧=3
= (3−2)
=1

1 𝑑 2𝑋 𝑧
• 𝜆1= 1! 𝑑𝑧 𝑧−3 𝑧
|𝑧=3
Numerical Example
2𝑋 𝑧 (−𝑧 2 +5𝑧−5) − 𝑧−2 2 +𝑧−1 𝑧−2+1 1
• 𝑧−3 = = =− 𝑧−2 + = − 𝑧 − 2 + 1 + 𝑧−2
𝑧 (𝑧−2) 𝑧−2 𝑧−2

𝑑 2𝑋 𝑧 1
• 𝑑𝑧
𝑧−3 𝑧
= −1 − (𝑧−2)2

• Substituting z=3 we get 𝜆1 =-2

𝑋(𝑧) 1 2 1
• = − +
𝑧 𝑧−2 𝑧−3 𝑧−3 2

𝑧 2𝑧 𝑧
• 𝑋 𝑧 = − + with ROC 2 < |z|<3
𝑧−2 𝑧−3 𝑧−3 2

𝑧
• corresponds to 2𝑛 𝑢 𝑛 with ROC 𝑧 > 2
𝑧−2

2𝑧
• − corresponds to −2 × −3𝑛 𝑢 −𝑛 − 1 with ROC 𝑧 < 3
𝑧−3

𝑧
• For the term we make use of the differentiation property
𝑧−3 2
Numerical Example
𝑧
• Consider 𝑋෨ 𝑧 = 𝑧−3

𝑑 𝑑 𝑧 𝑑 𝑧−3+3 𝑑 3 −3
• 𝑋෨ 𝑧 = 𝑑𝑧 𝑧−3 = 𝑑𝑧 = 𝑑𝑧 1 + 𝑧−3 = (𝑧−3)2
𝑑𝑧 𝑧−3

𝑑 3𝑧
• Now −z 𝑑𝑧 𝑋෨ 𝑧 = 𝑧−3 2

𝑧
• For 𝑋෨ 𝑧 = , the inverse Z transform is −3𝑛 𝑢 −𝑛 − 1
𝑧−3

𝑧 1
• So, for the inverse Z transform is 3 −𝑛 3𝑛 𝑢 −𝑛 − 1 = −𝑛 3𝑛−1 𝑢 −𝑛 − 1
𝑧−3 2

• Inverse Z transform for X(z) is 2𝑛 𝑢 𝑛 + 2 × 3𝑛 𝑢 −𝑛 − 1 −𝑛 3𝑛−1 𝑢 −𝑛 − 1