YouTube - IMPLearn

Realisation of discrete time system

• A digital filter transfer function can be realized in a variety ways
• Realization of FIR filters
• Realization of IIR filters
• Basic elements required for implementation of an LTI digital system are adder , multiplier and
memory for storing elements
• In digital implementation the delay operation can be implemented by providing a storage register by
each unit delay is required

𝒙𝟏 𝒏 𝒂
𝒙𝟏 𝒏 + 𝒙𝟐 𝒏 𝒙 𝒏 𝒂𝒙 𝒏 Z-1
𝒙 𝒏 𝒙 𝒏−𝟏

𝒙𝟐 𝒏

Adder Multiplier Unit delay

[Link]
YouTube - IMPLearn

FIR filter realization

• Mainly four types of realisation are there

1. Direct form realisation
2. Cascade form realisation
3. Linear phase realisation
4. Lattice structure realisation

[Link]
 YouTube - IMPLearn
Direct form realization
The direct form of FIR may be obtained by using the equation of linear convolution
𝑁−1

𝑦 𝑘 = ෍ℎ 𝑘 𝑥 𝑛−𝑘 = ℎ 0 𝑥 𝑛 + ℎ 1 𝑥 𝑛 − 1 + ⋯+ ℎ 𝑁 − 1 𝑥 𝑛 − 𝑁 − 1
𝑘=0

Taking Z-transform
𝑌 𝑧 = ℎ 0 𝑋 𝑧 + ℎ 1 𝑋 𝑧 𝑧 −1 + ⋯ + ℎ 𝑁 − 1 𝑋 𝑧 𝑧 − 𝑁−1

𝑿 𝒛
Z-1 Z-1 Z-1 Z-1

𝒉 𝟎 𝒉 𝟏 𝒉 𝟐 𝒉 𝑵−𝟐 𝒉 𝑵−𝟏

𝒀 𝒛

[Link]
 YouTube - IMPLearn
Q) Determine the direct form realization of system function
𝐻 𝑧 = 1 + 2𝑧 −1 − 3𝑧 −2 − 4𝑧 −3 + 5𝑧 −4
𝑌 𝑧
𝐻 𝑧 =
𝑋 𝑧
𝐻 𝑧 = 1 + 2𝑧 −1 − 3𝑧 −2 − 4𝑧 −3 + 5𝑧 −4

𝑌 𝑧 = 𝑋 𝑧 1 + 2𝑧 −1 − 3𝑧 −2 − 4𝑧 −3 + 5𝑧 −4

𝑿 𝒛
Z-1 Z-1 Z-1 Z-1

𝟏 𝟐 −𝟑 −𝟒 𝟓

𝒀 𝒛

[Link]
 YouTube - IMPLearn
Cascade form realization
Q) Obtain cascade form realization of system function 𝐻 𝑧 = 1 + 2𝑧 −1 − 𝑧 −2 1 + 𝑧 −1 − 𝑧 −2

𝐻 𝑧 = 𝐻1 𝑧 𝐻2 𝑧

𝑌1 𝑧 𝑿𝟏 𝒛 𝒀𝟏 𝒛 𝑿𝟐 𝒛 𝒀 𝒛
𝐻1 𝑧 =
𝑋1 𝑧

𝑌2 𝑧
𝐻2 𝑧 =
𝑋2 𝑧 Z-1 Z-1
𝟐 𝟏

𝑌1 𝑧 = 𝑋1 𝑧 1 + 2𝑧 −1 − 𝑧 −2
Z-1 Z-1
𝑌1 𝑧 = 𝑋1 𝑧 + 2𝑋 𝑧 𝑧 −1 −𝑋 𝑧 𝑧 −2
−𝟏 −𝟏
𝑌2 𝑧 = 𝑋2 𝑧 1 + 𝑧 −1 − 𝑧 −2

𝑌2 𝑧 = 𝑋2 𝑧 + 𝑋2 𝑧 𝑧 −1 − 𝑋2 𝑧 𝑧 −2
[Link]
 YouTube - IMPLearn
Cascade form realization
5
Q) Obtain cascade form realization of system function 𝐻 𝑧 = 1 + 𝑧 −1 + 2𝑧 −2 + 2𝑧 −3
2

5 −1
𝐻 𝑧 = 1 + 𝑧 + 2𝑧 −2 + 2𝑧 −3
2 𝑿 𝒛 𝒀 𝒛
𝑧3 5 −1
= 3 1 + 𝑧 + 2𝑧 −2 + 2𝑧 −3
𝑧 2
1 3 5 2 Z-1
= 3 𝑧 + 𝑧 + 2𝑧1 + 2 Z-1 𝟏
𝑧 2
𝟐
The term inside the bracket equal to zero when z=-2 𝟐

So the first term can be (z+2)

Z-1
1 1 𝟏
= 3 𝑧+2 𝑧2 + 𝑧 + 1
𝑧 2

1 1 −1
= 3 𝑧 1 + 2𝑧 𝑧 1 + 𝑧 + 𝑧 −2
−1 2
𝑧 2
[Link]
 YouTube - IMPLearn
Linear phase realization
1 1 1 1 1
Q) Realise the system function 𝐻 𝑧 = + 𝑧 −1 + 𝑧 −2 + 𝑧 −3 + 𝑧 −4 + 𝑧 −5 + 𝑧 −6
2 3 4 3 2
ℎ 0 ℎ 1 𝒉 𝟐 𝒉 𝟑 𝒉 𝟒 𝒉 𝟓 𝒉 𝟔

For a linear phase FIR filter ℎ 𝑛 = ℎ 𝑁 − 1 − 𝑛

𝑁=7
1
ℎ 0 = ℎ 7−1−0 =ℎ 6 =
2 𝑿 𝒛
1 Z-1 Z-1 Z-1
ℎ 1 = ℎ 7−1−1 =ℎ 5 =
3
ℎ 2 = ℎ 7−1−2 =ℎ 4 =1
1
ℎ 3 = ℎ 7−1−3 =ℎ 3 = 𝟏 𝟏
4 𝟏 𝟑
𝟐 Z-1 Z-1 Z-1
𝐻 𝑧
1 −6
1 −1 𝟏
= 1+𝑧 + 𝑧 + 𝑧 −5 𝟒
2 3
−2 −4
1 −3
+ 1. 𝑧 + 𝑧 + 𝑧
4
𝒀 𝒛 [Link]
 YouTube - IMPLearn
Linear phase realization
Q) Realise the system function using minimum number of multiplier
1 1 1 1
𝐻 𝑧 = 1 + 𝑧 −1 + 𝑧 −2 + 𝑧 −3 + 𝑧 −4 + 𝑧 −5
3 4 4 3
ℎ 0 ℎ 1 𝒉 𝟐 𝒉 𝟑 𝒉 𝟒 𝒉 𝟓

For a linear phase FIR filter ℎ 𝑛 = ℎ 𝑁 − 1 − 𝑛

𝑁=6

ℎ 0 = ℎ 6−1−0 =ℎ 5 =1
𝑿 𝒛
1 Z-1 Z-1
ℎ 1 = ℎ 6−1−1 =ℎ 4 =
3
1
ℎ 2 = ℎ 6−1−2 =ℎ 3 =
4 Z-1
𝟏 𝟏
𝟑 𝟒
𝐻 𝑧
1 −1 Z-1 Z-1
= 1 1 + 𝑧 −5 + 𝑧 + 𝑧 −4
3
1 −2
+ 𝑧 + 𝑧 −3
4

𝒀 𝒛 [Link]
 YouTube - IMPLearn
Linear phase realization
Q) Realise the system function using minimum number of multiplier
1 1
𝐻 𝑧 = 1 + 𝑧 −1 1 + 𝑧 −1 + 𝑧 −2 + 𝑧 −3
2 2

𝑿 𝒛 Z-1

Z-1
Z-1
Z-1

𝒀 𝒛

[Link]
YouTube - IMPLearn
Conversion of lattice coefficient to direct form filter coefficient

The general equation

𝑦 𝑛 = 𝑥 𝑛 + ෍ 𝛼𝑚 𝑘 𝑥 𝑛 − 𝑘
𝑘=1

The equations to convert filter coefficients to direct form FIR filter coefficients are

𝛼𝑚 0 = 1

𝛼𝑚 𝑚 = 𝐾𝑚

𝛼𝑚 𝑘 = 𝛼𝑚−1 𝑘 + 𝛼𝑚 𝑚 𝛼𝑚−1 𝑚 − 𝑘

[Link]
 YouTube - IMPLearn
Q) Consider an FIR filter lattice filter with coefficient K1 = ½ ,K2 = 1/3 , K3 = ¼. Determine the
FIR filter coefficients for direct form structure?

Solution 𝑚

𝑦 𝑛 = 𝑥 𝑛 + ෍ 𝛼𝑚 𝑘 𝑥 𝑛 − 𝑘
Number of stages (m) = 3 𝑘=1

𝑦 𝑛 = 𝑥 𝑛 + ෍ 𝛼𝑚 𝑘 𝑥 𝑛 − 𝑘 = 𝑥 𝑛 + 𝛼3 1 𝑥 𝑛 − 1 + 𝛼3 2 𝑥 𝑛 − 2 + 𝛼3 3 𝑥 𝑛 − 3
𝑘=1

For m=3 For m=2 For m=1

𝛼3 0 = 1 𝛼2 0 = 1 𝛼1 0 = 1
1 1 1
𝛼3 3 = 𝛼2 2 = 𝛼1 1 =
4 3 2
and k=2 and k=1
For m=3 and k=1
𝛼3 2 𝛼2 1
= 𝛼2 2 + 𝛼3 3 𝛼2 1 𝛼3 1 = 𝛼2 1 + 𝛼3 3 𝛼2 2
= 𝛼1 1 + 𝛼2 2 𝛼1 1
1 1 1 1 1 1 2 2 1 1 3
= + 𝛼2 1 = = + . = = + . = [Link]
3 4 2 2 3 2 3 3 4 3 4
YouTube - IMPLearn
Conversion of direct form FIR filter coefficient to lattice coefficient

The general equation

𝑦 𝑛 = 𝑥 𝑛 + ෍ 𝛼𝑚 𝑘 𝑥 𝑛 − 𝑘
𝑘=1

The equations to convert filter coefficients to lattice form FIR filter coefficients are

𝛼𝑚−1 0 = 1

𝐾𝑚 = 𝛼𝑚 𝑚

𝛼𝑚 𝑘 − 𝛼𝑚 𝑚 𝛼𝑚 𝑚 − 𝑘
𝛼𝑚−1 𝑘 = 2 𝑓𝑜𝑟 1 ≤ 𝑘 ≤ 𝑚 − 1
1 − 𝛼𝑚 𝑚
[Link]
 YouTube - IMPLearn
4 3 2
A FIR filter is given by the difference equation 𝑦 𝑛 = 2𝑥 𝑛 + 5 𝑥 𝑛 − 1 + 2 𝑥 𝑛 − 2 + 3 𝑥 𝑛 − 3 Determine
its lattice form

Solution
𝑚
3
𝑦 𝑛 = 𝑥 𝑛 + ෍ 𝛼𝑚 𝑘 𝑥 𝑛 − 𝑘
𝑦 𝑛 = 𝑥 𝑛 + ෍ 𝛼𝑚 𝑘 𝑥 𝑛 − 𝑘 𝑘=1
𝑘=1

= 𝑥 𝑛 + 𝛼3 1 𝑥 𝑛 − 1 + 𝛼3 2 𝑥 𝑛 − 2 + 𝛼3 3 𝑥 𝑛 − 3

2 3 1
𝑦 𝑛 =2 𝑥 𝑛 + 𝑥 𝑛−1 + 𝑥 𝑛−2 + 𝑥 𝑛−3
5 4 3

Comparing the two equations we get

2 3 1
𝛼3 1 = 𝛼3 2 = 𝛼 3 3 = = 𝑘3
5 4 3

𝛼2 0 = 1
[Link]
 YouTube - IMPLearn 2 3 1
𝛼3 1 = 𝛼3 2 = 𝛼3 3 = = 𝑘3 𝛼2 0 = 1
For m=3, k=1 5 4 3

2 1 3
𝛼3 1 − 𝛼3 3 𝛼3 2 − . 𝑚
𝛼2 1 = = 5 3 4
1 − 𝛼32 3 = 0.1687 𝑦 𝑛 = 𝑥 𝑛 + ෍ 𝛼𝑚 𝑘 𝑥 𝑛 − 𝑘
1 2
1− 3 𝑘=1

For m=3, k=2

3 1 2
𝛼3 2 − 𝛼3 3 𝛼3 1 − .
𝑘2 = 𝛼2 2 = = 4 3 5
= 0.6937
1 − 𝛼22 3 1 2
1− 3

For m=2, k=1

𝛼2 1 − 𝛼2 2 𝛼2 1 0.1687 − (0.6937)0.1687
𝑘1 = 𝛼1 1 = =
1 − 𝛼22 2 1 − 0.6937 2

= 0.0996
[Link]
YouTube - IMPLearn

IIR filter Realisation

• Mainly four types of realisation are there

1. Direct form I realisation
2. Direct form II realisation
3. Cascade form realisation
4. Parallel form realisation

[Link]
 YouTube - IMPLearn
Direct form I realization
Let us consider an IIR system described by the difference equation
𝑁 𝑀

𝑦 𝑛 = − ෍ 𝑎𝑘 𝑦 𝑛 − 𝑘 + ෍ 𝑏𝑘 𝑥 𝑛 − 𝑘 𝒙 𝒏 𝒚 𝒏
𝒃𝟎 𝒘 𝒏
𝑘=1 𝑘=0

= −𝑎1 𝑦 𝑛 − 1 − 𝑎2 𝑦 𝑛 − 2 …
− 𝑎𝑁 𝑦 𝑛 − 𝑁 + 𝑤 𝑛 Z-1 Z-1
𝒃𝟏 −𝒂𝟏
Where 𝒙 𝒏−𝟏 𝒚 𝒏−𝟏

𝑤 𝑛
Z-1 Z-1
= 𝑏0 𝑥 𝑛 + 𝑏1 𝑥 𝑛 − 𝑁 + ⋯
+ 𝑏𝑀 𝑥 𝑛 − 𝑀
𝒃𝑴−𝟏 −𝒂𝑵−𝟏

Z-1 Z-1
𝒃𝑴 −𝒂𝑵
𝒙 𝒏−𝑴 𝒚 𝒏−𝑴
[Link]
 YouTube - IMPLearn
Direct form I realization
Q) Realise the second order digital filter
𝑦 𝑛 = 2𝑟 cos 𝜔0 𝑦 𝑛 − 1 − 𝑟 2 𝑦 𝑛 − 2 + 𝑥 𝑛 − 𝑟 cos 𝜔0 𝑥 𝑛 − 1
Solution

𝑤 𝑛 = 𝑥 𝑛 − 𝑟 cos 𝜔0 𝑥 𝑛 − 1 𝑦 𝑛 = 2𝑟 cos 𝜔0 𝑦 𝑛 − 1 − 𝑟 2 𝑦 𝑛 − 2 + 𝑤 𝑛

𝒙 𝒏 𝒚 𝒏
𝒘 𝒏

Z-1 Z-1
−𝑟 cos 𝜔0 2𝑟 cos 𝜔0

Z-1
𝑟2

[Link]
 YouTube - IMPLearn
Direct form I realization
Q) Obtain the direct form I realisation for the system described by difference equation
𝑦 𝑛 = 0.5𝑦 𝑛 − 1 − 0.25𝑦 𝑛 − 2 + 𝑥 𝑛 + 0.4𝑥 𝑛 − 1
Solution

𝑤 𝑛 = 𝑥 𝑛 + 0.4𝑥 𝑛 − 1 𝑦 𝑛 = 0.5𝑦 𝑛 − 1 − 0.25𝑦 𝑛 − 2 + 𝑤 𝑛

𝒙 𝒏 𝒚 𝒏
𝒘 𝒏

Z-1 Z-1
0.4 0.5

Z-1
−0.25

[Link]
 YouTube - IMPLearn
Direct form II realization
Let us consider an IIR system described by the difference equation
𝑁 𝑀

𝑦 𝑛 = − ෍ 𝑎𝑘 𝑦 𝑛 − 𝑘 + ෍ 𝑏𝑘 𝑥 𝑛 − 𝑘
𝑘=1 𝑘=0 𝑌 𝑧 = 𝑏0 𝑊 𝑧 + 𝑏1 𝑧 −1 𝑊 𝑧 + ⋯ + 𝑏𝑀 𝑧 −𝑀 𝑊 𝑧
The system function can be represented as where
σ𝑀
𝑘=0 𝑏𝑘 𝑧
−𝑘
𝐻 𝑧 = 𝑊 𝑧 + 𝑎1 𝑧 −1 𝑊 𝑧 + 𝑎2 𝑧 −2 𝑊 𝑧 … + 𝑎𝑁 𝑧 −𝑁 𝑊 𝑧 = 𝑋 𝑧
1 + σ𝑁
𝑘=1 𝑎𝑘 𝑧
−𝑘

Let
𝑊 𝑧 = 𝑋 𝑧 − 𝑎1 𝑧 −1 𝑊 𝑧 − 𝑎2 𝑧 −2 𝑊 𝑧 … − 𝑎𝑁 𝑧 −𝑁 𝑊 𝑧
𝑌 𝑧 𝑌 𝑧 𝑊 𝑧
= .
𝑋 𝑧 𝑊 𝑧 𝑋 𝑧 Taking inverse z transform we get
𝑀
𝑌 𝑧 𝑤 𝑛 = 𝑥 𝑛 − 𝑎1 𝑤 𝑛 − 1 − 𝑎2 𝑤 𝑛 − 2 … − 𝑎𝑁 𝑤 𝑛 − 𝑁
= ෍ 𝑏𝑘 𝑧 −𝑘
𝑊 𝑧
𝑘=0
𝑦 𝑛 = 𝑏0 𝑤 𝑛 + 𝑏1 𝑤 𝑛 − 1 + ⋯ + 𝑏𝑀 𝑤 𝑛 − 𝑀
𝑊 𝑧 1
=
𝑋 𝑧 1 + σ𝑁
𝑘=1 𝑎𝑘 𝑧
−𝑘 [Link]
YouTube - IMPLearn
Direct form II realization
𝑤 𝑛 = 𝑥 𝑛 − 𝑎1 𝑤 𝑛 − 1 − 𝑎2 𝑤 𝑛 − 2 … − 𝑎𝑁 𝑤 𝑛 − 𝑁

𝑦 𝑛 = 𝑏0 𝑤 𝑛 + 𝑏1 𝑤 𝑛 − 1 + ⋯ + 𝑏𝑀 𝑤 𝑛 − 𝑀

𝒙 𝒏 𝒚 𝒏

Z-1
−𝒂𝟏 𝒃𝟏

Z-1

−𝒂𝑵−𝟏 𝒃𝑴−𝟏

Z-1
−𝒂𝑵 𝒃𝑴
[Link]
 YouTube - IMPLearn
Direct form II realization
Q) Realise the second order digital filter using direct form II 𝑌 𝑧 𝑌 𝑧 𝑊 𝑧
= .
𝑦 𝑛 = 2𝑟 cos 𝜔0 𝑦 𝑛 − 1 − 𝑟 2 𝑦 𝑛 − 2 + 𝑥 𝑛 − 𝑟 cos 𝜔0 𝑥 𝑛 − 1 𝑋 𝑧 𝑊 𝑧 𝑋 𝑧
Solution
Taking Z - Transform

𝑌 𝑧 = 2𝑟 cos 𝜔0 𝑌 𝑧 𝑧 −1 − 𝑟 2 𝑌 𝑧 𝑧 −2 + 𝑋 𝑧 − 𝑟 cos 𝜔0 𝑋 𝑧 𝑧 −1

𝑌 𝑧 1 − 2𝑟 cos 𝜔0 𝑧 −1 + 𝑟 2 𝑧 −2 = 𝑋 𝑧 1 − 𝑟 cos 𝜔0 𝑧 −1

𝑌 𝑧 1 − 𝑟 cos 𝜔0 𝑧 −1
=
𝑋 𝑧 1 − 2𝑟 cos 𝜔0 𝑧 −1 + 𝑟 2 𝑧 −2 𝑊 𝑧 1
=
𝑌 𝑧 𝑋 𝑧 1 − 2𝑟 cos 𝜔0 𝑧 −1 + 𝑟 2 𝑧 −2
= 1 − 𝑟 cos 𝜔0 𝑧 −1
𝑊 𝑧
𝑊 𝑧 − 2𝑟 cos 𝜔0 𝑊 𝑧 𝑧 −1 + 𝑟 2 𝑊 𝑧 𝑧 −2 = 𝑋 𝑧
𝑌 𝑧 = 𝑊 𝑧 1 − 𝑟 cos 𝜔0 𝑧 −1
𝑊 𝑧 = 𝑋 𝑧 + 2𝑟 cos 𝜔0 𝑧 −1 𝑊 𝑧 − 𝑟 2 𝑧 −2 𝑊 𝑧
Inverse transform
Inverse transform
𝑦 𝑛 = 𝑤 𝑛 − 𝑟 cos 𝜔0 𝑤 𝑛 − 1 𝑤 𝑛 = 𝑥 𝑛 + 2𝑟 cos 𝜔0 𝑥 𝑛 − 1 − 𝑟 2 𝑥 𝑛 − 2
[Link]
YouTube - IMPLearn
Direct form II realization
𝑦 𝑛 = 𝑤 𝑛 − 𝑟 cos 𝜔0 𝑤 𝑛 − 1 𝑧 −1 𝑤 𝑛 = 𝑥 𝑛 + 2𝑟 cos 𝜔0 𝑥 𝑛 − 1 − 𝑟 2 𝑥 𝑛 − 2

𝒙 𝒏 𝒘 𝒏 𝒚 𝒏

Z-1
𝟐𝒓 𝒄𝒐𝒔𝝎𝟎 −𝒓 𝒄𝒐𝒔𝝎𝟎

Z-1
−𝒓𝟐

[Link]
 YouTube - IMPLearn
Direct form II realization
Q) Determine the direct form II realisation for the following system 𝑌 𝑧 𝑌 𝑧 𝑊 𝑧
𝑦 𝑛 = −0.1𝑦 𝑛 − 1 + 0.72𝑦 𝑛 − 2 + 0.7𝑥 𝑛 − 0.252𝑥 𝑛 − 1 = .
𝑋 𝑧 𝑊 𝑧 𝑋 𝑧
Solution
Taking Z - Transform

𝑌 𝑧 = −0.1𝑌 𝑧 𝑧 −1 + 0.72𝑌 𝑧 𝑧 −2 + 0.7𝑋 𝑧 − 0.252𝑋 𝑧 𝑧 −1

𝑌 𝑧 1 + 0.1𝑧 −1 − 0.72𝑧 −2 = 𝑋 𝑧 0.7 − 0.252𝑧 −1

𝑌 𝑧 0.7 − 0.252𝑧 −1
=
𝑋 𝑧 1 + 0.1𝑧 −1 − 0.72𝑧 −2 𝑊 𝑧 1
=
𝑌 𝑧 𝑋 𝑧 1 + 0.1𝑧 −1 − 0.72𝑧 −2
= 0.7 − 0.252𝑧 −1
𝑊 𝑧
𝑊 𝑧 + 0.1𝑊 𝑧 𝑧 −1 − 0.72𝑊 𝑧 𝑧 −2 = 𝑋 𝑧
𝑌 𝑧 = 𝑊 𝑧 0.7 − 0.252𝑧 −1
𝑊 𝑧 = 𝑋 𝑧 − 0.1𝑊 𝑧 𝑧 −1 + 0.72𝑊 𝑧 𝑧 −2
Inverse transform
Inverse transform
𝑦 𝑛 = 0.7𝑤 𝑛 − 0.252𝑤 𝑛 − 1 𝑤 𝑛 = 𝑥 𝑛 − 0.1𝑤 𝑛 − 1 + 0.72𝑤 𝑛 − 2
[Link]
YouTube - IMPLearn
Direct form II realization

𝑦 𝑛 = 0.7𝑤 𝑛 − 0.252𝑤 𝑛 − 1 𝑤 𝑛 = 𝑥 𝑛 − 0.1𝑤 𝑛 − 1 + 0.72𝑤 𝑛 − 2

𝒙 𝒏 𝒘 𝒏 𝟎. 𝟕 𝒚 𝒏

Z-1
−𝟎. 𝟏 −𝟎. 𝟐𝟓𝟐

Z-1
𝟎. 𝟕𝟐

[Link]
 YouTube - IMPLearn
Cascade form realization
Let us consider an IIR system with system function

𝒚𝟏 𝒏 𝒚 𝒏
= 𝒙𝟐 𝒏
𝒙 𝒏 𝒚 𝒏 𝒙 𝒏 = 𝒙𝟏 𝒏 𝒃𝒌𝟎 𝒃𝒎𝟎
H1(z) H2(z)

Z-1 Z-1
𝑏𝑘0 + 𝑏𝑘1 𝑧 −1 + 𝑏𝑘2 𝑧 −2
𝐻1 𝑧 = −𝒂𝒌𝟏 −𝒃𝒌𝟏 −𝒂𝒎𝟏 −𝒃𝒎𝟏
1 + 𝑎𝑘1 𝑧 −1 + 𝑎𝑘2 𝑧 −2

𝑏𝑚0 + 𝑏𝑚1 𝑧 −1 + 𝑏𝑚2 𝑧 −2 Z-1 Z-1

𝐻2 𝑧 =
1 + 𝑎𝑚1 𝑧 −1 + 𝑎𝑚2 𝑧 −2 −𝒂𝒌𝟐 −𝒃𝒌𝟐 −𝒂𝒎𝟐 −𝒃𝒎𝟐

[Link]
 YouTube - IMPLearn
Cascade form realization
Q) Realise the system with difference equation
3 1 1
𝑦 𝑛 = 𝑦 𝑛 − 1 − 𝑦 𝑛 − 2 + 𝑥 𝑛 + 𝑥 𝑛 − 1 in cascade form
4 8 3
𝑌 𝑧
Solution 𝐻 𝑧 =
𝑋 𝑧
Taking Z - Transform

3 1 1
𝑌 𝑧 = 𝑌 𝑧 𝑧 − 𝑌 𝑧 𝑧 + 𝑋 𝑧 + 𝑋 𝑧 𝑧 −1
−1 −2
4 8 3
3 −1 1 −2 1 −1
𝑌 𝑧 1− 𝑧 + 𝑧 =𝑋 𝑧 1+ 𝑧 1
4 8 3 1 + 𝑧 −1
𝐻1 𝑧 = 3
1 1
1 + 𝑧 −1 1 − 𝑧 −1
𝑌 𝑧 3 2
=
𝑋 𝑧 3 1
1 − 𝑧 −1 + 𝑧 −2
4 8 1
1 𝐻2 𝑧 =
1
𝑌 𝑧 1 + 𝑧 −1 1 − 𝑧 −1
= 3 4
𝑋 𝑧 1 1
1 − 𝑧 −1 1 − 𝑧 −1
2 4 [Link]
 YouTube - IMPLearn

1 1
1 + 𝑧 −1 𝐻2 𝑧 =
1
3
𝐻1 𝑧 =
1 1 − 𝑧 −1
4
1 − 𝑧 −1
2
1 1 1
𝑌1 𝑧 = 𝑋1 𝑧 + 𝑋1 𝑧 𝑧 + 𝑌1 𝑧 𝑧 −1
−1
𝑌2 𝑧 = X2 𝑧 + 𝑌2 𝑧
3 2 4
1 1 1
𝑦1 𝑛 = 𝑥1 𝑛 + 𝑥1 𝑛 − 1 + 𝑦1 𝑛 − 1 𝑦2 𝑛 = x2 𝑛 + 𝑦2 𝑛 − 1
3 2 4

𝒚𝟏 𝒏
𝒙 𝒏 = 𝒙𝟏 𝒏 = 𝒙𝟐 𝒏 𝒚𝟐 𝒏

Z-1 Z-1
𝟏 𝟏 𝟏
𝟑 𝟐 𝟒

[Link]
 YouTube - IMPLearn
Parallel form realization
A parallel form realisation of an IIR system can be obtained by performing a partial expansion
𝑁
𝐶𝑘 𝐶1 𝐶2 𝐶𝑁
𝐻 𝑧 =𝐶+෍ 𝐻 𝑧 =𝐶+ + + ⋯ +
1 − 𝑃𝑘 𝑧 −1 1 − 𝑃1 𝑧 −1 1 − 𝑃2 𝑧 −1 1 − 𝑃𝑁 𝑧 −1
𝑘=1

𝑌 𝑧 = 𝑐𝑋 𝑧 + 𝐻1 𝑧 𝑋 𝑧 + 𝐻2 𝑧 𝑋 𝑧 + ⋯ + 𝐻𝑁 𝑧 𝑋 𝑧

𝑪
𝒙 𝒏

H1(z)

H2(z)

Hk(z)
𝒚 𝒏
[Link]
 YouTube - IMPLearn
Parallel form realization
Q) Realise the system given by difference equation
𝑦 𝑛 = −0.1𝑦 𝑛 − 1 + 0.72𝑦 𝑛 − 2 + 0.7𝑥 𝑛 − 0.25𝑥 𝑛 − 2 in parallel form

Taking Z - Transform

𝑌 𝑧 = −0.1𝑌 𝑧 𝑧 −1 + 0.72𝑌 𝑧 𝑧 −2 + 0.7𝑋 𝑧 − 0.25𝑋 𝑧 𝑧 −2

𝑌 𝑧 1 + 0.1𝑧 −1 − 0.72𝑧 −2 = 𝑋 𝑧 0.7 − 0.25𝑧 −2

0.7−0.25𝑧 −2
𝐻 𝑧 =
1+ 0.1𝑧 −1 −0.72𝑧 −2

−0.035𝑧 −1 +0.35
= 0.35 +
1+0.1𝑧 −1 −0.72𝑧 −2

0.206 0.144
= 0.35 + +
1 + 0.9𝑧 −1 1 − 0.8𝑧 −1
[Link]
YouTube - IMPLearn
Parallel form realization
0.206 0.144
𝐻 𝑧 = 0.35 + +
1 + 0.9𝑧 −1 1 − 0.8𝑧 −1

𝟎. 𝟑𝟓
𝒙 𝒏 𝟎. 𝟐𝟎𝟔

Z-1

−𝟎. 𝟗

𝟎. 𝟏𝟒𝟒 𝒚 𝒏

Z-1
𝟎. 𝟖
[Link]
YouTube - IMPLearn

Multi rate DSP

• The processing of a discrete time signal at different sampling rates in

different parts of a system is called multi rate DSP
• Discrete time system that employ sampling rate conversion while
processing the discrete time signal are called multi rate DSP system
• The process of converting a signal from one sampling rate to another
sampling rate are of two types
• Down sampling or decimation
• Up sampling or interpolation

[Link]
YouTube - IMPLearn

Down sampling
• It is the process of reducing the sampling rate by an integer factor D or M
• Down sampled signal of x(n) can be obtained by simply keeping every
Mth sample and removing (M-1) in between samples

𝒙 𝒏 𝒚 𝒏 = 𝒙 𝑴𝒏
ꜜM

𝑥 𝑛 = 1, −1, 2, 4, 0, 3, 2, 1, 5 for M=2

𝑥 𝑀𝑛 = 1, 2, 0, 2, 5

[Link]
YouTube - IMPLearn

Spectrum of the down sampled signal

𝑀−1
1 𝑗
𝜔−2𝜋𝑘
𝑌 𝑒 𝑗𝜔 = ෍𝑋 𝑒 𝑀
𝑀
𝑘=0

• If the Fourier transform of the input-signal of a down samples in 𝑋 𝑒 𝑗𝜔 , then the

Fourier transform Y 𝑒 𝑗𝜔 of the output signal y(n) is a sum of M uniformly shifted and
stretched version of 𝑋 𝑒 𝑗𝜔 scaled by a factor of 1/M

[Link]
 YouTube - IMPLearn
𝜋 𝜋
Q) Consider a spectrum of input signal 𝑋 𝑒 𝑗𝜔 with a bandwidth of − to shown , when the signal is down
2 2
sampled by a factor D, sketch the spectrum of a down sampled signal for sampling rate reduction factor
D=2,3

𝟑𝝅 𝝅 𝝅 𝟑𝝅
-𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅
𝟐 𝟐 𝟐 𝟐

For M/D=2
When D=2
1
1 𝑗
𝜔−2𝜋𝑘
𝑌 𝑒 𝑗𝜔 = ෍𝑋 𝑒 2 Bandwidth =[Link] =2π
2
𝑘=0

1 𝑗
𝜔 1 𝑗
𝜔−2𝜋
𝑌 𝑒 𝑗𝜔 = 𝑋 𝑒 2 + 𝑋 𝑒 2
2 2

[Link]
 YouTube - IMPLearn
1 𝜔 1 𝑗
𝜔 1 𝑗
𝜔−2𝜋
𝑋 𝑒
𝑗
2 𝑌 𝑒 𝑗𝜔 = 𝑋 𝑒 2 + 𝑋 𝑒 2
2 2 2

𝟏
𝟐

𝟕𝝅 𝟓𝝅 𝟑𝝅 𝝅 𝝅 𝟑𝝅 𝟓𝝅 𝟕𝝅
−𝟒𝝅 − −𝟑𝝅 − −𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅 𝟑𝝅 𝟒𝝅
𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐
1 𝑗
𝜔−2𝜋
𝑋 𝑒 2
2

𝟕𝝅 𝟓𝝅 𝟑𝝅 𝝅 𝝅 𝟑𝝅 𝟓𝝅 𝟕𝝅
−𝟒𝝅 − −𝟑𝝅 − −𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅 𝟑𝝅 𝟒𝝅
𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐

𝟕𝝅 𝟓𝝅 𝟑𝝅 𝝅 𝝅 𝟑𝝅 𝟓𝝅 𝟕𝝅
−𝟒𝝅 − −𝟑𝝅 − −𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅 𝟑𝝅 𝟒𝝅
𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 [Link]
𝟐 𝟐
 YouTube - IMPLearn
𝜋 𝜋
Q) Consider a spectrum of input signal 𝑋 𝑒 𝑗𝜔 with a bandwidth of − to shown , when the signal is down
2 2
sampled by a factor D, sketch the spectrum of a down sampled signal for sampling rate reduction factor
D=2,3

𝟑𝝅 𝝅 𝝅 𝟑𝝅
-𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅
𝟐 𝟐 𝟐 𝟐

For M/D=3
When D=3
2
1 𝑗
𝜔−2𝜋𝑘
𝑌 𝑒 𝑗𝜔
= ෍𝑋 𝑒 3 Bandwidth =[Link] =3π
3
𝑘=0

1 𝑗
𝜔 1 𝑗
𝜔−2𝜋 1 𝑗
𝜔−4𝜋
𝑌 𝑒 𝑗𝜔 = 𝑋 𝑒 3 + 𝑋 𝑒 3 + 𝑋 𝑒 3
3 3 3

[Link]
 1 𝑗
𝜔
1 𝜔 1 𝜔−2𝜋 1 𝜔−4𝜋
YouTube - IMPLearn 𝑋 𝑒 3 𝑌 𝑒 𝑗𝜔
= 𝑋 𝑒
𝑗
3 + 𝑋 𝑒
𝑗
3 + 𝑋 𝑒
𝑗
3
3 𝟏 3 3 3
𝟑

𝟕𝝅 𝟓𝝅 𝟑𝝅 𝝅 𝝅 𝟑𝝅 𝟓𝝅 𝟕𝝅
−𝟒𝝅 − −𝟑𝝅 − −𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅 𝟑𝝅 𝟒𝝅
𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐
1 𝑗
𝜔−2𝜋
𝑋 𝑒 3
3

𝟕𝝅 𝟓𝝅 𝟑𝝅 𝝅 𝝅 𝟑𝝅 𝟓𝝅 𝟕𝝅
−𝟒𝝅 − −𝟑𝝅 − −𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅 𝟑𝝅 𝟒𝝅
𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐
1 𝑗
𝜔−4𝜋
𝑋 𝑒 3
3

𝟕𝝅 𝟓𝝅 𝟑𝝅 𝝅 𝝅 𝟑𝝅 𝟓𝝅 𝟕𝝅
−𝟒𝝅 − −𝟑𝝅 − −𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅 𝟑𝝅 𝟒𝝅
𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐

𝟕𝝅 𝟓𝝅 𝟑𝝅 𝝅 𝝅 𝟑𝝅 𝟓𝝅 𝟕𝝅
−𝟒𝝅 − −𝟑𝝅 − −𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅 [Link]
𝟑𝝅 𝟒𝝅
𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐
YouTube - IMPLearn

Anti-aliasing filters

• In order to avoid aliasing the input signal should be band limited to π/D for
decimation by a factor of D

Anti-aliasing
filter

𝒙 𝒏 𝒚 𝒏 = 𝒙 𝑴𝒏
h(n) ꜜM

[Link]
YouTube - IMPLearn

Up sampling
• It is the process of increasing the sampling rate by an integer factor I or L
• Up sampled signal of x(n) can be obtained by a factor of L by L-1 equally
spaced zeros between each pairs of samples

𝒏
𝒙 𝒏 𝒚 𝒏 =𝒙
ꜛL 𝑳

𝑥 𝑛 = 1, 2, 3, 4, 5 for I/L=3

𝑛
𝑥 = 1, 0, 0, 2, 0, 0, 3, 0, 0, 4,0,0
𝐿

[Link]
YouTube - IMPLearn

Spectrum of the up sampled signal

𝑌 𝑒 𝑗𝜔 = 𝑋 𝑒 𝑗𝜔𝐼

• The term 𝑋 𝑒 𝑗𝜔𝐼 is the frequency compressed version of 𝑋 𝑒 𝑗𝜔 by a factor I.

• If the frequency response is periodic with 2π , the 𝑋 𝑒 𝑗𝜔𝐼 will repeat I times in a period
of 0 to 2π in the spectrum of up sampled

[Link]
 YouTube - IMPLearn
Q) The spectrum of discrete time signal is shown below. Draw the spectrum of the signal if it is upscaled by
I=2,3

𝟕𝝅 𝟓𝝅 𝟑𝝅 𝝅 𝝅 𝟑𝝅 𝟓𝝅 𝟕𝝅
−𝟒𝝅 − −𝟑𝝅 − −𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅 𝟑𝝅 𝟒𝝅
𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐 𝟐

[Link]
 YouTube - IMPLearn
For I or L=2

𝑌 𝑒 𝑗𝜔 = 𝑋 𝑒 𝑗2𝜔

When I=2

Bandwidth =BW/I =π

𝟑𝝅 𝝅 𝝅 𝟑𝝅
-𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅
𝟐 𝟐 𝟐 𝟐

For I or L=3

𝑌 𝑒 𝑗𝜔 = 𝑋 𝑒 𝑗3𝜔
When I=3

Bandwidth =BW/I =2π/3

𝟑𝝅 𝝅 𝝅 𝟑𝝅 [Link]
-𝟐𝝅 − −𝝅 − 𝟎 𝝅 𝟐𝝅
𝟐 𝟐 𝟐 𝟐
YouTube - IMPLearn

Anti-imaging filters
• When up sampled by a factor of I, the output spectrum will have I images in each
𝜋
period with each image bandwidth to
𝐼
𝜋
• Since the frequency spectrum in the range to 0 to are unique and we have to
𝐼
filter the other images
• Hence the output of up samples is passed through a lowpass filter with band
𝜋
width
𝐼
• Since the lowpass filter is designed to avoid multiple images in output spectrum ,
it also called anti-imaging filter

Anti-imaging
filter
𝒙 𝒏 𝒚 𝒏
ꜛI h(n)

[Link]