4.

DIGITAL FILTER STRUCTURES

Rohit Kumar
May 2026

1
Q1. Obtain the Direct Form-II Realization of the
LTI System
Given Difference Equation:
3 3 1
y(n) = − y(n − 1) + y(n − 2) + y(n − 3) + x(n) + 3x(n − 1) + 2x(n − 2)
8 32 64

Step 1: Obtain Transfer Function

Taking Z-transform on both sides,
3 3 1
Y (z) = − z −1 Y (z) + z −2 Y (z) + z −3 Y (z) + X(z) + 3z −1 X(z) + 2z −2 X(z)
8 32 64
Rearranging,
 
3 3 1
Y (z) 1 + z −1 − z −2 − z −3 = X(z)(1 + 3z −1 + 2z −2 )
8 32 64

Hence,

Y (z) 1 + 3z −1 + 2z −2
H(z) = =
X(z) 1 + 83 z −1 − 32
3 −2
z − 64 1 −3
z

Comparing with

b0 + b1 z −1 + b2 z −2
H(z) =
1 + a1 z −1 + a2 z −2 + a3 z −3
we obtain

b0 = 1, b1 = 3, b2 = 2

3 3 1
a1 = , a2 = − , a3 = −
8 32 64

Step 2: Direct Form-II Realization

Let the intermediate sequence be w(n).
Then
3 3 1
w(n) = x(n) − w(n − 1) + w(n − 2) + w(n − 3)
8 32 64
Output equation:

y(n) = w(n) + 3w(n − 1) + 2w(n − 2)

2
Advantages of Direct Form-II
1. Requires minimum memory locations.
2. Canonical realization structure.

3. Number of delays equals order of denominator.

4. Widely used in DSP processors.

Block Diagram
+----(3)----+
| |
v |
x(n)--->(+)----->w(n)---->(+)----> y(n)
^ | ^
| | |
| z^-1--(3)--+
| |
| z^-1--(2)--+
|
-3/8
|
z^-1
|
+3/32
|
z^-2
|
+1/64
|
z^-3

Conclusion
The given third-order IIR system is realized using Direct Form-II structure with
only three delay elements, making it a canonical realization.

3
Q2. Consider FIR Filter with Lattice Coefficients

k1 = 0.65, k2 = −0.34, k3 = 0.8

Determine the direct form FIR coefficients and realize the system.

Introduction
A lattice structure represents an FIR filter using reflection coefficients ki . The
direct form coefficients can be obtained recursively from the lattice coefficients.

Step 1: First Stage

For

k1 = 0.65

A1 (z) = 1 + 0.65z −1

Hence
(1)
a0 = 1

(1)
a1 = 0.65

Step 2: Second Stage

Using

A2 (z) = A1 (z) + k2 z −2 A1 (z −1 )

Substituting

A2 (z) = (1 + 0.65z −1 ) − 0.34z −2 (1 + 0.65z)

= 1 + 0.429z −1 − 0.34z −2

Step 3: Third Stage

Using

A3 (z) = A2 (z) + k3 z −3 A2 (z −1 )

Substituting

4
 k3 = 0.8

After simplification,

A3 (z) = 1 + 0.157z −1 + 0.003z −2 + 0.8z −3

Therefore,

H(z) = 1 + 0.157z −1 + 0.003z −2 + 0.8z −3

Direct Form Coefficients

h(0) = 1

h(1) = 0.157

h(2) = 0.003

h(3) = 0.8

Difference Equation

y(n) = x(n) + 0.157x(n − 1) + 0.003x(n − 2) + 0.8x(n − 3)

Direct Form Realization

x(n) ----->o----->z^-1----->z^-1----->z^-1
| | | |
(1) (0.157) (0.003) (0.8)
| | | |
+---------+---------+---------+
|
v
y(n)

Conclusion
The direct form FIR coefficients corresponding to the given lattice coefficients
are

H(z) = 1 + 0.157z −1 + 0.003z −2 + 0.8z −3

which can be realized using a tapped-delay line FIR structure.

5
Q3. Obtain the Parallel Realization of the LTI
System

(1 + z −1 )(1 + 2z −1 )
H(z) = 1 −1
(1 + 8z + 21 z −1 )(1 − 14 z −1 )
)(1

Introduction
Parallel realization is obtained by expressing the transfer function as a sum of
first-order sections using partial fraction expansion.

Step 1: Expand Numerator

(1 + z −1 )(1 + 2z −1 ) = 1 + 3z −1 + 2z −2

Thus,

1 + 3z −1 + 2z −2
H(z) = 1 −1
(1 + 8z )(1+ 21 z −1 )(1 − 14 z −1 )

Step 2: Partial Fraction Expansion

Assume
A B C
H(z) = + +
1 + 81 z −1 1 + 21 z −1 1 − 14 z −1

Multiplying both sides by the denominator and comparing coefficients gives

24
A=
5

B = −2

9
C=−
5
Hence
24 9
5 2 5
H(z) = − −
1 + 81 z −1 1+ 1 −1
2z 1− 1 −1
4z

6
Parallel Realization
The system consists of three first-order sections connected in parallel.

H(z) = H1 (z) + H2 (z) + H3 (z)

where
24
5
H1 (z) =
1 + 18 z −1

2
H2 (z) = −
1 + 12 z −1

9 1
H3 (z) = − 1 − z −1
5 4

Block Diagram

H1(z)
x(n) ---------> [ ] --------\
\
H2(z) >---- (+) ---> y(n)
x(n) ---------> [ ] --------/
\
\
H3(z) --------\
x(n) ---------> [ ] ----------/

Advantages
1. Low sensitivity to coefficient quantization.
2. Easy implementation of higher-order systems.
3. Each branch can be analyzed independently.

4. Better numerical stability.

Conclusion
The given transfer function is realized as the parallel combination of three first-
order IIR sections obtained using partial fraction expansion.

7
Q4. Realize the Cascade Form Realization of the
FIR System with Minimum Number of Multipli-
ers
  
1 1 −1 1 −2 1 1 −1 1 −2
H(z) = + z + z + z + z
5 2 5 7 4 7

Introduction
Cascade realization is obtained by expressing the overall FIR transfer function as
the product of lower-order FIR sections. Each section is realized independently
and connected in series.

Given Sections
1 1 −1 1 −2
H1 (z) = + z + z
5 2 5

1 1 −1 1 −2
H2 (z) = + z + z
7 4 7
Therefore,

H(z) = H1 (z)H2 (z)

Observation
Both sections are symmetric:
1
h1 (0) = h1 (2) =
5

1
h2 (0) = h2 (2) =
7
Hence linear-phase implementation can be used to reduce the number of
multipliers.

Cascade Realization
First section:
1 1 1
y1 (n) = x(n) + x(n − 1) + x(n − 2)
5 2 5
Second section:
1 1 1
y(n) = y1 (n) + y1 (n − 1) + y1 (n − 2)
7 4 7

8
Block Diagram

x(n) ---> [ H1(z) ] ---> y1(n) ---> [ H2(z) ] ---> y(n)

where
1 1 −1 1 −2
H1 (z) = + z + z
5 2 5
and
1 1 −1 1 −2
H2 (z) = + z + z
7 4 7

Advantages
1. Reduced multiplier requirement.
2. Modular realization.

3. Easy implementation.
4. Improved numerical accuracy.

Conclusion
The FIR system is realized as two second-order FIR sections connected in cas-
cade. This realization requires fewer multipliers than direct implementation.

9
Q5. Realize the Given IIR System in Cascade
Form
1 + 0.25z −1
H(z) =
(1 − 2z −1 + 0.25z −2 )(1 − 3z −1 + 0.2z −2 )

Introduction
Cascade realization is obtained by expressing the transfer function as the prod-
uct of second-order sections.

Step 1: Express Transfer Function

Given
H(z) = H1 (z)H2 (z)
where
1 + 0.25z −1
H1 (z) =
1 − 2z −1 + 0.25z −2
and
1
H2 (z) =
1 − 3z −1 + 0.2z −2

Step 2: Difference Equations

For first section,

Y1 (z) = H1 (z)X(z)
Therefore,

y1 (n) − 2y1 (n − 1) + 0.25y1 (n − 2) = x(n) + 0.25x(n − 1)

Hence,

y1 (n) = 2y1 (n − 1) − 0.25y1 (n − 2) + x(n) + 0.25x(n − 1)

For second section,

Y (z) = H2 (z)Y1 (z)

Thus,

y(n) − 3y(n − 1) + 0.2y(n − 2) = y1 (n)

or
y(n) = 3y(n − 1) − 0.2y(n − 2) + y1 (n)

10
Cascade Structure

x(n)
|
v
+---------+
| H1(z) |
+---------+
|
v
y1(n)
|
v
+---------+
| H2(z) |
+---------+
|
v
y(n)

Advantages
1. Less coefficient sensitivity.
2. Easier implementation.
3. Better numerical stability.

4. Preferred for higher-order IIR systems.

Conclusion
The given fourth-order IIR filter is realized as a cascade connection of two
second-order sections.

11
Q6. Consider FIR Filter

H(z) = 1 + 2.88z −1 + 3.4048z −2 + 1.74z −3 + 0.4z −4

Sketch the lattice realization of the filter.

Introduction
Lattice realization is a modular structure used for FIR filters. It is based on
reflection coefficients k1 , k2 , . . . , kM .
Advantages:

• Numerically stable.
• Modular structure.
• Suitable for adaptive filtering.
• Low sensitivity to coefficient quantization.

Given Transfer Function

A4 (z) = 1 + 2.88z −1 + 3.4048z −2 + 1.74z −3 + 0.4z −4

Using step-down recursion (as shown in module notes), reflection coefficients

are obtained as

k4 = 0.4

k3 = 0.688

k2 = 1.218

k1 = 0.79

Lattice Coefficients

k1 = 0.79

k2 = 1.218

k3 = 0.688

k4 = 0.4

12
Four-Stage Lattice Structure

x(n)
|
|---- Stage 1 ----|---- Stage 2 ----|
| k1=0.79 k2=1.218
|
|---- Stage 3 ----|---- Stage 4 ----|
| k3=0.688 k4=0.4
|
v
y(n)

Each stage contains:

1. One delay element z^-1
2. Forward path
3. Backward path
4. Reflection coefficient ki

General Equations
Forward prediction:

Am (n) = Am−1 (n) + km Bm−1 (n − 1)

Backward prediction:

Bm (n) = km Am−1 (n) + Bm−1 (n − 1)

for

m = 1, 2, 3, 4

Advantages of Lattice Realization

1. Stable implementation.
2. Modular architecture.

3. Low round-off noise.

4. Easy coefficient updating.
5. Widely used in adaptive DSP systems.

13
Conclusion
The FIR filter is realized using a four-stage lattice structure with reflection
coefficients

k1 = 0.79, k2 = 1.218, k3 = 0.688, k4 = 0.4

which provides an efficient and numerically robust implementation.

14
Q7. Obtain the Cascade Form Realization for the
IIR System

y(n) = −0.1y(n − 1) + 0.2y(n − 2) + 3x(n) + 3.6x(n − 1) + 0.6x(n − 2)

Introduction
Cascade realization is obtained by expressing the transfer function as a prod-
uct of lower-order sections. It offers improved numerical stability and reduced
coefficient sensitivity.

Step 1: Obtain Transfer Function

Taking Z-transform,

Y (z) + 0.1z −1 Y (z) − 0.2z −2 Y (z) = 3X(z) + 3.6z −1 X(z) + 0.6z −2 X(z)

Hence,

Y (z) 3 + 3.6z −1 + 0.6z −2

H(z) = =
X(z) 1 + 0.1z −1 − 0.2z −2

Factor numerator:

3 + 3.6z −1 + 0.6z −2 = 3(1 + z −1 )(1 + 0.2z −1 )

Factor denominator:

1 + 0.1z −1 − 0.2z −2 = (1 + 0.5z −1 )(1 − 0.4z −1 )

Therefore,

1 + z −1 1 + 0.2z −1
  
H(z) = 3
1 + 0.5z −1 1 − 0.4z −1

Cascade Sections

1 + z −1
H1 (z) =
1 + 0.5z −1

1 + 0.2z −1
H2 (z) =
1 − 0.4z −1

H(z) = 3H1 (z)H2 (z)

15
Block Diagram

x(n)
|
v
+---------+
| H1(z) |
+---------+
|
v
+---------+
| H2(z) |
+---------+
|
v
(×3)
|
v
y(n)

Advantages
1. Better numerical stability.
2. Easier implementation.
3. Less coefficient sensitivity.

4. Suitable for higher-order IIR systems.

Conclusion
The given second-order IIR system is realized as the cascade of two first-order
sections.

16
Q8. Obtain the Direct Form-I and Direct Form-II
Structures
Given

y(n) = 0.5y(n − 1) − 0.25y(n − 2) + x(n) + 0.4x(n − 1)

Step 1: Transfer Function

Taking Z-transform,

Y (z) − 0.5z −1 Y (z) + 0.25z −2 Y (z) = X(z) + 0.4z −1 X(z)

Thus,

Y (z) 1 + 0.4z −1
H(z) = =
X(z) 1 − 0.5z −1 + 0.25z −2

Direct Form-I Realization

General form:
b0 + b1 z −1
H(z) =
1 + a1 z −1 + a2 z −2
Comparing,

b0 = 1

b1 = 0.4

a1 = −0.5

a2 = 0.25

Difference Equation

y(n) = 0.5y(n − 1) − 0.25y(n − 2) + x(n) + 0.4x(n − 1)

17
Direct Form-I Structure

Input Delay Line

x(n)--->z^-1---->
| |
1 0.4
| |
+---->()

Output Feedback

y(n)<---z^-1<---z^-1
| |
0.5 -0.25
| |
+---->()

Direct Form-II Realization

Direct Form-II combines input and output delay lines.
Intermediate variable:

w(n) = x(n) + 0.5w(n − 1) − 0.25w(n − 2)

Output equation:

y(n) = w(n) + 0.4w(n − 1)

Direct Form-II Structure

x(n)-->()--->w(n)--->z^-1----->z^-1
| | |
0.5 0.4 -0.25
| |
+--------+
|
v
y(n)

18
Comparison
Parameter Direct Form-I Direct Form-II
Delay Elements 3 2
Memory Required More Less
Complexity Higher Lower
Canonical Form No Yes

Conclusion
Direct Form-II realization requires fewer delay elements and is therefore pre-
ferred in practical DSP implementations.

19
Quick Revision Notes for Viva and Theory Ques-
tions
1. Direct Form-I Structure
Definition:
Direct implementation of the difference equation using separate delay lines
for input and output sequences.
Features:

• Separate feedforward and feedback sections.

• Easy to understand.
• Requires more memory.
• Non-canonical realization.

Advantage:
Simple implementation.
Disadvantage:
Requires large number of delay elements.

2. Direct Form-II Structure

Definition:
Canonical realization obtained by combining delay elements of Direct Form-
I.
Features:

• Minimum memory requirement.

• Widely used in DSP processors.
• Canonical structure.

Advantage:
Uses minimum number of delays.
Disadvantage:
More sensitive to round-off noise.

3. Cascade Structure
Definition:
Transfer function is factorized into first-order or second-order sections con-
nected in series.
Advantages:

• Better numerical stability.

20
 • Lower coefficient sensitivity.
• Easy debugging.

Applications:
Higher-order IIR filters.

4. Parallel Structure
Definition:
Transfer function is decomposed into partial fractions and realized as parallel
branches.
Advantages:

• Independent processing of sections.

• Reduced quantization effects.

• Easier implementation of large systems.

5. Lattice Structure
Definition:
FIR realization using reflection coefficients.
Advantages:

• Numerically stable.
• Modular structure.
• Suitable for adaptive filtering.
• Low sensitivity to coefficient quantization.

Applications:
Speech processing, adaptive filters, linear prediction.

21