Analysis of Rational

Systems/Filters

Chirayu Athalye

DT-LTI Systems

Linear Shift-Invariant
Difference Equations

Frequency Response
Analysis of Rational Systems/Filters Types of FIR Filters

References

Chirayu D. Athalye

Department of Electrical & Electronics Engineering,

BITS Pilani, K K Birla Goa Campus.

2024
 Analysis of Rational
Systems/Filters
Transfer Function and Frequency Response
Chirayu Athalye
1 Consider a Discrete Time LTI system (DT-LTI) with the impulse response h : Z → C.
DT-LTI Systems
2 The output of this system to an input sequence x : Z → C is given by Linear Shift-Invariant
Difference Equations
y [n] = (h ∗ x)[n] ∀n ∈ Z. Frequency Response

3 Therefore, using the convolution property of z-Transform, we have Types of FIR Filters

 References


Y (z) = H(z) X (z) ∀z ∈ ROC Z y [n] ,

where H(z) is known as the System/Transfer Function of the DT-LTI system.

4 Similarly, using the convolution property of DTFT, we have
Y (e iω ) = H(e iω ) X (e iω ),
where H(e iω ) is known as the Frequency Response of the DT-LTI system.

Question
Consider two DT-LTI systems with impulse responses h1 : Z → C and h2 : Z → C. Find the
Transfer Function (TF) of the resultant system if these two systems are connected in
1 Cascade 2 Parallel.
 Analysis of Rational
Systems/Filters
Theorem (Equivalent Conditions for Causality) Chirayu Athalye

For a DT-LTI system, the following are equivalent: DT-LTI Systems

1 The system is causal. Linear Shift-Invariant
Difference Equations
2 The impulse response h : Z → C is a causal signal, i.e., h[n] = 0 for n < 0. Frequency Response
3 ROC of H(z) is of the form {z ∈ Ce : ρ1 < |z| ≤ ∞}, where ρ1 ≥ 0. Types of FIR Filters

References

For a causal DT-LTI system, if TF is rational, then does it have to be proper?

Theorem (Equivalent Conditions for BIBO Stability)

For a DT-LTI system, the following are equivalent:
1 The system is BIBO stable.
∞
X
2 The impulse response h : Z → C is absolutely summable, i.e., |h[n]| < ∞.
n=−∞
3 ROC of H(z) contains the unit circle.
Whereas, a causal DT-LTI system with (proper) rational transfer function H(z) is BIBO stable
if and only if all the poles of H(z) are inside the open unit disc {z ∈ C : |z| < 1}.
 Analysis of Rational
Systems Described by Linear Constant-Coefficient Difference Equations Systems/Filters

Chirayu Athalye
N
X M
X
y [n] + ak y [n − k] = bk x[n − k] DT-LTI Systems

k=1 k=0 Linear Shift-Invariant

! ! Difference Equations
N M
X X Frequency Response
∴ 1+ ak z −k Y (z) = bk z −k X (z)
Types of FIR Filters
k=1 k=0
References

Y (z)
Subsequently, H(z) = and ROC can be inferred from the additional information about
X (z)
the system such as stability or causality.
Remark: The unilateral z-Transform is used to solve linear constant-coefficient difference
equations with nonzero initial conditions.

Homework
1 Find the impulse response of an unstable system governed by
1
y [n] − y [n − 1] = x[n] + x[n − 1].
3
2 For the initial conditions y [−1] = 1 and y [−2] = 2, and x[n] := (0.5)n u[n], solve the
following difference equation: y [n] − 5y [n − 1] + 6y [n − 2] = 2x[n] − 5x[n − 1].
 Analysis of Rational
Systems/Filters

Chirayu Athalye

DT-LTI Systems

Linear Shift-Invariant
Difference Equations

Frequency Response

Types of FIR Filters

References

(a) Unwrapped (Continuous) Phase

(b) Wrapped Phase

]H(e iω ) = ARG H(e iω ) + 2πkω .

 Analysis of Rational
Systems/Filters
Linear Phase Systems/Filters
Chirayu Athalye

DT-LTI system with frequency response H(e iω ) is said to be linear phase if

DT-LTI Systems

]H(e iω ) = −αω ∀ω ∈ [−π, π), (where α ∈ R). Linear Shift-Invariant

Difference Equations

Remark: Zero phase systems are a special class of linear phase systems. Frequency Response

Types of FIR Filters

References
Generalized Linear Phase Systems/Filters
DT-LTI system is said to be generalized linear phase if its frequency response is of the form:
H(e iω ) = A(e iω ) e −i(αω+β) ∀ω ∈ [−π, π),
iω
where A(e ) is a real-valued (need not be nonnegative) function of ω, and α, β ∈ R.

Remark (Generalized Linear Phase Filters)

1 If for some ω1 ∈ [−π, π), we have A(e iω1 ) < 0, then A(e iω1 ) = |A(e iω1 )| e ±iπ .
2 The phase response of a generalized linear phase system is a piecewise affine function:
]H(e iω ) = −αω − β(ω) ∀ω ∈ [−π, π),
where α ∈ R and β(·) is a piecewise constant function on [−π, π).
 Analysis of Rational
Systems/Filters

Chirayu Athalye
Group Delay and Phase Delay
DT-LTI Systems
dφ τg (·) and τp (·) quantify how signal’s
1 Group Delay: τg (ω) := −φ0 (ω) = − (ω). Linear Shift-Invariant
Difference Equations
dω frequency components are delayed in
Frequency Response
φ(ω) time when passing through an LTI
2 Phase Delay: τp (ω) := − . Types of FIR Filters
ω system.
References

Remark
1 For a linear phase system, the group delay τg (·) is a constant function.
2 Whereas, for a generalized linear phase system, τg (·) is constant almost1 everywhere.
3 For linear phase systems, group delay = phase delay.

Consider a DT-LTI system with the impulse response h : Z → C. For the input sequence
x[n] := Aenv [n] cos(ω1 n + θ), where Aenv [·] > 0 and ω1 , θ ∈ R, the output can be approximated
as:
y [n] ≈ H(e iω1 ) Aenv [n − τg (ω1 )] cos ω1 (n − τp (ω1 )) + θ


∀n ∈ Z.

1
(excluding countable discontinuities)
 Analysis of Rational
Systems/Filters

Chirayu Athalye
Bounded Real Transfer Functions
DT-LTI Systems
A causal stable transfer function H(z) with real-coefficients is said to be bounded real if Linear Shift-Invariant
Difference Equations
iω
|H(e )| ≤ 1 ∀ω ∈ R. Frequency Response

Types of FIR Filters

Remark: Any causal stable TF with real-coefficients can be converted to a bounded real TF. References

Minimum-Phase, Maximum-Phase, and Mixed-Phase Transfer Functions

A causal stable transfer function H(z) is said to be:
1 Minimum-phase if all the zeros of H(z) are inside the open unit disc.
2 Maximum-phase if all the zeros of H(z) are outside the open unit disc.
3 Mixed-phase if it is neither minimum-phase nor maximum-phase.

Remark
1
1 A minimum-phase system Hmin (z) has a causal and stable inverse system Hmin (z)
.
 Analysis of Rational
Systems/Filters

Chirayu Athalye

DT-LTI Systems

Linear Shift-Invariant
Difference Equations

Frequency Response

Types of FIR Filters

References

z +b bz + 1
(a) H1 (z) = (b) H2 (z) =
z +a z +a

(c) Phase Responses for a = 0.5 and b = −0.8.

 Analysis of Rational
Systems/Filters

Chirayu Athalye

DT-LTI Systems

Linear Shift-Invariant
Difference Equations

Frequency Response

Types of FIR Filters

ha [n] = {1.56, 2.89, 3.39, 2.19, 0.81},
References
hb [n] = {0.81, 2.19, 3.39, 2.89, 1.56},
(a) Minimum Phase System (b) Maximum Phase System
hc [n] = {1.26, 2.51, 3.5, 2.58, 1},
hd [n] = {1, 2.58, 3.5, 2.51, 1.26}.

Check the magnitude and phase

plots using MATLAB.

(c) Mixed Phase System (d) Mixed Phase System

Figure: Systems with the Same Magnitude Response but
Different Phase Responses.
 Analysis of Rational
Systems/Filters
Consider a causal FIR system with the impulse response of length (M + 1).
M Chirayu Athalye
X
Its TF is given by H(z) = h[n]z −n . Thus, the order of the system is M. DT-LTI Systems
n=0 Linear Shift-Invariant
Difference Equations

Frequency Response
Type 1 FIR System (Impulse Response: Odd Length and Symmetric)
Types of FIR Filters

Type 1 FIR system has a symmetric impulse response References

h[n] = h[M − n], 0 ≤ n ≤ M,

where M is an even integer. Its frequency response is
M
X
H(e iω ) = h[n]e −iωn
n=0


iωM iωM iωM M
= e− 2 h[0]e 2 + h[M]e − 2 + h[1]e iω 2 −1


M
+ h[M − 1]e −iω 2 −1 + · · · · +h[ M2 ] . Figure: Type 1 FIR with M = 6.

Using the symmetry condition, we get

M/2
(
iω − iωM
X h[ M2 ], for m = 0
H(e ) = e 2 a[m] cos(ωm), where a[m] =
m=0
2h[ M2 − m], for m = 1, . . . , M
2
.
 Analysis of Rational
Type 2 FIR System (Impulse Response: Even Length and Symmetric) Systems/Filters

Chirayu Athalye

Type 2 FIR system has a symmetric impulse response DT-LTI Systems

h[n] = h[M − n], 0 ≤ n ≤ M, Linear Shift-Invariant

Difference Equations

where M is an odd integer. Its frequency response is Frequency Response

M Types of FIR Filters

X
H(e iω ) = h[n]e −iωn References

n=0
(M+1)/2
iωM X
= e− b[m] cos ω(m − 12 ) ,


2

m=1
Figure: Type 2 FIR with M = 5.
where b[m] = 2h[ M+1
2
− m] for m = 1, . . . , M+1
2
.

Remark: Note that since cos π(m − 12 ) = 0 for all m ∈ Z, we have H(e iπ ) = 0; therefore,

Type 2 FIR system cannot be used to design a highpass filter.

For symmetric FIR systems (Types 1 and 2),

XM M
X 0
X
H(z) = h[n] z −n = h[M − n] z −n = h[`] z `−M = z −M H(z −1 ).
n=0 n=0 `=M
 Analysis of Rational
Systems/Filters

Chirayu Athalye

DT-LTI Systems

Linear Shift-Invariant
(a) Impulse Response (a) Impulse Response Difference Equations

Frequency Response

Types of FIR Filters

References

(b) Magnitude Response (b) Magnitude Response

(c) Phase Response (c) Phase Response

Figure: Type 1 FIR System. Figure: Type 2 FIR System.
 Analysis of Rational
Type 3 FIR Filter (Impulse Response: Odd Length and Antisymmetric) Systems/Filters

Chirayu Athalye

Type 3 FIR filter has an antisymmetric impulse response DT-LTI Systems

h[n] = −h[M − n], 0 ≤ n ≤ M, Linear Shift-Invariant

Difference Equations

where M is an even integer. Its frequency response is Frequency Response

XM Types of FIR Filters

H(e iω ) = h[n]e −iωn References

n=0
M/2
iωM X
= ie − 2 c[m] sin(ωm),
m=1

where c[m] = 2h[ M2 − m] for m = 1, . . . , M

2
. Figure: Type 3 FIR with M = 6.

Remark: Since sin(0m) = sin(πm) = 0 for all m ∈ Z, we have H(e i0 ) = H(e iπ ) = 0.

Therefore, Type 3 FIR system cannot be used to design a lowpass or highpass filter.

For antisymmetric FIR systems (Types 3 and 4 ),

XM M
X 0
X
H(z) = h[n] z −n = − h[M − n] z −n = − h[`] z `−M = −z −M H(z −1 ).
n=0 n=0 `=M
 Analysis of Rational
Systems/Filters
Type 4 FIR Filter (Impulse Response: Even Length and Antisymmetric)
Chirayu Athalye

DT-LTI Systems
Type 4 FIR filter has an antisymmetric impulse response
Linear Shift-Invariant
Difference Equations
h[n] = −h[M − n], 0 ≤ n ≤ M,
Frequency Response
where M is an odd integer. Its frequency response is Types of FIR Filters
M
X References
H(e iω ) = h[n]e −iωn
n=0
(M+1)/2
iωM X
= ie − d[m] sin ω(m − 12 ) ,


2

m=1

where d[m] = 2h[ M+1

2
− m] for m = 1, . . . , M+1
2
.
Figure: Type 4 FIR with M = 5.
i0
Remark: Note that H(e ) = 0; therefore, Type 4 FIR system cannot be used to design a
lowpass filter.

Do all FIR systems have generalized linear phase response?

No; FIR systems with generalized linear phase response can be categorized into the above four types.
 Analysis of Rational
Systems/Filters

Chirayu Athalye

DT-LTI Systems

Linear Shift-Invariant
(a) Impulse Response (a) Impulse Response Difference Equations

Frequency Response

Types of FIR Filters

References

(b) Magnitude Response (b) Magnitude Response

(c) Phase Response (c) Phase Response

Figure: Type 3 FIR System. Figure: Type 4 FIR System.

 Analysis of Rational
Systems/Filters
Remark (Symmetric FIR Systems (Types 1 and 2))
Chirayu Athalye
1 For symmetric FIR systems, we have: H(z) = z −M H(z −1 ).
DT-LTI Systems
2 Therefore, zeros of symmetric FIR systems occur in reciprocal pairs. Linear Shift-Invariant
Difference Equations
3 Notice that Type 2 FIR systems, where M is odd, have zero at z = −1.
Frequency Response

Types of FIR Filters

Remark (Antisymmetric FIR Systems (Types 3 and 4 )) References

−M −1
1 For antisymmetric FIR systems, we have: H(z) = −z H(z ).
2 Therefore, zeros of symmetric FIR systems occur in reciprocal pairs.
3 Notice that antisymmetric FIR systems have zero at z = 1. In addition, Type 3 FIR
systems (where M is even) have zero at z = −1.

Remark (Zeros of Generalized Linear Phase FIR Systems)

1 For generalized linear phase FIR systems, zeros occur in reciprocal pairs.
2 Furthermore, if the impulse response is real-valued, then zeros occur in conjugate
reciprocal quadruples.
3 However, for zeros on the unit circle and real-axis, conjugate reciprocal quadruples is
basically the reciprocal pair.
 Analysis of Rational
Systems/Filters

Chirayu Athalye

DT-LTI Systems

Linear Shift-Invariant
Difference Equations

Frequency Response

Types of FIR Filters

References

(a) Type 1 FIR (b) Type 2 FIR

(c) Type 3 FIR (d) Type 4 FIR

Figure: Plots of Zeros of Generalized Linear Phase FIR Systems with R-valued Impulse Response.
 Analysis of Rational
Systems/Filters

Chirayu Athalye

DT-LTI Systems

Linear Shift-Invariant
Difference Equations
References
Frequency Response

A.V. Oppenheim, R.W. Schafer, and J.R. Buck, Discrete-Time Signal Processing (2nd edition), Upper Saddle Types of FIR Filters

River, NJ, USA: Prentice Hall, 1999. References

S.K. Mitra, Digital Signal Processing: A Computer-Based Approach (4th edition), NY, USA: McGraw-Hill,
2010.

Readings
1 [Discrete-Time Signal Processing (Oppenheim et al.), Chapter 5].
2 [Digital Signal Processing (S.K. Mitra), Chapter 7].