Download as pdf or txt
Download as pdf or txt
You are on page 1of 75

Chapter 5

Transform Analysis of Linear Time-Invariant Systems

Der-Feng Tseng

Department of Electrical Engineering


National Taiwan University of Science and Technology
(through the courtesy of Prof. Peng-Hua Wang of National Taipei University)

February 19, 2015

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 1 / 75


Outline

1 5.1 Frequency Response

2 5.2 System Functions

3 5.3 Frequency Response

4 5.4 Magnitude and Phase

5 5.5 All-Pass System

6 5.6 Minimum-Phase System

7 5.7 Generalized Linear Phase

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 2 / 75


About The figures

All the figures are from Discrete-Time Signal Processing, 2e, by


Oppenheim, Schafer, and Buck,Prentice Hall, Inc.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 3 / 75


5.1 Frequency Response

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 4 / 75


Frequency Response
The Fourier transforms of the system input and output are related by

Y (ejω ) = H(ejω )X(ejω )

With the frequency response expressed in the polar form, we have

|Y (ejω )| = |H(ejω )| · |X(ejω )|, ∢Y (ejω ) = ∢H(ejω ) + ∢X(ejω )

|H(ejω )| is the magnitude response or the gain of the system.


∢H(ejω ) is the phase response or the phase shift of the system.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 5 / 75


Ideal Lowpass Filter
The frequency response of the ideal lowpass filter is
(
jω 1, |ω| < ωc
Hlp (e ) =
0, ωc < |ω| ≤ π.

The corresponding impulse response is


sin ωc n
hlp [n] = , −∞ < n < ∞.
πn
The ideal lowpass filter is noncausal. It is NOT computational
realizable.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 6 / 75


Ideal Highpass Filter
The frequency response of the ideal highpass filter is
(
jω 0, |ω| < ωc
Hhp (e ) =
1, ωc < |ω| ≤ π.

Since Hhp (ejω ) = 1 − Hlp (ejω ), the corresponding impulse response


is
sin ωc n
hhp [n] = δ[n] − hlp [n] = δ[n] − , −∞ < n < ∞.
πn

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 7 / 75


Delay
The impulse response and the frequency response of the ideal delay
filter is
hid [n] = δ[n − nd ], Hid (ejω ) = e−jωnd .
The frequency response of the ideal lowpass filter with delay nd is
(
e−jωnd , |ω| < ωc
Hlp (ejω ) =
0, ωc < |ω| ≤ π.

The corresponding impulse response is


sin ωc (n − nd )
hlp [n] = , −∞ < n < ∞.
π(n − nd )

Ideal delay ⇒ Linear phase

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 8 / 75


Group Delay
The linearity of the phase response can be measured by the group
delay defined by
d
τ (ω) = grd[H(ejω )] = − ∢H(ejω ).

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 9 / 75


Example 5.1

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 10 / 75


Example 5.1

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 11 / 75


Example 5.1

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 12 / 75


5.2 System Functions

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 13 / 75


Linear Difference Equations
The linear constant-coefficient difference equations
N
X M
X
ak y[n − k] = bm x[n − m]
k=0 m=0

We can calculate the output recursively by


N M
!
1 X X
y[n] = − ak y[n − k] + bm x[n − m]
a0 m=0
k=1

We want to characterize the LTI systems by the linear difference


equation.
Take z-transform, we have
M
X
bm z −m
Y (z)
H(z) = = m=0
X(z) N
X
ak z −k
k=0

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 14 / 75


Example 5.2
Suppose that the system function of a LTI system is
(1 + z −1 )2
H(z) = 1 −1 ,
(1 − 2z + 34 z −1 )
)(1

find the difference equation that is satisfied by the input and output of
this system.

Solution. Since
1 + 2z −1 + z −2 Y (z)
H(z) = = ,
1 + 14 z −1 − 83 z −2 X(z)
we have

y[n] + 14 y[n − 1] − 38 y[n − 2] = x[n] + 2x[n − 1] + x[n − 2]

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 15 / 75


ROC

Causal ⇒ ROC is outside the outermost pole


Anti-causal ⇒ ROC is inside the innermost pole.
Stable ⇒ ROC include the unit circle.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 16 / 75


Example 5.3
Consider the LTI system with input and output related through the
difference equation

y[n] − 25 y[n − 1] + y[n − 2] = x[n].

Discuss the stability, causality and ROC.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 17 / 75


Example 5.3
Solution.
1 1
H(z) = =
1 − 25 z −1 + z −2 (1 − 12 z −1 )(1 − 2z −1 )

ROC = {z : |z| > 2}: causal, not stable.


ROC = {z : |z| < 21 }: not causal, not stable.
1
ROC = {z : 2 < |z| < 2}: not causal, stable.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 18 / 75


Inverse System
Given H(z), the inverse system Hi (z) satisfies

H(z)Hi (z) = 1

⇒ h[n] ∗ hi [n] = δ[n]


⇒ Regions of convergence of H(z) and Hi (z) must
overlap.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 19 / 75


Example 5.4
Let H(z) be
1 − 0.5z −1
H(z) = .
1 − 0.9z −1
Find the inverse system Hi (z) and discuss the ROC.

Solution.
1 − 0.9z −1
Hi (z) =
1 − 0.5z −1

ROC of Hi (z) is either |z| > 0.5 or |z| < 0.5.

If the ROC of H(z) is |z| > 0.9, the ROC of Hi (z) is |z| > 0.5.
If the ROC of H(z) is |z| < 0.9, we have two inverse systems.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 20 / 75


Example 5.5

Let H(z) be
z −1 − 0.5
H(z) = , |z| > 0.9
1 − 0.9z −1
Find the inverse system Hi (z) and discuss the ROC.

Solution.
1 − 0.9zz −1 −2 + 1.8z −1
Hi (z) = =
z − 0.5
−1 1 − z −1
ROC of Hi (z) is either |z| > 0.5 or |z| < 0.5.

Both ROCs of |z| > 2 and |z| < 2 overlap with |z| > 0.9, so both are
valid inverse system.
If the ROC of Hi (z) is |z| < 2,
hi1 [n] = 2n+1 u[−n − 1] − 1.8 · 2n−1 u[−n]. (stable, noncausal)
If the ROC of Hi (z) is |z| > 2,
hi2 [n] = −2n+1 u[n] + 1.8 · 2n−1 u[n − 1]. (unstable, causal)
Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 21 / 75
Impulse Response
Suppose a causal system function can be expressed in the form
M−N N
X X Ak
H(z) = Br z −r +
r=0
1 − dk z −1
k=1

Its impulse response is


M−N
X N
X
h[n] = Br δ[n − r] + Ak dnk u[n]
r=0 k=1

If at least one nonzero pole of H(z) is not canceled by a zero, one


terms of the form Ak dnk u[n] exists. ⇒ infinite impulse response (IIR)
system.
H(z) has no poles except at z = 0 ⇒ finite impulse response (FIR)
system.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 22 / 75


Example 5.6
Consider a causal system whose input and output satisfy the difference
equation
y[n] − ay[n − 1] = x[n]
The system function is
1
H(z) =
1 − az −1

The ROC is |z| > |a|, the


condition for stability is |a| < 1.
h[n] = an u[n]

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 23 / 75


Example 5.7
Consider the following impulse response and transfer function
(
an , 0 ≤ n ≤ M 1 − aM+1 z −(M+1)
h[n] = H(z) =
0, otherwise, 1 − az −1

The zeros are


aej2πk/(M +1) , k = 0, 1, . . . , M.
The pole at z = a is canceled by
a zero.
The difference
PM kequation is
y[n] = k=0 a x[n − k]
An equivalent equation is
y[n] − ay[n − 1] =
x[n] − aM +1 x[n − M − 1]

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 24 / 75


5.3 Frequency Response

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 25 / 75


Linear Difference Equations
The frequency response to a system function
M
X M
X
bm z −m bm e−jωm
m=0 m=0
H(z) = N
is H(ejω ) = N
X X
ak z −k e−jωk
k=0 k=0

The frequency response can be factored as


M
Y
  (1 − cm e−jω )
b0 m=1
H(ejω ) =
a0 YN
(1 − dk e−jω )
k=1

Magnitude response: |H(ejω )|


Gain in dB = 20 log10 |H(ejω )|.
Attenuation in dB = −20 log10 |H(ejω )| = − Gain in dB.
Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 26 / 75
Phase Response
Phase response
M
X N
X
∢H(ejω ) = ∢[b0 /a0 ] + ∢[1 − cm e−jω ] − ∢[1 − dk e−jω ]
m=1 k=1

Group delay
M N
X d X d
grd[H(ejω )] = arg[1 − cm e−jω ] − arg[1 − dk e−jω ]
m=1
dω dω
k=1
M 2 N
X |dm | − ℜ{dm e −jω
} X |ck |2 − ℜ{ck e−jω }
= 2

m=1
1 + |dm | − 2ℜ{dm e −jω } 1 + |ck |2 − 2ℜ{ck e−jω }
k=1

where arg[·] represents the continuous phase.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 27 / 75


Phase Response

Principle value: −π < ARG[H(ejω )] ≤ π


continuous phase response
∢H(ejω ) = ARG[H(ejω )] + 2πr(ω)

where r(ω) is a positive or negative integer.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 28 / 75


Single Zero

H(ejω ) = 1 − rejθ e−jω


|H(ejω )|2 = 1 + r 2 − 2r cos(ω − θ)
Principle value of phase response
 
jω r sin(ω − θ)
ARG[H(e )] = arctan
1 − r cos(ω − θ)

Group delay
r2 − r cos(ω − θ)
grd[H(ejω )] =
1 + r2 − 2r cos(ω − θ)

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 29 / 75


Single Zero

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 30 / 75


Example 5.8
Consider the following second order system
1 1
H(z) = =
(1 − rejθ z −1 )(1 − re−jθ z −1 ) 1 − 2r cos θz −1 + r2 z −2

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 31 / 75


Example 5.8

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 32 / 75


Example 5.10
Consider the following second order system
0.05634(1 + z −1 )(1 − 1.0166z −1 + z −2 )
H(z) =
(1 − 0.683z −1)(1 − 1.4461z −1 + 0.7957z −2)

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 33 / 75


Example 5.10

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 34 / 75


5.4 Magnitude and Phase

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 35 / 75


Linear Difference Equations

Can we determine the frequency response from magnitude response?


Given H(ejω ), the squared magnitude response |H(ejω )|2 can be
evaluated from

|H(ejω )|2 = H(ejω )H ∗ (ejω ) = H(z)H ∗ (1/z ∗)|z=ejω

Let C(z) = H(z)H ∗ (1/z ∗ )

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 36 / 75


Example 5.11
Consider the following two systems
2(1 − z −1 )(1 + 0.5z −1)
H1 (z) =
(1 − 0.8ejπ/4 z −1 )(1 − 0.8e−jπ/4 z −1 )
(1 − z −1 )(1 + 2z −1 )
H2 (z) =
(1 − 0.8ejπ/4 z −1 )(1 − 0.8e−jπ/4 z −1 )

We can show that C1 (z) = C2 (z). That is, the magnitude responses of
both systems are the same.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 37 / 75


Example 5.12

Suppose we are given the pole-zero plot for C(z) in the figure below. We
want to determine a causal stable H(z).

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 38 / 75


Example 5.12

3 poles and 3 zeros for H(z)


Poles: (P1 , P2 , P3 ) for causality and stability
Zeros:
Real coefficient: (z3 , z1 , z2 ) or (z6 , z1 , z2 ) or (z3 , z4 , z5 ) or (z6 , z4 , z5 )
⇒ 4 choices.
Complex coefficient: additional (z3 , z1 , z5 ) or (z3 , z2 , z4 ) or (z6 , z1 , z5 )
or (z6 , z2 , z4 ) ⇒ 8 choices.
If the number of poles and zeros are not restricted, we have infinite
choice of H(z). For example, if

z −1 − a∗
H(z) = H1 (z) ,
1 − az −1
then H(z) and H1 (z) have the same C(z).

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 39 / 75


5.5 All-Pass System

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 40 / 75


All-Pass System
A first order all-pass system function is
z −1 − a∗ a∗ + z −1
Hap (z) = or Hap (z) =
1 − az −1 1 + az −1
A general all-pass system function is
a∗n + a∗n−1 z −1 + a∗n−2 z −2 + · · · + a∗0 z −n
Hap (z) =
a0 + a1 z −1 + a2 z −2 + · · · + an z −n
where a0 = 1.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 41 / 75


Example 5.13

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 42 / 75


Example 5.13

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 43 / 75


Properties of All-Pass System
Any all pass system can be factored as a cascade of first order all-pass
systems.
Mr M
Y z −1 − dk Yc z −1 − e∗k z −1 − ek
Hap (z) = A
1 − dk z −1 1 − ek z −1 1 − e∗k z −1
k=1 k=1

The group delay of causal and stable all-pass system is positive.


1 − r2
 
e−jω − re−jθ
grd = ≥0
1 − rejθ e−jω |1 − rejθ e−jω |2

The phase response of causal and stable all-pass system is negative.


Z ω
arg[Hap (ejω )] = − grd[Hap (ejφ )]dφ + arg[Hap (ej0 )] ≤ 0
0

since Hap (ej0 ) = A, arg[Hap (ej0 )] = 0.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 44 / 75


Frequency response

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 45 / 75


5.6 Minimum-Phase System

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 46 / 75


Minimum-Phase System

A minimum-phase has all its poles and zeros inside the unit circle.
Any stable and causal system can be factored as a cascade of a
minimum-phase system and an all-pass system.

H(z) = Hmin (z)Hap (z)


Put all zeros and poles of H(z) inside the unit circle into Hmin (z).

Suppose 1/c , |c| < 1 is a zero of H(z) outside the unit circle. Put
1/c∗ to Hap (z). Put c as a pole of Hap (z) and as a zero of Hmin (z).

′ z −1 − c∗
H(z) = Hmin (z)(z −1 − c∗ ) = [Hmin

(z)(1 − cz −1 )]
1 − cz −1

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 47 / 75


Example 5.14

1 + 13 z −1 z −1 + 31
  
1 + 3z −1
H1 (z) = = 3
1 + 21 z −1 1 + 12 z −1 1 + 31 z −1

(1 + 23 e+jπ/4 z −1 )(1 + 32 e−jπ/4 z −1 )


H2 (z) =
1 − 13 z −1
9 (z −1 + 32 e−jπ/4 )(z −1 + 32 ejπ/4 )
=
4 1 − 13 z −1
" #
9 (1 + 23 e−jπ/4 z −1 )(1 + 32 ejπ/4 z −1 )
=
4 1 − 13 z −1
" #
(z −1 + 32 e−jπ/4 )(z −1 + 32 ejπ/4 )
×
(1 + 23 e−jπ/4 z −1 )(1 + 23 ejπ/4 z −1 )

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 48 / 75


System Compensation

Suppose Hd (z) is a distorting system, we want to compensate it by


cascading a causal and stable system Hc (z).
Perfect compensation is possible only if Hd (z) is a minimum-phase
system.
Suppose Hd (z) = Hdmin (z)Hap (z), let
1
Hc (z) = (z),
Hdmin
then G(z) = Hap (z).
Magnitude is exactly compensated, but phase is modified.
Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 49 / 75
Example 5.15

Hd (z) = (1 − 0.9ej0.6π z −1 )(1 − 0.9e−j0.6π z −1 )


× (1 − 1.25ej0.8π z −1 )(1 − 1.25e−j0.8π z −1 )
= (1 − 0.9ej0.6π z −1 )(1 − 0.9e−j0.6π z −1 )(1.25)2
× (z −1 − 0.8ej0.8π )(z −1 − 0.8e−j0.8π )
⇒ Hmin (z) = (1.25)2 (1 − 0.9ej0.6π z −1 )(1 − 0.9e−j0.6π z −1 )
× (1 − 0.8ej0.8π z −1 )(1 − 0.8e−j0.8π z −1 )
(z −1 − 0.8ej0.8π )(z −1 − 0.8e−j0.8π )
Hap (z) =
(1 − 0.8ej0.8π z −1 )(1 − 0.8e−j0.8π z −1 )

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 50 / 75


Example 5.15

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 51 / 75


Example 5.15

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 52 / 75


Example 5.15

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 53 / 75


Example 5.15

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 54 / 75


Minimum Phase-Lag
Minimum phase-lag property

arg[H(ejω )] = arg[Hmin (ejω )] + arg[Hap (ejω )] ≤ arg[Hmin (ejω )]

Among all system functions that have the same magnitude response,
the minimum-phase system has the maximal phase response.
The phase-lag function is the negative of the phase response. Thus,
the minimum-phase system has the minimum phase-lag response. This
is the origin of the minimum-phase system.
Minimum group-delay property

grd[H(ejω )] = grd[Hmin (ejω )] + grd[Hap (ejω )] ≥ grd[Hmin (ejω )]

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 55 / 75


Minimum Energy-Delay

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 56 / 75


Minimum Energy-Delay

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 57 / 75


Minimum Energy-Delay

|h[0]| ≤ |hmin [0]|


Total energy
∞ π
1
X Z
|h[n]|2 = |H(ejω )|2 dω
n=0
2π π
π ∞
1
Z X
= |Hmin (ejω )|2 dω = |hmin [n]|2
2π π n=0

Partial energy
m
X m
X
|h[n]|2 ≤ |hmin [n]|2
n=0 n=0

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 58 / 75


5.7 Generalized Linear Phase

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 59 / 75


Pure Delay
Consider Hid (ejω ) = e−jωα , |ω| < π, we have

|Hid (ejω )| = 1, ∢Hid (ejω ) = −αω, grd[Hid (ejω )] = α

The impulse response is


sin π(n − α)
Hid [n] = , −∞ < n < ∞
π(n − α)

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 60 / 75


Non-Integer Delay

Consider H(ejω ) = |H(ejω )|e−jωα , |ω| < π


For example, a lowpass filter
(
jω e−jωα , |ω| < ωc ,
Hlp (e ) =
0, ωc < |ω| < π.

Its impulse response is


sin ωc (n − α)
Hlp [n] = , −∞ < n < ∞
ωc (n − α)

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 61 / 75


Example 5.16

sin ωc (n − α)
Hlp [n] = , −∞ < n < ∞
ωc (n − α)

If α = nd is an integer, then Hlp [2nd − n] = Hlp [n] → The impulse


response is symmetric about n = nd

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 62 / 75


Example 5.16

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 63 / 75


Generalized Linear Phase
Generalized linear-phase system

H(ejω ) = A(ejω )e−jαω+jβ

grd[H(ejω )] = α, arg[H(ejω )] = β − αω, 0 < ω < π


A necessary condition for generalized linear phase

X
h[n] sin[ω(n − α) + β] = 0
n=−∞

β = 0, π, 2α = M =an integer, h[2α − n] = h[n]


β = π/2, 3π/2, 2α = M =an integer, h[2α − n] = −h[n]

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 64 / 75


Causal G Linear Phase
If the system is causal, then

X
h[n] sin[ω(n − α) + β] = 0
n=0

Since h[2α − n] = ±h[n], we have h[n] = 0 for n < 0 and n > M .


That is
h[n] = ±h[M − n]

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 65 / 75


Type I Linear Phase System

M is even, h[n] = h[M − n]


Frequency response
 
M M/2
X X
H(ejω ) = h[n]e−jωn = e−jωM/2  a[k] cos ωk 
n=0 k=0

where

a[0] = h[M/2]
a[k] = 2h[M/2 − k], k = 1, 2, . . . , M/2.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 66 / 75


Example 5.17

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 67 / 75


Type II Linear Phase System

M is odd, h[n] = h[M − n]


Frequency response
 
M
X (M+1)/2
X 
H(ejω ) = h[n]e−jωn = e−jωM/2 b[k] cos[ω(k − 21 )]
 
n=0 k=0

where
b[k] = 2h[(M + 1)/2 − k] k = 1, 2, . . . , (M + 1)/2.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 68 / 75


Example 5.18

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 69 / 75


Type III Linear Phase System

M is even, h[n] = −h[M − n]


Frequency response
 
M M/2
X X
H(ejω ) = h[n]e−jωn = je−jωM/2  c[k] sin ωk 
n=0 k=1

where
c[k] = 2h[M/2 − k] k = 1, 2, . . . , M/2.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 70 / 75


Example 5.19

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 71 / 75


Type IV Linear Phase System

M is odd, h[n] = −h[M − n]


Frequency response
 
M
X (M+1)/2
X 
H(ejω ) = h[n]e−jωn = je−jωM/2 d[k] sin[ω(k − 21 )]
 
n=0 k=1

where
d[k] = 2h[(M + 1)/2 − k] k = 1, 2, . . . , (M + 1)/2.

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 72 / 75


Example 5.20

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 73 / 75


FIR Linear Phase System

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 74 / 75


Pole-Zero plot

Der-Feng Tseng (NTUST) DSP Chapter 5 February 19, 2015 75 / 75

You might also like