S-88 3104 13.5

1
Copyright 2010, S. K. Mitra

1
Nyquist Nyquist Filters Filters
Under certain conditions, a lowpassfilter
can be designed to have a number of zero-
valued coefficients
When used as interpolation filters these
filters preserve the nonzero samples of the
up-sampler output at the interpolator output
Moreover, due to the presence of these
zero-valued coefficients, these filters are
computationally more efficient than other
lowpassfilters of same order
2
L Lth th- -Band Filters Band Filters
These filters, called theNyquist filtersor
Lth-band filters, are often used in single-rate
and multi-rate signal processing
Consider the factor-of-L interpolator shown
below
The input-output relation of the interpolator
in the z-domain is given by
L ] [n x
] [n y
) (z H
] [n x
u
) ( ) ( ) (
L
z X z H z Y =
3
If H(z) is realized in theL-band polyphase
form, then we have
Assume that thek-thpolyphasecomponent
of H(z) is a constant, i.e., :

=

=
1
0
) ( ) (
L
i
L
i
i
z E z z H
= ) (z E
k
) ( . . . ) ( ) ( ) (
1
) 1 (
1
1
0
L
k
k L L
z E z z E z z E z H

+ + + =
) ( . . . ) (
1
) 1 (
1
) 1 ( L
L
L L
k
k k
z E z z E z z

+
+
+ + + +
4
Then we can expressY(z) as
As a result,
Thus, the input samples appear at the output
without any distortion for all values of n,
whereas, in-between output samples
are determined by interpolation
=

+ =
1
0
) ( ) ( ) ( ) (
L
L L L k
z X z E z z X z z Y
l
l
l
k l
] [ ] [ n x k Ln y = +
) 1 ( L
5
A filter with the above property is called a
Nyquist filter or anLth-band filter
Its impulse response has many zero-valued
samples, making it computationally
attractive
For example, the impulse response of an
Lth-bandfilter for k =0 satisfies the
following condition
= ] [Ln h
otherwise , 0
0 , = n
6
Figure below shows a typical impulse
response of athird-band filter (L =3)
Lth-band filterscan be either FIR or IIR
filters
n
0 3 6
_
3
_
6
h[n]
2
7
If the0-th polyphasecomponent of H(z) is a
constant, i.e., then it can be shown
that
(assuming =1/L)
Since the frequency response of is
the shifted version of ,
the sum of all of theseL uniformly shifted
versions of add up to a constant
= ) (
0
z E

=
= =
1
0
1 ) (
L
k
k
L
L zW H
) (
k
L
zW H
) (
) / 2 ( L k j
e H

) (
j
e H
) (
j
e H
0 2
H(zWL) H(zWL
2) H(zWL
L 1)
_
H(z) H(z)
8
Half Half- -Band Filters Band Filters
AnLth-bandfilter for L =2 is called ahalf-
band filter
The transfer function of a half-band filter is
thus given by
with its impulse response satisfying
) ( ) (
2
1
1
z E z z H

+ =
= ] 2 [ n h
otherwise , 0
0 , = n
9
The condition
reduces to
(assuming =0.5)
If H(z) has real coefficients, then
Hence
) ( ) (
2
1
1
z E z z H

+ =
1 ) ( ) ( = + z H z H
) ( ) (
) (
=
j j
e H e H
1 ) ( ) (
) (
= +
j j
e H e H
10
and add up
to1 for all
Or, in other words, exhibits a
symmetry with respect to the half-band
frequency/2, hence the name half-band
filter
) (
) 2 / ( j
e H ) (
) 2 / ( + j
e H
) (
j
e H
11
Figure below illustrates this symmetry for a
half-band lowpassfilter for which passband
and stopbandripples are equal, i.e.,
and passbandand stopbandedges are
symmetric with respect to/2, i.e.,
s p
=
= +
s p
1
1
+
_
p
s
/2

H(e )
j
12
Attractive property: About 50% of the
coefficients of h[n] are zero
This reduces the number of multiplications
required in its implementation significantly
For example, if N =101, an arbitrary Type 1
FIR transfer function requires about 50
multipliers, whereas, aType 1half-band
filter requires only about 25 multipliers
3
13
An FIR half-band filter can be designed
with linear phase
However, there is a constraint on its length
Consider a zero-phase half-band FIR filter
for which , with
Let the highest nonzero coefficient beh[R]
] [ * ] [ n h n h = 1 | | =
14
ThenR is odd as a result of the condition
ThereforeR =2K+1 for some integer K
Thus the length of h[n] is restricted to be of
the form2R+1 =4K+3 [unlessH(z) is a
constant]
= ] 2 [ n h
otherwise , 0
0 , = n
15
Design of Linear Design of Linear- -Phase Phase
A lowpasslinear-phaseLth-band FIR filter
can be readily designed via the windowed
Fourier series approach
In this approach, the impulse response
coefficients of the lowpassfilter are chosen
as where is the
impulse response of an ideal lowpassfilter
with a cutoff at /L and w[n] is a suitable
window function
] [ ] [ ] [ n w n h n h
LP
= ] [n h
LP
16
Now, the impulse response of an ideal Lth-
band lowpassfilter with a cutoff at
is given by
It can be seen from the above that

= n
n
L n
n h
LP
,
) / sin(
] [
L
c
/ =
. . . , 2 , for 0 ] [ L L n n h
LP
= =
17
Hence, the coefficient condition of theLth-
band filter
is indeed satisfied
Hence, an Lth-band FIR filter can be
designed by applying a suitable window
w[n] to
= ] [Ln h
otherwise , 0
0 , = n
] [n h
LP
18
There are many other candidates for Lth-
band FIR filters
Program 13_8can be used to design an Lth-
band FIR filter using the windowed Fourier
series approach
The program employs the Hamming
window
However, other windows can also be used
4
19
Figure below shows the gain response of a
half-band filter of length-23 designed using
Program 13_8
0 0.2 0.4 0.6 0.8 1
-80
-60
-40
-20
0
/
G
a
i
n
,

d
B
20
The filter coefficients are given by
As expected, h[n] =0for
; 0 ] 10 [ ] 10 [ ; 002315 . 0 ] 11 [ ] 11 [ = = = = h h h h
; 0 ] 8 [ ] 8 [ ; 005412 . 0 ] 9 [ ] 9 [ = = = = h h h h
; 0 ] 6 [ ] 6 [ ; 001586 . 0 ] 7 [ ] 7 [ = = = = h h h h
; 0 ] 4 [ ] 4 [ ; 003584 . 0 ] 5 [ ] 5 [ = = = = h h h h
; 0 ] 2 [ ] 2 [ ; 089258 . 0 ] 3 [ ] 3 [ = = = = h h h h
; 5 . 0 ] 0 [ ; 3122379 . 0 ] 1 [ ] 1 [ = = = h h h
10 , 8 , 6 , 4 , 2 = n
21
We show below the gain response of a
length 234
th
-band lowpassfilter designed
using the M-file firnyquist firnyquist with 2
different roll-off factors
0 0.2 0.4 0.6 0.8 1
-80
-60
-40
-20
0
R = 0.4 R = 0.1
/
G
a
in
, d
B
Gain response of 4th-band FIR filter
22
The problem of designing a real-coefficient
half-band FIR filter can be transformed into
the design of a single passbandFIR filter
with no stopbandwhich can be easily
designed using the Parks-McClellan
algorithm
An inverse transformation of the wideband
filter then yields the half-band FIR filter
23
Let the specifications of the real-coefficient
half-band filter G(z) of order N be as
follows:
Passbandedge at , stopbandedge at ,
passbandripple of , and stopbandripple
of
Now for a half-band filter ,
and the order N is even with
N/2odd
= =
s p
= +
s p
p

24
Now, consider the design of a wide-band
linear-phase FIR filter F(z) of degree N/2
with a passbandfrom 0to , a transition
band from to , and a passbandripple
of 2
Since N/2is odd, F(z) has a zero at
Let f[n] denote the impulse response of F(z)
1 = z
p
2
p
2
5
25
Define
G G( (z z) ) can be seen to be the transfer function
of a causal half-band lowpassFIR filter
with an impulse response
)] ( [ ) (
/ 2 2
2
1
z F z z G
N
+ =

=
=
2 2
1
2
2
1
0
2
N
N
n
n odd n
even n n f
n g
,
, ,
], / [
] [
26
The plots below the magnitude response of an
wide-band lowpassfilter of degree 13with a
passbandfrom0to 0.85 and a transition band
from0.9 to and the magnitude response of the
derived half-band filter
0 0.2 0.4 0.6 0.8 1
0
0.2
0.4
0.6
0.8
1
/
M
a
g
n
itu
d
e
Half-Band Lowpass Filter
0 0.2 0.4 0.6 0.8 1
0
0.2
0.4
0.6
0.8
1
Wide-Band Lowpass Filter
/
M
a
g
n
itu
d
e
27
Design of Half Design of Half- -Band IIR Filters Band IIR Filters
Recall that an odd-order bounded real (BR)
lowpassIIR transfer function G(z) =
P(z)/D(z) with a symmetric numerator and
satisfying the power-symmetry condition
can be decomposed in the form
where and are stable allpass
functions
1
1 1
= +

) ( ) ( ) ( ) ( z G z G z G z G
)] ( ) ( [ ) (
2
1
1 2
0
2
1
z z z z G A A

+ =
) (z
0
A ) (z
1
A
28
It follows from the power-symmetry
condition that G(z) is a half-band lowpass
transfer function
A Butterworth half-band lowpassIIR filter
G(z) can be designed by first designing an
odd-order analog Butterworth lowpassfilter
with a 3-dB cutoff frequency at and
then applying a bilinear transformation
We next consider the design of an elliptic
IIR half-band filter
1 =
c
29
It can be shown that any odd-order elliptic
lowpasshalf-band filter G(z) with a
frequency response specification given by
and satisfying the conditions
is a power-symmetric transfer function
p
j
p
e G

0 1 2 1 for , ) (

s s
j
e G for , ) (
) ( ,
p p s s p
= = + 1 4
2
30
It can be shown that the poles of the elliptic
lowpasshalf-band filter lie on the imaginary
axis
Using the pole-interlacing property, we can
readily identify the expressions for
and
) (z
0
A
) (z
1
A
6
31
Design Steps:
Since , only
one of the bandedgesand one of the ripples
can be specified
Let the specified stopbandedgeand
stopbandripplebe and , respectively
Then and are determined using the
equations at the top of the slide
) ( ,
p p s s p
= = + 1 4
2
s

32
Define
and compute
) / tan(
) / tan(
2
2
s
p
r
=
2
1 r r = '
) ' (
) ' (
r
r
q
+
=
1 2
1
0
13
0
9
0
5
0 0
150 15 2 q q q q q + + + =
2
2
2
1

=
s
s
D
33
Next, the estimate of the order of G(z) is
determined by choosing the smallest odd
integer satisfying
Now the integer value of N is almost always
higher than the quantity on the RHS of the
above equation
) / ( log
) ( log
q
D
N
1
16
10
10

34
As a result, the corresponding value of
will be smaller than the original specified
value
To determine the actual value of , the
actual value of the parameter D is first
computed from
16
10
1
10
) / ( log q N
D =
s

35
From the actual value of D, the actual value
of is computed by solving
From the new value of , the actual value
of is obtained from
s
2
2
2
1

=
s
s
D
) (
p p s
= 1 4
2
s

36
Next, the poles of the two allpassfilters are
computed as follows:
+
+
=
=
+
0
0
1 4 1
2 1 2 1
1 2 1 2
2
i
i i
i
i i i
k
N ki q
N k i q q
) / cos( ) (
) / ) sin(( ) (
) ( /
=

r
k k
k
r b
2
1 1
2
) (
) /(
2
1 2
k k k
b c + =
) /( ) (
k k k
c c + =

2 2
1
7
37
In general, the two infinite sums in the
expression for converge after the
addition of 5or 6terms
The poles of the two allpassfilters are on
the imaginary axis at and are
inside the unit circle, as the parameters
are distinct with magnitudes less than1
k
j z =
k

38
Using the pole-interlacing property, then
poles of and are selected
Their corresponding zeros are at the mirror-
image locations
Example
We consider the design of an elliptic half-
band lowpassfilter
) (z
0
A ) (z
1
A
39
The specifications are:
The transfer functions of the two allpass
sections of the half-band filter designed
using Program 13_9are given by
= 6 0.
s
016 0. =
s
1
1
1
1
1
0
71454 0 1
71454 0
23647 0 1
23647 0
+
+
=
+
+
=
z
z
z
z
z
z
.
.
) ( ,
.
.
) ( A A
40
The pole-zero plot and the magnitude
response of the designed elliptic lowpass
half-band filter are shown below:
-1.5 -1 -0.5 0 0.5 1 1.5
-1
-0.5
0
0.5
1
Real Part
Im
a
g
in
a
ry
P
a
rt
The pole-zero plot
0 0.2 0.4 0.6 0.8 1
0
0.2
0.4
0.6
0.8
1
/
M
a
g
n
itu
d
e
Half-Band IIR Lowpass Filter

S-88 3104 13.5

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

S-88 3104 13.5

Uploaded by

Copyright:

Available Formats

1

Copyright 2010, S. K. Mitra

Copyright 2010, S. K. Mitra

Copyright 2010, S. K. Mitra

Copyright 2010, S. K. Mitra

Copyright 2010, S. K. Mitra

Copyright 2010, S. K. Mitra

Copyright 2010, S. K. Mitra

You might also like