4.2.6.

Design of Equiripple Linear-Phase FIR Digital

Filters

Previous Methods:
The windowing method and the frequency-sampling
method are relatively simple techniques for designing
linear-phase FIR filter. Here, a major problem, is a lack
of precise control of the critical frequencies such cut off
frequencies of pass-band and stop-band.

1
New Method:

The new filter design method described in this section is

formulated as a Chebyshev approximation problem. It is
viewed as an optimum design criterion in the sense that
the maximum weighted approximation error between
the desired frequency response and the actual frequency
response is minimized. The resulting filter designs have
ripples in both the pass-band and the stop-band.

2
New Method:

The new filter design method described in this section is

formulated as a Chebyshev approximation problem. It is
viewed as an optimum design criterion in the sense that
the maximum weighted approximation error between
the desired frequency response and the actual frequency
response is minimized. The resulting filter designs have
ripples in both the pass-band and the stop-band.

3
New Method:

The new filter design method described in this section is

formulated as a Chebyshev approximation problem. It is
viewed as an optimum design criterion in the sense that
the maximum weighted approximation error between
the desired frequency response and the actual frequency
response is minimized. The resulting filter designs have
ripples in both the pass-band and the stop-band.

To describe the design procedure, let us recall the

following basic filter specifications

4
 Basic Filter Specifications: A Review
H  e j 

1  1
1 : pass-band ripple
1  1
pass band

transition band  2 : stop-band ripple

2
stop band
P S  5
Design of Equiripple Linear-Phase FIR Digital Filters
(Summary)

Method: formulated as a Chebyshev approximation

problem.

Optimum design criterion: minimizing the maximum

weighted approximation error between the desired
frequency response and the actual frequency response.

Solution: an iterative method based on an alternation

theorem and Remez exchange algorithm application.
Result: filter possessing ripples (equirpple) in both the
pass-band and the stop-band. 6
 Filter description:

M: the number of impulse response coefficients

M 1 M 1
y ( n)   h( k ) x ( n  k ) H (e j )   h(k )e j k
k 0 k 0

Filter specifications:

Pass-band: 1  1  H r ( )  1  1

Stop-band:  2  H r ( )   2

7
 Four different cases of linear phase FIR filters:

Case 1: symmetric impulse response [h(k)=h(M-1-k)],

M odd.
Case 2: symmetric impulse response [h(k)=h(M-1-k)],
M even.
Case 3: antisymmetric impulse response
[h(k)=-h(M-1-k)], M odd.
Case 4: antisymmetric impulse response
[h(k)=-h(M-1-k)], M even.

8
 Frequency response of the linear phase FIR filter

We have shown, that the frequency response of linear

phase FIR digital filters can be expressed as follows:
M 1
 j
j
H (e )  H r ( ) e 2

where H r ( ) is the real-valued frequency response.

9
 L
H r ( )  Q ( ) P ( ) where P( )   (k )cos  k
k 0

The  (k ) represent the parameters of the filter which are

linearly related to the unit sample response h(k).

Case 1: Q( )  1 L=(M-1)/2


Case 2: Q( )  cos L=(M/2)-1
2

Case 3: Q( )  sin L=(M-3)/2
2

Case 4: Q( )  sin L=(M/2)-1
2
10
A. The actual real-valued frequency response of the
filter:
H r ( )

B. The desired (ideal) real-valued frequency response

of the filter:
1 for  in pass  band
H dr ( )  
0 for  in stop  band

C. The weighting function of the approximation

error:
 2 /  1 for  in pass  band
W ( )  
 1 for  in stop  band
11
 Comments on weighting function:

The weighting function on the approximation error allows

to choose the relative size of the approximation errors in the
different frequency bands (i.e. in the pass-band and in the
stop-band) differently. In particular, it is convenient to
normalize
W ( )
to the unity in the stop-band
W ( )  1
and set
W ( )   2 /  1
in the pass-band.
12
With regard to previous specifications of the desired
(ideal) and actual real-valued frequency responses and the
weighting function on the approximation error, we can
now define the weighted approximation error as
E ( ) 

 W ( )  H dr ( )  H r ( ) 

 W ( )  H dr ( )  Q( ) P( )  

 H dr ( ) 
 W ( )Q( )   P( ) 
 Q( )  13
The modified real-valued desired frequency response:
H dr ( )
H dr ( ) 
Q( )

The modified weighting function

on the approximation error:
W ( )  W ( )Q( )

The weighted approximation error:

E ( )  W ( )  H dr ( )  P ( )  
 
 L

 W ( )  H dr ( )    ( k )cos  k 
 k 0 
14
for all different types of linear-phase FIR filters.
Given the error function E ( ), the Chebyshev’s
approximation problem is basically to determine the
filter parameters  (k ) that minimize the maximum
absolute value E ( ) over the frequency bands in which
the approximation is to be performed (set: S). In
mathematical terms we seek the solution of the problem:

min  max E ( )  
over  ( k )   S 

 min  max W ( )  H dr ( )  P( )   

over  ( k ) 
 S   

  L

 min  max W ( )  H dr ( )   (k )cos  k  
over  ( k )

S
 k 0 
15
 Note A.:
Here, S represents the set (disjoint union) of frequency
bands over which the optimization is to be performed.
Basically, the set S consists of the pass-bands and stop-
bands of the desired frequency response of the filter.

min  max E ( )  
over  ( k )   S 

 min  max W ( )  H dr ( )  P( )   

over  ( k ) 
 S   

  L

 min  max W ( )  H dr ( )   ( k ) cos  k  
over  ( k )

 S
 k 0 
16
 Note B.:
The described criterion refered to as Chebyshev’s mini-
max criterion leads to an equiripple filter i.e. to a filter
whose magnitude response oscillates uniformly
between the tolerance bounds of each band.

min  max E ( )  
over  ( k )   S 

 min  max W ( )  H dr ( )  P( )   

over  ( k ) 
 S   

  L

 min  max W ( )  H dr ( )   ( k ) cos  k  
over  ( k )

 S
 k 0 
17