Optimal FIR Filters

7: Optimal FIR filters
Optimal Filters
Alternation Theorem
Chebyshev Polynomials
Maximal Error Locations
Remez Exchange
Algorithm
Determine Polynomial
Example Design
FIR Pros and Cons
Summary
MATLAB routines
DSP and Digital Filters (2014-5490)
Optimal FIR: 7 1 / 11
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange
We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

around h[0], i.e. h[n] = h[n].
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange

j
H() = H(e ) =
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
M
2
h[n]ejn
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange

j
H() = H(e ) =
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
M
2
jn
h[n]e
= h[0] + 2
P M2
1
h[n] cos n
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

j
P M2
jn
P M2
= h[0] + 2 1 h[n] cos n

j

H() is real but not necessarily positive (unlike H(e ) ).
H() = H(e ) =
M
2
h[n]e
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

j
P M2
jn
P M2
= h[0] + 2 1 h[n] cos n

j


Weighted error: e() = s() H() d() where d() is the target.
H() = H(e ) =
M
2
h[n]e
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

j
P M2
jn
P M2
= h[0] + 2 1 h[n] cos n

j


Choose s() to control the error variation with .
H() = H(e ) =
M
2
h[n]e
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

j
P M2
jn
P M2
= h[0] + 2 1 h[n] cos n

j


H() = H(e ) =
M
2
h[n]e
Example: lowpass filter
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

j
P M2
jn
P M2
= h[0] + 2 1 h[n] cos n

j


H() = H(e ) =
M
2
h[n]e
d() =
1 0 1
0 2
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

j
P M2
jn
P M2
= h[0] + 2 1 h[n] cos n

j


H() = H(e ) =
M
2
h[n]e
d() =
1 0 1
0 2
(
1 0 1
s() = 1
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

j
P M2
jn
P M2
= h[0] + 2 1 h[n] cos n

j


H() = H(e ) =
M
2
h[n]e
d() =
1 0 1
0 2
(
1 0 1
s() = 1
2
e() = 1 when H() lies at the edge of the specification.
Optimal Filters
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

j
P M2
jn
P M2
= h[0] + 2 1 h[n] cos n

j


H() = H(e ) =
M
2
h[n]e
d() =
1 0 1
0 2
(
1 0 1
s() = 1
2
e() = 1 when H() lies at the edge of the specification.
Minimax criterion: h[n] = arg minh[n] max |e()|: minimize max error
Alternation Theorem
Optimal Filters
Alternation Theorem
Remez Exchange
Want to find the best fit line: with the smallest maximal error.
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
8
6
4
2
2
Alternation Theorem
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Best fit line always attains the
maximal error three times with
alternate signs
Example Design
FIR Pros and Cons
Summary
MATLAB routines
8
6
4
2
2
Alternation Theorem
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
alternate signs
8
6
4
2
2
Proof:
Assume the first maximal deviation from the line is negative as shown.
Alternation Theorem
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
alternate signs
8
6
4
2
2
Proof:
There must be an equally large positive deviation; or else just move the line
downwards to reduce the maximal deviation.
Alternation Theorem
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
alternate signs
8
6
4
2
2
Proof:
This must be followed by another maximal negative deviation; or else you
can rotate the line and reduce the deviations.
Alternation Theorem
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
alternate signs
8
6
4
2
2
Proof:
Alternation Theorem:
A polynomial fit of degree n to a bounded set of points is minimax if and
only if it attains its maximal error at n + 2 points with alternating signs.
Alternation Theorem
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
alternate signs
8
6
4
2
2
Proof:
There may be additional maximal error points.
Alternation Theorem
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
alternate signs
8
6
4
2
2
Proof:
There may be additional maximal error points.
Fitting to a continuous function is the same as to an infinite number of
points.
Optimal Filters
Alternation Theorem
Remez Exchange
H() = H(e ) = h[0] + 2
P M2
1
h[n] cos n
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
H() = H(e ) = h[0] + 2

But
P M2
1
h[n] cos n
cos 2 = 2 cos2 1
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
H() = H(e ) = h[0] + 2

But
P M2
1
h[n] cos n
cos 2 = 2 cos2 1
cos 3 = 4 cos3 3 cos
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
H() = H(e ) = h[0] + 2
P M2
1
h[n] cos n
But cos n = Tn (cos ): Chebyshev polynomial of 1st kind
cos 2 = 2 cos2 1 = T2 (cos )

cos 3 = 4 cos3 3 cos = T3 (cos )
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
H() = H(e ) = h[0] + 2
P M2
1
h[n] cos n
cos 2 = 2 cos2 1 = T2 (cos )

Recurrence Relation:
Tn+1 (x) = 2xTn (x) Tn1 (x) with T0 (x) = 1, T1 (x) = x
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
H() = H(e ) = h[0] + 2
P M2
1
h[n] cos n
cos 2 = 2 cos2 1 = T2 (cos )


Proof: cos (n + ) + cos (n ) = 2 cos cos n
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
H() = H(e ) = h[0] + 2
P M2
1
h[n] cos n
cos 2 = 2 cos2 1 = T2 (cos )


So H() is an M
2 order polynomial in cos : alternation theorem applies.
Optimal Filters
Alternation Theorem
Remez Exchange
H() = H(e ) = h[0] + 2
h[n] cos n
cos 2 = 2 cos2 1 = T2 (cos )

Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2

So H() is an M
Example: Low-pass filter of length 5 (M = 4)
1.4
M=4
1.2
1
|g|
0.8
0.6
0.4
0.2
0
0
0.5
1.5
2.5
3.5
Optimal Filters
Alternation Theorem
Remez Exchange
H() = H(e ) = h[0] + 2
h[n] cos n
cos 2 = 2 cos2 1 = T2 (cos )

Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2

So H() is an M
1.4
M=4
M=4
1.2
|g| (dB)
1
|g|
0.8
0.6
-5
-10
0.4
0.2
0
0
-15
0.5
1.5
2.5
3.5
0.5
1.5
2.5
Optimal Filters
Alternation Theorem
Remez Exchange
H() = H(e ) = h[0] + 2
h[n] cos n
cos 2 = 2 cos2 1 = T2 (cos )

Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2

So H() is an M
1.4
M=4
M=4
1.2
M=2
0
1
|g|
0.8
0.6
-5
g
|g| (dB)
0.5
-10
0.4
0
0.2
0
0
-15
0.5
1.5
2.5
3.5
0.5
1.5
2.5
-1
-0.5
0
cos()
0.5
Optimal Filters
Alternation Theorem
Remez Exchange
Maximal error locations occur either at band

edges or when dH
d = 0
1.2
M=18
1
0.8
|H|
0.6
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
0.4
0.2
0
0
0.5
1.5
2.5
Optimal Filters
Alternation Theorem
Remez Exchange

edges or when dH
d = 0
1.2
M=18
1
0.8
|H|
0.6
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
0.4
0.2
0
0
0.5
1.5
2.5
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm

edges or when dH
d = 0
H() = h[0] + 2
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
1
1.2
M=18
1
0.8
h[n] cos n
|H|
0.6
0.4
0.2
0
0
0.5
1.5
2.5
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

edges or when dH
d = 0
P M2
H() = h[0] + 2
= P (cos )
M=18
1
0.8
h[n] cos n
where P (x) is a polynomial of order M

2 .
1.2
|H|
0.6
0.4
0.2
0
0
0.5
1.5
2.5
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

edges or when dH
d = 0
P M2
H() = h[0] + 2
= P (cos )
M=18
1
0.8
h[n] cos n

2 .
dH
d
1.2
|H|
0.6
0.4
0.2
0
0
0.5
1.5
2.5
= P (cos ) sin
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

edges or when dH
d = 0
P M2
H() = h[0] + 2
= P (cos )
h[n] cos n
= P (cos ) sin
= 0 at = 0, and at most
M=18
1
0.8
0.6
0.4
0.2

2 .
dH
d
1.2
|H|
M
2
0
0
0.5
1.5
2.5
1 zeros of polynomial P (x).
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

edges or when dH
d = 0
P M2
H() = h[0] + 2
= P (cos )
h[n] cos n
= P (cos ) sin
0.6
0.4
0.2
M
2
With two bands, we have at most
M=18
0.8

2 .
dH
d
1.2
|H|
0
0
0.5
1.5
2.5
M
2
+ 3 maximal error frequencies.
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

edges or when dH
d = 0
P M2
H() = h[0] + 2
= P (cos )
M=18
1
0.8
h[n] cos n
= P (cos ) sin
0.6
0.4
0.2

2 .
dH
d
1.2
|H|
M
2
0
0
0.5
1.5
2.5
With two bands, we have at most M

2 + 3 maximal error frequencies.
We require M
2 + 2 of alternating signs for the optimal fit.
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

edges or when dH
d = 0
P M2
H() = h[0] + 2
= P (cos )
M=18
1
0.8
h[n] cos n
= P (cos ) sin
0.6
0.4
0.2

2 .
dH
d
1.2
|H|
M
2
0
0
0.5
1.5
2.5

We require M
Only three possibilities exist (try them all):
(a) = 0 + two band edges + M

2 1 zeros of P (x).
1
zeros
of
P
(x).
(b) = + two band edges + M
2
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

edges or when dH
d = 0
P M2
H() = h[0] + 2
= P (cos )
M=18
1
0.8
h[n] cos n
= P (cos ) sin
0.6
0.4
0.2

2 .
dH
d
1.2
|H|
M
2
0
0
0.5
1.5
2.5

We require M
Only three possibilities exist (try them all):
(a) = 0 + two band edges + M

2 1 zeros of P (x).
1
zeros
of
P
(x).
(b) = + two band edges + M
2
(c) = {0 and } + two band edges + M

2 2 zeros of P (x).
Remez Exchange Algorithm

Optimal Filters
Alternation Theorem
Remez Exchange
1. Guess the positions of the M

2 + 2 maximal error frequencies and give
alternating signs to the errors (e.g. choose evenly spaced ).
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

2. Determine the error magnitude, , and the M
2 + 1 coefficients of
the polynomial that passes through the maximal error locations.
M=4
Iteration 1
1.5
|g|
1
0.5
0
-0.5
0
0.5
1.5
2.5

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

3. Find the local maxima of the error
function by evaluating
e() = s() H() d() on a dense set of .
M=4
Iteration 1
1.5
|g|
1
0.5
0
-0.5
0
0.5
1.5
2.5

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

4. Update the maximal error frequencies to be an alternating subset of

the local maxima + band edges + {0 and/or }.
M=4
Iteration 1
1.5
|g|
1
0.5
0
-0.5
0
0.5
1.5
2.5

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


If maximum error is > , go back to step 2.
M=4
Iteration 1
1.5
|g|
1
0.5
0
-0.5
0
0.5
1.5
2.5

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


If maximum error is > , go back to step 2.
M=4
Iteration 1
1.5
M=4
Iteration 2
1
|g|
|g|
1
0.5
0
0.5
-0.5
0
0.5
1.5
2.5
0.5
1.5
2.5

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


If maximum error is > , go back to step 2. (typically 15 iterations)
M=4
Iteration 1
1.5
M=4
Iteration 2
1
|g|
|g|
1
0.5
0
0.5
-0.5
0
0.5
1.5
2.5
0.5
1.5
2.5

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


M=4
Iteration 1
1.5
M=4
Iteration 2
M=4
Iteration 3
0.5
0
|g|
|g|
|g|
1
0.5
0.5
-0.5
0
0.5
1.5
2.5
0.5
1.5
2.5
0.5
1.5
2.5

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


M=4
Iteration 1
1.5
M=4
Iteration 2
M=4
Iteration 3
0.5
0
|g|
|g|
|g|
1
0.5
0.5
-0.5
0
0.5
1.5
2.5
0.5
1.5
2.5
0.5
1.5
2.5

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


5. Evaluate H() on M + 1 evenly spaced and do an IDFT to get h[n].
M=4
Iteration 1
1.5
M=4
Iteration 2
M=4
Iteration 3
0.5
0
|g|
|g|
|g|
1
0.5
0.5
-0.5
0
0.5
1.5
2.5
0.5
1.5
2.5
0.5
1.5
2.5
Optimal Filters
Alternation Theorem
Remez Exchange
For each extremal frequency, i for 1 i M

2 +2
d(i ) = H(i ) +
(1)i
s(i )
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
(1)i
s(i )
= h[0] + 2
P M2
n=1
h[n] cos ni +
(1)i
s(i )
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
(1)i
s(i )
= h[0] + 2
P M2
n=1
h[n] cos ni +
(1)i
s(i )
Method 1:
M
Solve M
2 + 2 equations in 2 + 2 unknowns for h[n] + .
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
(1)i
s(i )
= h[0] + 2
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
n=1
h[n] cos ni +
(1)i
s(i )
Method 1:
M
Solve M
In step 3, evaluate H() = h[0] + 2
P M2
n=1
h[n] cos ni
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
(1)i
s(i )
= h[0] + 2
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
n=1
h[n] cos ni +
(1)i
s(i )
Method 1: (Computation time M 3 )

M
Solve M
P M2
n=1
h[n] cos ni
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
(1)i
s(i )
= h[0] + 2
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
n=1
h[n] cos ni +
(1)i
s(i )

M
Solve M
P M2
n=1
h[n] cos ni
Method 2: Dont calculate h[n] explicitly

Q
1
Multiply equations by ci = j6=i cos cos
and add:
P M2 +2
i=1
i

P M2
ci h[0] + 2 n=1 h[n] cos n +
(1)i
s(i )
P M2 +2
i=1
ci d(i )
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
(1)i
s(i )
= h[0] + 2
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
n=1
h[n] cos ni +
(1)i
s(i )

M
Solve M
P M2
n=1
h[n] cos ni

Q
1
and add:
i
j

PM
P M2
(1)i
2 +2
h[n]
cos
n
+
c
h[0]
+
2
=
i
n=1
i=1
i=1 ci d(i )
s(i )
All terms involving h[n] sum to zero leaving
P M2 +2 (1)i ci
P M2 +2
i=1
i=1 ci d(i )
s(i ) =
P M2 +2
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
(1)i
s(i )
= h[0] + 2
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
n=1
h[n] cos ni +
(1)i
s(i )

M
Solve M
P M2
n=1
h[n] cos ni

Q
1
and add:
i
j

PM
P M2
(1)i
2 +2
h[n]
cos
n
+
c
h[0]
+
2
=
i
n=1
i=1
i=1 ci d(i )
s(i )
P M2 +2 (1)i ci
P M2 +2
i=1
i=1 ci d(i )
s(i ) =
P M2 +2
Solve for
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
(1)i
s(i )
= h[0] + 2
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
n=1
h[n] cos ni +
(1)i
s(i )

M
Solve M
P M2
n=1
h[n] cos ni

Q
1
and add:
i
j

PM
P M2
(1)i
2 +2
h[n]
cos
n
+
c
h[0]
+
2
=
i
n=1
i=1
i=1 ci d(i )
s(i )
P M2 +2 (1)i ci
P M2 +2
i=1
i=1 ci d(i )
s(i ) =
P M2 +2
Solve for then calculate the H(i )
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
(1)i
s(i )
= h[0] + 2
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
n=1
h[n] cos ni +
(1)i
s(i )

M
Solve M
P M2
n=1
h[n] cos ni

Q
1
and add:
i
j

PM
P M2
(1)i
2 +2
h[n]
cos
n
+
c
h[0]
+
2
=
i
n=1
i=1
i=1 ci d(i )
s(i )
P M2 +2 (1)i ci
P M2 +2
i=1
i=1 ci d(i )
s(i ) =
P M2 +2
Solve for then calculate the H(i ) then use Lagrange interpolation:
H() = P (cos ) =
P M2 +2
i=1
H(i )
cos cos j
j6=i cos i cos j
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
(1)i
s(i )
= h[0] + 2
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
n=1
h[n] cos ni +
(1)i
s(i )

M
Solve M
P M2
n=1
h[n] cos ni

Q
1
and add:
i
j

PM
P M2
(1)i
2 +2
h[n]
cos
n
+
c
h[0]
+
2
=
i
n=1
i=1
i=1 ci d(i )
s(i )
P M2 +2 (1)i ci
P M2 +2
i=1
i=1 ci d(i )
s(i ) =
P M2 +2
P M2 +2
cos cos
H() = P (cos ) = i=1 H(i ) j6=i cos i cos jj

M
2 + 1 -polynomial going through all the H(i ) [actually order
M
2 ]
Optimal Filters
Alternation Theorem
Remez Exchange

2 +2
d(i ) = H(i ) +
(1)i
s(i )
= h[0] + 2
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
P M2
n=1
h[n] cos ni +
(1)i
s(i )

M
Solve M
P M2
n=1
h[n] cos ni
Method 2: Dont calculate h[n] explicitly (Computation time M 2 )

Q
1
and add:
i
j

PM
P M2
(1)i
2 +2
h[n]
cos
n
+
c
h[0]
+
2
=
i
n=1
i=1
i=1 ci d(i )
s(i )
P M2 +2 (1)i ci
P M2 +2
i=1
i=1 ci d(i )
s(i ) =
P M2 +2
P M2 +2
cos cos
H() = P (cos ) = i=1 H(i ) j6=i cos i cos jj

M
2 + 1 -polynomial going through all the H(i ) [actually order
M
2 ]
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Filter Specifications:
Bandpass = [0.5, 1],
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
M=36
|H|
0.8
0.6
0.4
0.2
0
0
0.5
1.5
2.5
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Bandpass = [0.5, 1], Transition widths: = 0.2
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
M=36
|H|
0.8
0.6
0.4
0.2
0
0
0.5
1.5
2.5
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Stopband Attenuation: 25 dB and 15 dB
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
M=36
|H|
0.8
0.6
0.4
0.2
0
0
0.5
1.5
2.5
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Passband Ripple: 0.3 dB
Example Design
FIR Pros and Cons
Summary
MATLAB routines
M=36
|H|
0.8
0.6
0.4
0.2
0
0
0.5
1.5
2.5
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
Determine gain tolerances for each band:
25 dB = 0.056, 0.3 dB = 1 0.034, 15 dB = 0.178
M=36
|H|
0.8
0.6
0.4
0.2
0
0
0.5
1.5
2.5
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
25 dB = 0.056, 0.3 dB = 1 0.034, 15 dB = 0.178
Predicted order: M = 36
M=36
|H|
0.8
0.6
0.4
0.2
0
0
0.5
1.5
2.5
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
25 dB = 0.056, 0.3 dB = 1 0.034, 15 dB = 0.178
M
2 + 2 extremal frequencies are distributed between the bands
M=36
|H|
0.8
0.6
0.4
0.2
0
0
0.5
1.5
2.5
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
25 dB = 0.056, 0.3 dB = 1 0.034, 15 dB = 0.178
M
Filter meets specs ,
M=36
|H|
0.8
0.6
0.4
0.2
0
0
0.5
1.5
2.5
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
25 dB = 0.056, 0.3 dB = 1 0.034, 15 dB = 0.178
M
Filter meets specs ,; clearer on a decibel scale
M=36
|H| (dB)
|H|
0.8
0.6
0.4
-10
-15
-20
-25
0.2
0
0
M=36
-5
0.5
1.5
2.5
-30
0
0.5
1.5
2.5
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
25 dB = 0.056, 0.3 dB = 1 0.034, 15 dB = 0.178
M
Most zeros are on the unit circle + three reciprocal pairs
1
M=36
|H| (dB)
|H|
0.8
0.6
0.4
1
0.5
-10
-15
-20
-0.5
-25
0.2
0
0
M=36
-5
0.5
1.5
2.5
-30
0
-1
0.5
1.5
2.5
-1
Example Design
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
25 dB = 0.056, 0.3 dB = 1 0.034, 15 dB = 0.178
M
Most zeros are on the unit circle + three reciprocal pairs
Reciprocal pairs give a linear phase shift
1
M=36
|H| (dB)
|H|
0.8
0.6
0.4
1
0.5
-10
-15
-20
-0.5
-25
0.2
0
0
M=36
-5
0.5
1.5
2.5
-30
0
-1
0.5
1.5
2.5
-1
FIR Pros and Cons

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
Can have linear phase

no envelope distortion, all frequencies have the same delay ,
symmetric or antisymmetric: h[n] = h[n]n or h[n]n
antisymmetric filters have H(ej0 ) = H(ej ) = 0
symmetry means you only need M
2 + 1 multiplications
to implement the filter.
FIR Pros and Cons

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

Always stable ,
FIR Pros and Cons

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

Always stable ,
Low coefficient sensitivity ,
FIR Pros and Cons

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

Always stable ,
Optimal design method fast and robust ,
FIR Pros and Cons

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

Always stable ,
Normally needs higher order than an IIR filter /
dBatten
Filter order M 3.5
where is the most rapid transition
FIR Pros and Cons

Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

Always stable ,
Normally needs higher order than an IIR filter /
dBatten
Filter order M 3.5
where is the most rapid transition
dBatten
dBatten
fs = 3.5
fs2
Filtering complexity M fs 3.5
fs2 for a given specification in unscaled units.
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Optimal Filters: minimax error criterion
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
use weight function, s(), to allow different errors

in different frequency bands
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

symmetric filter has zeros on unit circle or in reciprocal pairs
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

Response of symmetric filter is a polynomial in cos
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

Alternation Theorem: M
2 + 2 maximal errors with alternating signs
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

Remez Exchange Algorithm (also known as Parks-McLellan Algorithm)
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines

multiple constant-gain bands separated by transition regions
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


very robust, works for filters with M > 1000
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


Efficient: computation M 2
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


can go mad in the transition regions
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


Modified version works on arbitrary gain function
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


Does not always converge
Summary
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
Example Design
FIR Pros and Cons
Summary
MATLAB routines


Does not always converge

For further details see Mitra: 10.
MATLAB routines
Optimal Filters
Alternation Theorem
Remez Exchange
Algorithm
firpm
firpmord
cfirpm
remez
optimal FIR filter design

estimate require order for firpm
arbitrary-response filter design
[obsolete] optimal FIR filter design
Example Design
FIR Pros and Cons
Summary
MATLAB routines

Optimal FIR Filters

Uploaded by

Document Information

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

Optimal FIR Filters

Uploaded by

Copyright:

Available Formats

7: Optimal FIR filters

7: Optimal FIR filters

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

= h[0] + 2 1 h[n] cos n

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

= h[0] + 2 1 h[n] cos n

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

= h[0] + 2 1 h[n] cos n

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

= h[0] + 2 1 h[n] cos n

Example: lowpass filter

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

= h[0] + 2 1 h[n] cos n

Example: lowpass filter

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

= h[0] + 2 1 h[n] cos n

Example: lowpass filter

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

= h[0] + 2 1 h[n] cos n

Example: lowpass filter

DSP and Digital Filters (2014-5490)

We restrict ourselves to zero-phase filters of odd length M + 1, symmetric

= h[0] + 2 1 h[n] cos n

Example: lowpass filter

DSP and Digital Filters (2014-5490)

DSP and Digital Filters (2014-5490)

DSP and Digital Filters (2014-5490)

DSP and Digital Filters (2014-5490)

DSP and Digital Filters (2014-5490)

DSP and Digital Filters (2014-5490)

DSP and Digital Filters (2014-5490)

DSP and Digital Filters (2014-5490)

H() = H(e ) = h[0] + 2

DSP and Digital Filters (2014-5490)

H() = H(e ) = h[0] + 2

DSP and Digital Filters (2014-5490)

H() = H(e ) = h[0] + 2

DSP and Digital Filters (2014-5490)

H() = H(e ) = h[0] + 2

But cos n = Tn (cos ): Chebyshev polynomial of 1st kind

cos 2 = 2 cos2 1 = T2 (cos )

DSP and Digital Filters (2014-5490)

H() = H(e ) = h[0] + 2

But cos n = Tn (cos ): Chebyshev polynomial of 1st kind

cos 2 = 2 cos2 1 = T2 (cos )

Tn+1 (x) = 2xTn (x) Tn1 (x) with T0 (x) = 1, T1 (x) = x

DSP and Digital Filters (2014-5490)

H() = H(e ) = h[0] + 2

But cos n = Tn (cos ): Chebyshev polynomial of 1st kind

cos 2 = 2 cos2 1 = T2 (cos )

Tn+1 (x) = 2xTn (x) Tn1 (x) with T0 (x) = 1, T1 (x) = x