Basic IIR Digital Filter

Structures
• The causal IIR digital filters we are
concerned with in this course are
characterized by a real rational transfer
function of z 1 or, equivalently by a constant
coefficient difference equation
• From the difference equation representation,
it can be seen that the realization of the
causal IIR digital filters requires some form
of feedback
Copyright © 2001, S. K. Mitra
 Basic IIR Digital Filter
Structures
• An N-th order IIR digital transfer function is
characterized by 2N+1 unique coefficients,
and in general, requires 2N+1 multipliers
and 2N two-input adders for implementation
• Direct form IIR filters: Filter structures in
which the multiplier coefficients are
precisely the coefficients of the transfer
function
Copyright © 2001, S. K. Mitra
 Direct Form IIR Digital Filter
Structures
• Consider for simplicity a 3rd-order IIR filter
with a transfer function

P( z ) p0  p1z 1  p2 z 2  p3 z 3
H ( z)  
D( z ) 1  d1z 1  d 2 z 2  d3 z 3

• We can implement H(z) as a cascade of two

filter sections as shown on the next slide

Copyright © 2001, S. K. Mitra

 Direct Form IIR Digital Filter
Structures
W (z )
X (z ) H1 (z ) H 2 (z ) Y (z )

where
W ( z)
H1( z )   P( z )  p0  p1z 1  p2 z  2  p3 z 3
X ( z)
Y ( z) 1 1
H 2 ( z)   
W ( z ) D( z ) 1  d1z 1  d 2 z  2  d3 z 3
Copyright © 2001, S. K. Mitra
 Direct Form IIR Digital Filter
Structures
• The filter section H1(z ) can be seen to be
an FIR filter and can be realized as shown
below
w[n]  p0 x[n]  p1x[n  1]  p2 x[n  2]  p3 x[n  3]

Copyright © 2001, S. K. Mitra

 Direct Form IIR Digital Filter
Structures
• The time-domain representation of H 2 (z ) is
given by
y[n]  w[n]  d1 y[n  1]  d 2 y[n  2]  d3 y[n  3]
Realization of H 2 (z )
follows from the
above equation
and is shown on
the right
Copyright © 2001, S. K. Mitra
 Direct Form IIR Digital Filter
Structures
• A cascade of the two structures realizing H1(z )
and H 2 (z ) leads to the realization of H (z )
shown below and is known as the direct
form I structure

Copyright © 2001, S. K. Mitra

 Direct Form IIR Digital Filter
Structures
• Note: The direct form I structure is
noncanonic as it employs 6 delays to realize
a 3rd-order transfer function
• A transpose of the
direct form I structure
is shown on the right
and is called the direct
form t I structure

Copyright © 2001, S. K. Mitra

 Direct Form IIR Digital Filter
Structures
• Various other noncanonic direct form
structures can be derived by simple block
diagram manipulations as shown below

Copyright © 2001, S. K. Mitra

 Direct Form IIR Digital Filter
Structures
• Observe in the direct form structure shown
below, the signal variable at nodes 1 and 1'
are the same, and hence the two top delays
can be shared

Copyright © 2001, S. K. Mitra

 Direct Form IIR Digital Filter
Structures
• Likewise, the signal variables at nodes 2
and 2' are the same, permitting the sharing
of the middle two delays
• Following the same argument, the bottom
two delays can be shared
• Sharing of all delays reduces the total
number of delays to 3 resulting in a canonic
realization shown on the next slide along
with its transpose structure
Copyright © 2001, S. K. Mitra
 Direct Form IIR Digital Filter
Structures

• Direct form realizations of an N-th order IIR

transfer function should be evident
Copyright © 2001, S. K. Mitra
 Cascade Form IIR Digital
Filter Structures
• By expressing the numerator and the
denominator polynomials of the transfer
function as a product of polynomials of
lower degree, a digital filter can be realized
as a cascade of low-order filter sections
• Consider, for example, H(z) = P(z)/D(z)
expressed as
P( z ) P1( z ) P2 ( z ) P2 ( z )
H ( z)  
D( z ) D1( z ) D2 ( z ) D3 ( z )
Copyright © 2001, S. K. Mitra
 Cascade Form IIR Digital
Filter Structures
• Examples of cascade realizations obtained
by different pole-zero pairings are shown
below

Copyright © 2001, S. K. Mitra

 Cascade Form IIR Digital
Filter Structures
• Examples of cascade realizations obtained
by different ordering of sections are shown
below

Copyright © 2001, S. K. Mitra

 Cascade Form IIR Digital
Filter Structures
• There are altogether a total of 36 different
cascade realizations of
P1 ( z ) P2 ( z ) P2 ( z )
H ( z) 
D1 ( z ) D2 ( z ) D3 ( z )
based on pole-zero-pairings and ordering
• Due to finite wordlength effects, each such
cascade realization behaves differently from
others

Copyright © 2001, S. K. Mitra

 Cascade Form IIR Digital
Filter Structures
• Usually, the polynomials are factored into a
product of 1st-order and 2nd-order
polynomials:
 1  1k z 1   2k z 2 
H ( z )  p0   1 2 

k  1  1k z   2 k z 
• In the above, for a first-order factor
 2k   2k  0

Copyright © 2001, S. K. Mitra

 Cascade Form IIR Digital
Filter Structures
• Consider the 3rd-order transfer function
 1 11z 1  1 12 z 1   22 z  2 
H ( z )  p0  
1  1

2 
 111z  1  12 z   22 z 
• One possible realization is shown below

Copyright © 2001, S. K. Mitra

 Cascade Form IIR Digital
Filter Structures
• Example - Direct form II and cascade form
realizations of
0.44 z 1  0.362 z  2  0.02 z 3
H ( z) 
1 0.4 z 1  0.18 z  2 0.2 z 3

  0.440.362 z 1 0.02 z 2  z 1 
1 2  1 
 1 0.8 z  0.5 z  10.4 z 
are shown on the next slide

Copyright © 2001, S. K. Mitra

Cascade Form IIR Digital
Filter Structures

Direct form II Cascade form

Copyright © 2001, S. K. Mitra

 Parallel Form IIR Digital Filter
Structures
• A partial-fraction expansion of the transfer
1
function in z leads to the parallel form I
structure
• Assuming simple poles, the transfer function
H(z) can be expressed as
  0 k  1k z 1 
H ( z)   0    1 2 
k  11k z  2 k z 
• In the above for a real pole  2k   1k  0
Copyright © 2001, S. K. Mitra
Parallel Form IIR Digital Filter
Structures
• A direct partial-fraction expansion of the
transfer function in z leads to the parallel
form II structure
• Assuming simple poles, the transfer function
H(z) can be expressed as
1 2

H ( z)   0   
 z  z 
2 
0k 2k
1
k  11k z  2 k z 
• In the above for a real pole  2k   2k  0
Copyright © 2001, S. K. Mitra
 Parallel Form IIR Digital Filter
Structures
• The two basic parallel realizations of a 3rd-
order IIR transfer function are shown below

Parallel form I Parallel form II

Copyright © 2001, S. K. Mitra
Parallel Form IIR Digital Filter
Structures
• Example - A partial-fraction expansion of
0.44 z 1  0.362 z  2  0.02 z 3
H ( z) 
1 0.4 z 1  0.18 z  2 0.2 z 3
1
in z yields
0.6  0.5  0.2 z 1
H ( z )   0.1  1

1 0.4 z 1 0.8 z 1  0.5 z  2

Copyright © 2001, S. K. Mitra

Parallel Form IIR Digital Filter
Structures
• The corresponding parallel form I realization
is shown below

Copyright © 2001, S. K. Mitra

Parallel Form IIR Digital Filter
Structures
• Likewise, a partial-fraction expansion of
H(z) in z yields
0.24 z 1 0.2 z 1  0.25 z 1
H ( z)  1

1 0.4 z 1 0.8 z 1  0.5 z  2

• The corresponding
parallel form II
realization is shown
on the right
Copyright © 2001, S. K. Mitra
 Realization Using MATLAB
• The cascade form requires the factorization
of the transfer function which can be
developed using the M-file zp2sos
• The statement sos = zp2sos(z,p,k)
generates a matrix sos containing the
coefficients of each 2nd-order section of the
equivalent transfer function H(z) determined
from its pole-zero form

Copyright © 2001, S. K. Mitra

 Realization Using MATLAB
• sos is an L  6 matrix of the form
 p01 p11 p21 d01 d11 d21 
 p02 p12 p22 d02 d12 d22 
sos   
p 
 0L p1L p2L d0L d1L d2L 
whose i-th row contains the coefficients { pi }
and {di }, of the the numerator and
denominator polynomials of the i-th 2nd-
order section Copyright © 2001, S. K. Mitra
 Realization Using MATLAB
• L denotes the number of sections
• The form of the overall transfer function is
given by
L L p  p z 1  p z  2
H ( z )   H i ( z )   0i 1i 2i
1  d z  2
i 1 d
i 1 0i  d1i z 2i
• Program 6_1 can be used to factorize an
FIR and an IIR transfer function

Copyright © 2001, S. K. Mitra

 Realization Using MATLAB
• Note: An FIR transfer function can be
treated as an IIR transfer function with a
constant numerator of unity and a
denominator which is the polynomial
describing the FIR transfer function

Copyright © 2001, S. K. Mitra

 Realization Using MATLAB
• Parallel forms I and II can be developed
using the functions residuez and
residue, respectively
• Program 6_2 uses these two functions

Copyright © 2001, S. K. Mitra

 Realization of Allpass Filters
• An M-th order real-coefficient allpass
transfer function AM (z ) is characterized by
M unique coefficients as here the numerator
is the mirror-image polynomial of the
denominator
• A direct form realization of AM (z ) requires
2M multipliers
• Objective - Develop realizations of AM (z )
requiring only M multipliers
Copyright © 2001, S. K. Mitra
 Realization Using Multiplier
Extraction Approach
• Now, an arbitrary allpass transfer function
can be expressed as a product of 2nd-order
and/or 1st-order allpass transfer functions
• We consider first the minimum multiplier
realization of a 1st-order and a 2nd-order
allpass transfer functions

Copyright © 2001, S. K. Mitra

First-Order Allpass Structures
• Consider first the 1st-order allpass transfe
function given by
t11  z 1, t22   z 1, t11t22  t12t21  1
• We shall realize the above transfer function
in the form a structure containing a single
multiplier d1 as shown below
Y2
X1
d1
Y1
X2 Copyright © 2001, S. K. Mitra
First-Order Allpass Structures
• We express the transfer function A1( z )  Y1 / X1
in terms of the transfer parameters of the
two-pair as
t12t21d1 t11  d1 (t11t22  t12t21 )
A1( z )  t11  
1 d1t22 1 d1t22
• A comparison of the above with
d1  z 1
A1( z ) 
1 d1z 1
yields
t11  z 1, t22   z 1, t11t22  t12t21  1
Copyright © 2001, S. K. Mitra
First-Order Allpass Structures
1 1
• Substituting t11  z and t22   z in
t11t22  t12t21  1 we get
t12t21  1  z 2
• There are 4 possible solutions to the above
equation:
Type 1A: t11  z 1, t22   z 1, t12  1  z 2 , t21  1
Type 1B:
t11  z 1, t22   z 1, t12  1  z 1, t21  1  z 1
Copyright © 2001, S. K. Mitra
 First-Order Allpass Structures

• Type 1A t : t11  z 1, t22   z 1, t12  1, t21  1  z 2

• Type 1B t :
t11  z 1, t22   z 1, t12  1  z 1, t21  1  z 1
• We now develop the two-pair structure for
the Type 1A allpass transfer function

Copyright © 2001, S. K. Mitra

First-Order Allpass Structures
• From the transfer parameters of this allpass
we arrive at the input-output relations:
1
Y2  X1  z X 2
1 2 1
Y1  z X1  (1  z ) X 2  z Y2  X 2
• A realization of the above two-pair is
sketched below

Copyright © 2001, S. K. Mitra

First-Order Allpass Structures
• By constraining the X 2 , Y2 terminal-pair
with the multiplier d1 , we arrive at the
Type 1A allpass filter structure shown
below

Type 1A

Copyright © 2001, S. K. Mitra

First-Order Allpass Structures
• In a similar fashion, the other three single-
multiplier first-order allpass filter structures
can be developed as shown below

Type 1B Type 1A t

Type 1B t
Copyright © 2001, S. K. Mitra
 Second-Order Allpass
Structures
• A 2nd-order allpass transfer function is
characterized by 2 unique coefficients
• Hence, it can be realized using only 2
multipliers
• Type 2 allpass transfer function:
d1d2  d1z 1 z 2
A2( z ) 
1  d1z 1  d1d 2 z 2

Copyright © 2001, S. K. Mitra

Type 2 Allpass Structures

Copyright © 2001, S. K. Mitra

 Type 3 Allpass Structures
• Type 3 allpass transfer function:
d2  d1z 1 z 2
A3( z )  1 2
1  d1z  d 2 z

Copyright © 2001, S. K. Mitra

Type 3 Allpass Structures

Copyright © 2001, S. K. Mitra

 Realization Using Multiplier
Extraction Approach
• Example - Realize
0.2 0.18 z 1  0.4 z  2  z 3
A3( z ) 
1 0.4 z 1  0.18 z  2 0.2 z 3
( 0.4 z 1 )(0.5 0.8 z 1  z 2 )

(10.4 z 1 )(1 0.8 z 1  0.5 z  2 )
• A 3-multiplier realization of the above
allpass transfer function is shown below

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• The stability test algorithm described earlier
in the course also leas to an elegant
realization of an Mth-order allpass transfer
function
• The algorithm is based on the development
of a series of (m  1)th-order allpass transfer
functions Am1(z ) from an mth-order allpass
transfer function Am (z ) for m  M , M  1, ...,1
Copyright © 2001, S. K. Mitra
 Realization Using Two-Pair
Extraction Approach
• Let
d m  d m1z 1  d m 2 z  2  ...  d1z ( m1)  z  m
Am ( z )  1 2 ( m 1) m
1 d1z  d 2 z  ...  d m1z dm z
• We use the recursion
Am ( z )  km
Am1( z )  z[ ], m  M , M  1, ...,1
1 km Am ( z )
where km  Am ()  dm
• It has been shown earlier that AM (z ) is
stable if and only if
km  1 for m  M , M  1, ...,1
2
Copyright © 2001, S. K. Mitra
 Realization Using Two-Pair
Extraction Approach
• If the allpass transfer function Am1(z ) is
expressed in the form
d m' 1  d m'  2 z 1  ...  d1' z ( m 2 )  z ( m1)
Am1( z )  ( m2)
1 d1' z 1  ...  d m'  2 z   d m' 1z ( m1)

then the coefficients of Am1(z ) are simply

related to the coefficients of Am (z ) through
d i  d m d m i
di 
' , 1  i  m 1
1  dm2
Copyright © 2001, S. K. Mitra
 Realization Using Two-Pair
Extraction Approach
• To develop the realization method we
express Am (z ) in terms of Am1(z ):
km  z 1 Am 1 ( z )
Am ( z ) 
1 km z 1 Am 1 ( z )

• We realize Am (z ) in the form shown below

Y2
X1 t11 t12
Am 1 (z )
t21 t22
Y1
Am (z ) X2
Copyright © 2001, S. K. Mitra
 Realization Using Two-Pair
Extraction Approach
• The transfer function Am( z )  Y1/ X1 of the
constrained two-pair can be expressed as
t11  (t11t22  t12t21 ) Am1 ( z )
Am( z ) 
1 t22 Am1 ( z )

• Comparing the above with

km  z 1 Am 1 ( z )
Am ( z ) 
1 km z 1 Am 1 ( z )
we arrive at the two-pair transfer parameters
Copyright © 2001, S. K. Mitra
 Realization Using Two-Pair
Extraction Approach
1
t11  km , t22   km z
1
t11t22  t12t21   z
1
• Substituting t11  km and t22   km z in the
equation above we get
2 1
t12t21  (1  km ) z
• There are a number of solutions for t12 and
t21

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• Some possible solutions are given below:
1 1
t11  km , t22   km z , t12  z , t21  1  km
2

1 1
t11  km , t22   km z , t12  (1  km ) z , t21  1  km
1 2 1
t11  km , t22   km z , t12  1  km z , t21  1  km
2

1 2 1
t11  km , t22   km z , t12  (1  km ) z , t21 1

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• Consider the solution
1 2 1
t11  km , t22   km z , t12  (1  km ) z , t21  1
• Corresponding input-output relations are
2 1
Y1  km X1  (1  km ) z X 2
1
Y2  X1  km z X 2
• A direct realization of the above equations
leads to the 3-multiplier two-pair shown on
the next slide
Copyright © 2001, S. K. Mitra
 Realization Using Two-Pair
Extraction Approach

• The transfer parameters

1 1
t11  km , t22   km z , t12  (1  km ) z , t21  1  km
lead to the 4-multiplier two-pair structure
shown below

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• Likewise, the transfer parameters
t11  km , t22   km z 1, t12  1  km2 z 1, t21  1  km2
lead to the 4-multiplier two-pair structure
shown below

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• A 2-multiplier realization can be derived by
manipulating the input-output relations:
2 1
Y1  km X1  (1  km ) z X 2
1
Y2  X1  km z X 2
• Making use of the second equation, we can
rewrite the first equation as
1
Y1  kmY2  z X 2

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• A direct realization of
1
Y1  kmY2  z X 2
1
Y2  X1  km z X 2
lead to the 2-multiplier two-pair structure,
known as the lattice structure, shown below

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• Consider the two-pair described by
1 1
t11  km , t22   km z , t12  (1  km ) z , t21  1  km
• Its input-output relations are given by
1
Y1  km X1  (1  km ) z X 2
1
Y2  (1  km ) X1  km z X 2
• Define
1
V1  km ( X1  z ) X 2
Copyright © 2001, S. K. Mitra
 Realization Using Two-Pair
Extraction Approach
• We can then rewrite the input-output
1
relations as Y1  V1  z X 2 and Y2  X1  V1
• The corresponding 1-multiplier
V1 realization
is shown below
V1

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• An mth-order allpass transfer function Am (z )
is then realized by constraining any one of
the two-pairs developed earlier by the
(m  1) th-order allpass transfer function Am1(z )

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• The process is repeated until the
constraining transfer function is A0 (z )  1
• The complete realization of AM (z ) based on
the extraction of the two-pair latttice is
shown below

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• It follows from our earlier discussion that
AM (z ) is stable if the magnitudes of all
multiplier coefficients in the realization are
less than 1, i.e., | km |  1 for m  M , M  1, ...,1
• The cascaded lattice allpass filter structure
requires 2M multipliers
• A realization with M multipliers is obtained if
instead the single multiplier two-pair is used
Copyright © 2001, S. K. Mitra
 Realization Using Two-Pair
Extraction Approach
• Example - Realize
 0.2  0.18z 1  0.4 z 2  z 3
A3( z ) 
1  0.4 z 1  0.18z 2  0.2 z 3
d1  d 2 z 1  d1z 2  z 3

1  d1z 1  d 2 z  2  d3 z 3

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• We first realize A3( z ) in the form of a
lattice two-pair characterized by the
multiplier coefficient k3  d3  0.2
and constrained by a 2nd-order allpass A2 ( z )
as indicated below

k3   0.2
Copyright © 2001, S. K. Mitra
 Realization Using Two-Pair
Extraction Approach
• The allpass transfer function A2 ( z ) is of the
form
d 2'  d1' z 1  z  2
A2 ( z )  ' 1 ' 2
1 d1z  d 2 z
• Its coefficients are given by
d1  d3d 2 0.4( 0.2)(0.18)

d1'   0.4541667
1 d32 1( 0.2) 2

d 2  d3d1 0.18( 0.2)(0.4)

d2 
'   0.2708333
1 d32 1( 0.2) 2

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• Next, the allpass A2 ( z ) is realized as a
lattice two-pair characterized by the
multiplier coefficient k2  d 2'  0.2708333
and constrained by an allpass A1( z ) as
indicated below

k3   0.2, k2 0.2708333
Copyright © 2001, S. K. Mitra
 Realization Using Two-Pair
Extraction Approach
• The allpass transfer function A1( z ) is of the
form d1"  z 1
A1( z ) 
1  d1" z 1
• It coefficient is given by
d1'  d 2' d1' d1' 0.4541667
d1"   '   0.3573771
1( d 2' ) 2 1 d 2 1.2708333

Copyright © 2001, S. K. Mitra

 Realization Using Two-Pair
Extraction Approach
• Finally, the allpass A1( z ) is realized as a
lattice two-pair characterized by the
multiplier coefficient k1  d1"  0.3573771
and constrained by an allpass A0( z )  1 as
indicated below

k3   0.2, k2 0.2708333
k1  0.3573771 Copyright © 2001, S. K. Mitra
 Cascaded Lattice Realization
Using MATLAB
• The M-file poly2rc can be used to realize
an allpass transfer function in the cascaded
lattice form
• To this end Program 6_3 can be employed

Copyright © 2001, S. K. Mitra