2008/6/7

Chapter 8

Digital Filter Structures

清大電機系林嘉文
[email protected]
03-5731152

Original PowerPoint slides prepared by S. K. Mitra 8-1

© The McGraw-Hill Companies, Inc., 2007

Block Diagram Representation

• The convolution sum description of an LTI discrete-time
system can, in principle, be used to implement the system
• Here the input
input-output
output relation involves a finite sum of
products:

• On the other hand, an FIR system can be implemented

using the convolution sum which is a finite sum of products:

• The implementation of an LTI digital filter can be either in

software or hardware form, depending on applications
• In either case, the signal variables and the filter coefficients
cannot be represented with infinite precision
Original PowerPoint slides prepared by S. K. Mitra 8-2
© The McGraw-Hill Companies, Inc., 2007

1
 2008/6/7

Block Diagram Representation

• A structural representation using interconnected basic
building blocks is the first step in the hardware or software
implementation of an LTI digital filter
• In the time domain, the input-output relations of an LTI
digital filter is given by the convolution sum

or, by the linear constant coefficient difference equation

• For the implementation of an LTI digital filter, the input-

output relationship must be described by a valid
computational algorithm

Original PowerPoint slides prepared by S. K. Mitra 8-3

© The McGraw-Hill Companies, Inc., 2007

Block Diagram Representation

• Consider the causal first-order LTI digital filter shown below

• The filter is described by the difference equation

y[n] = −d1y[n −1] + p0x[n] + p1x[n −1]
• Using the above equation we can compute y[n] for n ≥ 0
knowing y[−1] and the input x[n] for n ≥ −1
y[0] = −d1y[−1] + p0x[0] + p1x[−1]
y[1] = −d1y[0] + p0x[1] + p1x[0]
Original PowerPoint slides prepared by S. K. Mitra
… 8-4
© The McGraw-Hill Companies, Inc., 2007

2
 2008/6/7

Basic Building Blocks

• The computational algorithm of an LTI digital filter can be
conveniently represented in block diagram form using the
basic building blocks shown below

Original PowerPoint slides prepared by S. K. Mitra 8-5

© The McGraw-Hill Companies, Inc., 2007

Basic Building Blocks

• Advantages of block diagram representation
– Easy to write down the computational algorithm by
inspection
– Easy to analyze the block diagram to determine the
explicit relation between the output and input
– Easy to manipulate a block diagram to derive other
“equivalent” block diagrams yielding different
computational algorithms
– Easy to determine the hardware requirements
– Easier to develop block diagram representations from
the transfer function directly

Original PowerPoint slides prepared by S. K. Mitra 8-6

© The McGraw-Hill Companies, Inc., 2007

3
 2008/6/7

Analysis of Block Diagrams

• Write down the expressions for the output signals of each
adder as a sum of its input signals, and develop a set of
eq ations relating the filter inp
equations inputt and o
output
tp t signals in terms
of all internal signals
• Eliminate the unwanted internal variables to obtain the
expression for the output signal as a function of the input
signal and the filter parameters (multiplier coefficients)

Original PowerPoint slides prepared by S. K. Mitra 8-7

© The McGraw-Hill Companies, Inc., 2007

Analysis of Block Diagrams

• Example - Consider the following single-loop feedback
structure

• The output E(z) of the adder is

E(z) = X(z) + G2(z)Y(z)
• But from the figure
Y(z) = G1(z)E(z)
• Eliminating E(z) from the previous equations we arrive at
[1− G1(z)G2(z)]Y(z) = G1(z)X(z)
which leads to
Original PowerPoint slides prepared by S. K. Mitra 8-8
© The McGraw-Hill Companies, Inc., 2007

4
 2008/6/7

Analysis of Block Diagrams

• Example – Analyze the following cascade lattice structure

• The output signals of the four adders are given by

W1 = X − αS
S2
W2 = W1 − δS1
W3 = S1 + εW2
Y = βW1 + γS2
Original PowerPoint slides prepared by S. K. Mitra 8-9
© The McGraw-Hill Companies, Inc., 2007

Analysis of Block Diagrams

• From the figure we observe
S1 = z−1W2
S2 = z−1W3
• Substituting the last two relations in the first four equations
we get
W1 = X − αz−1W3
W2 = W1 − δz−1W2
W3 = z−11W2 + εW2
Y = βW1 + γz−1W3
• From the second equation we get W2 = W1/(1 + δz−1) and
from the third equation we get W3 = (ε + z−1)W2
Original PowerPoint slides prepared by S. K. Mitra 8-10
© The McGraw-Hill Companies, Inc., 2007

5
 2008/6/7

Analysis of Block Diagrams

• Combining the last two equations we get

• Substituting the above equation in

W1 = X − αz−1W3
Y = βW1 + γz−1W3
we finally arrive at

Original PowerPoint slides prepared by S. K. Mitra 8-11

© The McGraw-Hill Companies, Inc., 2007

The Delay-Free Loop Problem

• For physical realizability of the digital filter structure, it is
necessary that the block diagram representation contains
no dela
delay-free
free loops
• To illustrate the delay-free loop problem consider the
structure below

• Analysis of this structure yields

u[n] = w[n] + y[n]
y[n] = B(v[n] + Au[n])
• As a result, y[n] = B(v[n] + A(w[n] + y[n])) 8-12
Original PowerPoint slides prepared by S. K. Mitra
© The McGraw-Hill Companies, Inc., 2007

6
 2008/6/7

The Delay-Free Loop Problem

• The determination of the current value of y[n] requires the
knowledge of the same value
• However, this is physically impossible to achieve due to the
finite time required to carry out all arithmetic operations on
a digital machine
• Solution: Replace the portion of the overall structure
containing the delay-free loops by an equivalent realization
with no delay-free loops

Original PowerPoint slides prepared by S. K. Mitra 8-13

© The McGraw-Hill Companies, Inc., 2007

Canonic & Noncanonic Structures

• A digital filter structure is said to be canonic if the number
of delays in the block diagram representation is equal to
the order of the transfer function
• Otherwise, it is a noncanonic structure
• The structure shown below is noncanonic as it employs
two delays to realize a first-order difference equation
y[n] = −d1y[n −1] + p0x[n] + p1x[n −1]

Original PowerPoint slides prepared by S. K. Mitra 8-14

© The McGraw-Hill Companies, Inc., 2007

7
 2008/6/7

Equivalent Structures
• Two digital filter structures are defined to be equivalent if
they have the same transfer function
• A simple way to generate an equivalent structure from a
given realization is via the transpose operation:
– Reverse all paths
– Replace pick-off nodes by adders, and vice versa
– Interchange the input and output nodes

Original PowerPoint slides prepared by S. K. Mitra 8-15

© The McGraw-Hill Companies, Inc., 2007

Equivalent Structures
• A redrawn transposed structure is shown below

• All other methods for developing equivalent structures are

based on a specific algorithm for each structure
• There are literally an infinite number of equivalent
structures realizing the same transfer function
• Under infinite precision arithmetic any given realization of
a digital filter behaves identically to any other equivalent
structure
Original PowerPoint slides prepared by S. K. Mitra 8-16
© The McGraw-Hill Companies, Inc., 2007

8
 2008/6/7

Equivalent Structures
• However, in practice, due to the finite word-length
limitations, a specific realization behaves totally differently
from its other equivalent realizations
• Hence, it is important to choose a structure that has the
least quantization effects when implemented using finite
precision arithmetic
• One way to arrive at such a structure is to determine a
large number of equivalent structures, analyze the finite
word length effects in each case
word-length case, and select the one
showing the least effects
• In certain cases, it is possible to develop a structure that
by construction has the least quantization effects

Original PowerPoint slides prepared by S. K. Mitra 8-17

© The McGraw-Hill Companies, Inc., 2007

Basic FIR Digital Filter Structures

• A causal FIR filter of order N is characterized by a transfer
function H(z) given by

• In the time-domain the input-output relation of the above

FIR filter is given by

• An FIR filter of order N is characterized by N+1

coefficients
coe c e ts a
and,
d, in ge
general,
e a , require
equ e N+1 multipliers
utp e s aand
dN
two-input adders
• Structures in which the multiplier coefficients are precisely
the coefficients of the transfer function are called direct
form structures
Original PowerPoint slides prepared by S. K. Mitra 8-18
© The McGraw-Hill Companies, Inc., 2007

9
 2008/6/7

Direct Form FIR Filter Structures

• A direct form realization of an FIR filter can be readily
developed from the convolution sum description as
indicated belo
below for N = 4

• An analysis of this structure yields

y[n] = h[0]x[n] + h[1]x[n −1] + h[2]x[n − 2] + h[3]x[n −3] +
h[4]x[n − 4]
• The direct form structure is also known as a transversal
filter
Original PowerPoint slides prepared by S. K. Mitra 8-19
© The McGraw-Hill Companies, Inc., 2007

Direct Form FIR Filter Structures

• The transpose of the direct form structure shown earlier is
indicated below

• Both direct form structures are canonic with respect to

delays

Original PowerPoint slides prepared by S. K. Mitra 8-20

© The McGraw-Hill Companies, Inc., 2007

10
 2008/6/7

Cascade Form FIR Filter Structures

• A higher-order FIR transfer function can also be realized
as a cascade of second-order FIR sections and possibly a
first order section
first-order
• To this end we express H(z) as

• Both direct form structures are canonic with respect to

delays where K = N/2 if N is even,
even and K = (N+1)/2 if N is
odd, with β2K = 0

Original PowerPoint slides prepared by S. K. Mitra 8-21

© The McGraw-Hill Companies, Inc., 2007

Cascade Form FIR Filter Structures

• A cascade realization for N = 6 is shown below
• To this end we express
p H(z)
( ) as
無法顯示圖像。您的電腦可能沒有足夠的記憶體來開啟圖像，或圖像可能已毀損。請重新啟動您的電腦，並再次開啟檔案。如果仍然出現紅色 x，您可能必須刪除圖像，然後再次插入圖像。

• Each second-order section in the above structure can also

be realized in the transposed direct form

Original PowerPoint slides prepared by S. K. Mitra 8-22

© The McGraw-Hill Companies, Inc., 2007

11
 2008/6/7

Polyphase FIR Structures

• The polyphase decomposition of H(z) leads to a parallel
form structure
• To illustrate this approach, consider a causal FIR transfer
function H(z) with N = 8:
H(z) = h[0] + h[1]z−1 + h[2]z−2 + h[3]z−3 + h[4]z−4 + h[5]z−5 +
h[6]z−6 + h[7]z−7 + h[8]z−8
• H(z) can be expressed as a sum of two terms, with one
term containing the even-indexed coefficients and the
other containing the odd-indexed coefficients
H(z) = (h[0] + h[2]z−2 + h[4]z−4 + h[6]z−6 + h[8]z−8)
+ (h[1]z−1 + h[3]z−3 + h[5]z−5 + h[7]z−7)
= (h[0] + h[2]z−2 + h[4]z−4 + h[6]z−6 + h[8]z−8)
+ zprepared
Original PowerPoint slides
−1(h[1] + h[3]z−2 + h[5]z−4 + h[7]z−6)
by S. K. Mitra 8-23
© The McGraw-Hill Companies, Inc., 2007

Polyphase FIR Structures

• By using the notation
E0(z) = h[0] + h[2]z−1 + h[4]z−2 + h[6]z−3 + h[8]z−4
E1(z) = h[1] + h[3]z−1 + h[5]z−2 + h[7]z−3
we can express H(z) as
H(z) = E0(z2) + z−1E1(z2)
• The above decomposition is more commonly known as
the 2-branch polyphase decomposition

Original PowerPoint slides prepared by S. K. Mitra 8-24

© The McGraw-Hill Companies, Inc., 2007

12
 2008/6/7

Polyphase FIR Structures

• In a similar manner, by grouping the terms in the original
expression for H(z), we can reexpress it in the form
H(z) = E0(z3) + z−1E1(z3) + z−2E2(z3)
where now
E0(z) = h[0] + h[3]z−1 + h[6]z−2
E1(z) = h[1] + h[4]z−1 + h[7]z−2
E2(z) = h[2] + h[5]z−1 + h[8]z−2
• The 3-branch p polyphase
yp decomposition
p is shown below

Original PowerPoint slides prepared by S. K. Mitra 8-25

© The McGraw-Hill Companies, Inc., 2007

Polyphase FIR Structures

• In the general case, an L-branch polyphase decomposition
of an FIR transfer function of order N is of the form

where
with h[n] = 0 for n > N

• The subfilters Em(zL) are also FIR filters

Original PowerPoint slides prepared by S. K. Mitra 8-26

© The McGraw-Hill Companies, Inc., 2007

13
 2008/6/7

Linear Phase FIR Structures

• The symmetry (or antisymmetry) property of a linear-
phase FIR filter can be exploited to reduce the number of
multipliers into almost half of that in the direct form
• Consider a length-7 Type 1 FIR transfer function with a
symmetric impulse response:
H(z) = h[0] + h[1]z−1 + h[2]z−2 + h[3]z−3 + h[2]z−4 + h[1]z−5 + h[0]z−6
• Rewriting H(z) in the form
H(z) = h[0](1 + z−6)+ h[1](z−1 + z−5) + h[2](z−2 + z−4)+ h[3]z−3

Original PowerPoint slides prepared by S. K. Mitra 8-27

© The McGraw-Hill Companies, Inc., 2007

Linear Phase FIR Structures

• A similar decomposition can be applied to a Type 2 FIR
transfer function
• For example
example, a length-8 Type 2 FIR transfer function can
be expressed as :
H(z) = h[0](1 + z−7)+ h[1](z−1 + z−6) + h[2](z−2 + z−5)+ h[3](z−3 + z−4)

• Note: The Type 1 linear-phase structure for a length-7 FIR

filter requires 4 multipliers, whereas a direct form
realization requires 7 multipliers 8-28
Original PowerPoint slides prepared by S. K. Mitra
© The McGraw-Hill Companies, Inc., 2007

14
 2008/6/7

Tapped Delay Lines

• In some applications, such as musical and sound
processing, FIR filter structures of the form shown below
are employed

• The structure consists of a chain of M1 + M2 + M3 unit

delays with taps at the input
input, at the end of first M1 delays,
delays
at the end of next M2 delays, and at the output
• Signals at these taps are then multiplied by constants α0,
α1, α2, and α3 and added to form the output
• The structure is referred to as the tapped delay line
Original PowerPoint slides prepared by S. K. Mitra 8-29
© The McGraw-Hill Companies, Inc., 2007

Basic IIR Digital Filter Structures

• We concern about causal IIR digital filters characterized
by a real rational transfer function of z−1 or, equivalently by
a constant coefficient difference equation
• The realization of the causal IIR digital filters requires
some form of feedback
• 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

Original PowerPoint slides prepared by S. K. Mitra 8-30

© The McGraw-Hill Companies, Inc., 2007

15
 2008/6/7

Direct Form IIR Filter Structures

• Consider for simplicity a 3rd-order IIR filter with a transfer
function

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

as shown below

where

Original PowerPoint slides prepared by S. K. Mitra 8-31

© The McGraw-Hill Companies, Inc., 2007

Direct Form IIR Filter Structures

• The filter section can be seen to be an FIR filter and can
be realized as shown below
w[n] = p0x[n] + p1x[n − 1] + p2x[n − 2] + p3x[n − 3]
• The time-domain representation of is given by
y[n] = w[n] − d1y[n − 1] − d2y[n − 2] − d3y[n − 3]

Original PowerPoint slides prepared by S. K. Mitra 8-32

© The McGraw-Hill Companies, Inc., 2007

16
 2008/6/7

Direct Form IIR Filter Structures

• A cascade of the two structures realizing and leads to the
realization of shown below and is known as the direct
form I structure

transpose

• The direct form I structure is non-canonic 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 It structure
Original PowerPoint slides prepared by S. K. Mitra 8-33
© The McGraw-Hill Companies, Inc., 2007

Direct Form IIR Filter Structures

• Various other non-canonic direct form structures can be
derived by simple block diagram manipulations as shown
below

• Observe in the right-hand-side direct form structure, the

signal variable at nodes and are the same, and
hence the two top delays can be shared 8-34
Original PowerPoint slides prepared by S. K. Mitra
© The McGraw-Hill Companies, Inc., 2007

17
 2008/6/7

Cascade Form IIR 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

• Examples of cascade:

Original PowerPoint slides prepared by S. K. Mitra 8-35

© The McGraw-Hill Companies, Inc., 2007

Cascade Form IIR Filter Structures

• There are a total of 36 different cascade realizations of

based on pole-zero-pairings and ordering

• Due to finite word-length effects, each such cascade
realization behaves differently from others
• Usually, the polynomials are factored into a product of 1st-
order and 2nd-order polynomials

• In the above, for a first-order factor α2k = β2k = 0

Original PowerPoint slides prepared by S. K. Mitra 8-36
© The McGraw-Hill Companies, Inc., 2007

18
 2008/6/7

Cascade Form IIR Filter Structures

• Consider the 3rd-order transfer function

• One possible realization is shown below

Original PowerPoint slides prepared by S. K. Mitra 8-37

© The McGraw-Hill Companies, Inc., 2007

Parallel Form IIR Filter Structures

• A partial-fraction expansion of the transfer function in z−1
leads to the parallel form I structure
• Assuming simple poles
poles, the transfer function H(z) can be
expressed as

• In the above for a real pole α2k = γ1k = 0

• A direct partial-fraction expansion of the transfer function
in z leads
eads to the
t e pa
parallel
a e form
o II st
structure
uctu e
• Assuming simple poles, the transfer function H(z) can be
expressed as

• In the above for a real pole α2k = δ2k = 0

Original PowerPoint slides prepared by S. K. Mitra 8-38
© The McGraw-Hill Companies, Inc., 2007

19
 2008/6/7

Parallel Form IIR Filter Structures

• The two basic parallel realizations of a 3rd-order IIR
transfer function are shown below

Original PowerPoint slides prepared by S. K. Mitra 8-39

© The McGraw-Hill Companies, Inc., 2007

Parallel Form IIR Filter Structures

• Example - A partial-fraction expansion of

in z−1 yields

Original PowerPoint slides prepared by S. K. Mitra 8-40

© The McGraw-Hill Companies, Inc., 2007

20
 2008/6/7

Parallel Form IIR Filter Structures

• Likewise, a partial-fraction expansion of H(z) in z yields

• The corresponding parallel form II realization is shown

below

Original PowerPoint slides prepared by S. K. Mitra 8-41

© The McGraw-Hill Companies, Inc., 2007

Realizations of All-Pass Filter

• 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
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
• An arbitrary allpass transfer function can be expressed as
a product of 2nd
2nd-order
order and/or 1st
1st-order
order allpass transfer
functions
• We consider first the minimum multiplier realization of a
1st-order and a 2nd-order allpass transfer functions

Original PowerPoint slides prepared by S. K. Mitra 8-42

© The McGraw-Hill Companies, Inc., 2007

21
 2008/6/7

Digital Two-Pairs
• The LTI discrete-time systems considered so far are
single-input, single-output structures characterized by a
transfer function
• Often, such a system can be efficiently realized by
interconnecting two-input, two-output structures, more
commonly called two-pairs
• Figures below show two commonly used block diagram
representations of a two-pair

• Here Y1 and Y2 denote the two outputs, and X1 and X2

denote the two inputs, where the dependencies on the
variable z has been omitted for simplicity 8-43
Original PowerPoint slides prepared by S. K. Mitra
© The McGraw-Hill Companies, Inc., 2007

Digital Two-Pairs
• The input-output relation of a digital two-pair is given by

• In the above relation the matrix τ given by

is called the transfer matrix of the two-pair

• It follows from the input-output
input output relation that the transfer
parameters can be found as follows:

Original PowerPoint slides prepared by S. K. Mitra 8-44

© The McGraw-Hill Companies, Inc., 2007

22
 2008/6/7

Digital Two-Pairs
• An alternate characterization of the two-pair is in terms of
its chain parameters as

where the matrix Γ given by

is called the chain matrix of the two-pair

• The relation between the transfer parameters and the
chain parameters are given by

Original PowerPoint slides prepared by S. K. Mitra 8-45

© The McGraw-Hill Companies, Inc., 2007

Digital Two-Pairs
• Cascade Connection - Γ-cascade

Here

• As a result,

• Hence

Original PowerPoint slides prepared by S. K. Mitra 8-46

© The McGraw-Hill Companies, Inc., 2007

23
 2008/6/7

Digital Two-Pairs
• Cascade Connection - τ-cascade

Here

• As a result,

• Hence

Original PowerPoint slides prepared by S. K. Mitra 8-47

© The McGraw-Hill Companies, Inc., 2007

Digital Two-Pairs
• Constrained Two-Pair

• It can be shown that

Original PowerPoint slides prepared by S. K. Mitra 8-48

© The McGraw-Hill Companies, Inc., 2007

24
 2008/6/7

First-Order All-Pass Filter Structures

• Consider first the 1st-order allpass transfer function given
by

• We shall realize the above transfer function in the form a

structure containing a single multiplier d1 as shown below

• We express the transfer function A1(z) = Y1/ X1 in terms of

the transfer parameters of the two-pair as

Original PowerPoint slides prepared by S. K. Mitra 8-49

© The McGraw-Hill Companies, Inc., 2007

First-Order All-Pass Filter Structures

• A comparison of the above with

Yields

• Substituting t11 = z−1 and t22= −z−1 in t11t22 = −1 we get

t12t21 = 1 − z−2
• There are 4 possible solutions to the above equation:
Type 1A: t11 = z−1, t22= −zz−1, t12 = 1 − z−2, t21 = 1
Type 1B: t11 = z−1, t22= −z−1, t12 = 1 + z−1, t21 = 1 − z−1
Type 1At: t11 = z−1, t22= −z−1, t12 = 1, t21 = 1 − z−2
Type 1Bt: t11 = z−1, t22= −z−1, t12 = 1 − z−1, t21 = 1 + z−1
Original PowerPoint slides prepared by S. K. Mitra 8-50
© The McGraw-Hill Companies, Inc., 2007

25
 2008/6/7

First-Order All-Pass Filter Structures

• From the transfer parameters of this allpass we arrive at
the input-output relations:
Y2 = X1 − z−1X2
Y1 = z−1X1 + (1 − z2)X2 = z−1Y2 + X2
• A realization of the above two-pair is sketched below

• By constraining the X2, Y1, terminal-pair

terminal pair with the multiplier
d1, we arrive at the Type 1A allpass filter structure shown
below:

Original PowerPoint slides prepared by S. K. Mitra 8-51

© The McGraw-Hill Companies, Inc., 2007

First-Order All-Pass Filter Structures

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

Original PowerPoint slides prepared by S. K. Mitra 8-52

© The McGraw-Hill Companies, Inc., 2007

26
 2008/6/7

Second-Order All-Pass 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:

Original PowerPoint slides prepared by S. K. Mitra 8-53

© The McGraw-Hill Companies, Inc., 2007

Type-3 All-Pass Structures

• Type 3 allpass transfer function:

Original PowerPoint slides prepared by S. K. Mitra 8-54

© The McGraw-Hill Companies, Inc., 2007

27
 2008/6/7

Realization Using Multiplier

Extraction Approaches
• Example – Realize

• A 3-multiplier cascade realization of the above allpass

transfer function is shown below

Original PowerPoint slides prepared by S. K. Mitra 8-55

© The McGraw-Hill Companies, Inc., 2007

Realization Using Two-Pair

Extraction Approaches
• 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 M
−1,...,1
• Let

• We use the recursion

where km = Am(∞) = dm
• It has been shown earlier that AM(z) is stable if and only if

Original PowerPoint slides prepared by S. K. Mitra 8-56

© The McGraw-Hill Companies, Inc., 2007

28
 2008/6/7

Realization Using Two-Pair

Extraction Approaches
• If the allpass transfer function Am−1(z) is expressed in the
form

• then the coefficients of Am−1(z) are simply related to the

coefficients of Am(z) through

• To develop the realization method we express Am(z) in

terms of Am−1(z)

Original PowerPoint slides prepared by S. K. Mitra 8-57

© The McGraw-Hill Companies, Inc., 2007

Realization Using Two-Pair

Extraction Approaches
• The transfer function Am(z) = Y1/X1 of the constrained two-
pair can be expressed as

• Comparing the above with

we arrive at the two-pair transfer parameters

• Corresponding input-output relations are

Original PowerPoint slides prepared by S. K. Mitra 8-58

© The McGraw-Hill Companies, Inc., 2007

29
 2008/6/7

Realization Using Two-Pair

Extraction Approaches
• A direct realization of the above equations leads to the
following 3-multiplier two-pair

• The transfer parameters

lead to the 4
4-multiplier
multiplier two
two-pair
pair structure shown below

Original PowerPoint slides prepared by S. K. Mitra 8-59

© The McGraw-Hill Companies, Inc., 2007

Realization Using Two-Pair

Extraction Approaches
• Likewise, the transfer parameters

lead to the 4-multiplier two-pair structure shown below

Original PowerPoint slides prepared by S. K. Mitra 8-60

© The McGraw-Hill Companies, Inc., 2007

30
 2008/6/7

Realization Using Two-Pair

Extraction Approaches
• A 2-multiplier realization can be derived by manipulating
the input-output relations:

• Making use of the second equation, we can rewrite the

equations as
Y1 = kmY2 + z−1X2
Y2 = X1 − kmz−1X2
lead to the 2-multiplier two-pair structure, known as the
lattice structure, shown below

Original PowerPoint slides prepared by S. K. Mitra 8-61

© The McGraw-Hill Companies, Inc., 2007

Realization Using Two-Pair

Extraction Approaches
• Consider the two-pair described by
t11 = km, t22 = −kmz−1, t12 = (1 − km)z−1, t21 = 1 + km
• Its input-output relations are given by
Y1 = kmX1 + (1 − km)z−1X2
Y2 = (1 + km)X1 − kmz−1X2
• Define V1 = km(X1 − z−1X2)
• We can then rewrite the input-output relations as
Y1 = V1 + z−11X2 and Y2 = X1 + V1, leading to the following 11-
multimplier architecture

Original PowerPoint slides prepared by S. K. Mitra 8-62

© The McGraw-Hill Companies, Inc., 2007

31
 2008/6/7

Realization Using Two-Pair

Extraction Approaches
• An mth-order allpass transfer function Am(z) is then
realized by constraining any one of the two-pairs of the
(m 1)th order allpass transfer ffunction
(m-1)th-order nction Am−1(z)
( )

• The process is repeated until the constraining transfer

function is A0((z)) = 1
• The realization of Am(z) based on the extraction of the two-
pair lattice is shown below

Original PowerPoint slides prepared by S. K. Mitra 8-63

© The McGraw-Hill Companies, Inc., 2007

Realization Using Two-Pair

Extraction Approaches
• It follows from our earlier discussion that Am(z) is stable if
the magnitudes of all multiplier coefficients in the realization
are less than 11, ii.e.,
e |km| < 1,
1 for m = M
M, M − 1,
1 ...,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
• Example - Realize

Original PowerPoint slides prepared by S. K. Mitra 8-64

© The McGraw-Hill Companies, Inc., 2007

32
 2008/6/7

Realization Using Two-Pair

Extraction Approaches
• We first realize A3(z) in the form of a lattice two-pair
characterized by the multiplier coefficient k3 = d3 = −0.2 and
constrained b by a 2nd
2nd-order
order allpass A2(z)
( ) as indicated belo
below

• The allpass transfer function A2(z) is of the form

• Its coefficients are given by

Original PowerPoint slides prepared by S. K. Mitra 8-65

© The McGraw-Hill Companies, Inc., 2007

Realization Using Two-Pair

Extraction Approaches
• Next, the allpass A2(z) is realized as a lattice two-pair
characterized by the multiplier coefficient k2 = d2’ = −0.2708
and constrained bby an allpass A1(z)
( ) as indicated belo
below

• The allpass transfer function A1(z) is of the form

• Its coefficients is given by

Original PowerPoint slides prepared by S. K. Mitra 8-66

© The McGraw-Hill Companies, Inc., 2007

33
 2008/6/7

Realization Using Two-Pair

Extraction Approaches
• Finally, the allpass A1(z) is realized as a lattice two-pair
characterized by the multiplier coefficient k1 = d1” = −0.3574
and constrained b by an allpass A0(z)
( ) as indicated belo
below

Original PowerPoint slides prepared by S. K. Mitra 8-67

© The McGraw-Hill Companies, Inc., 2007

Tunable Lowpass and Highpass

Digital Filters
• We have shown earlier that the 1st-order lowpass transfer
function

and the 1st-order highpass transfer function

are doubly-complementary pair

• Moreover,
Moreover they can be expressed as
HLP(z) = 1/2[1 + A1(z)]
HHP(z) = 1/2[1 − A1(z)]
• where
is a 1st-order allpass transfer function
Original PowerPoint slides prepared by S. K. Mitra 8-68
© The McGraw-Hill Companies, Inc., 2007

34
 2008/6/7

Tunable Lowpass and Highpass

Digital Filters
• A realization of HLP(z) and HHP(z) based on the allpass-
based decomposition is shown below

• The 1st-order allpass filter can be realized using any one of

the 4 single-multiplier allpass structures
• In the following example
example, the 3-dB cutoff frequency can be
varied by changing the multiplier coefficient α

Original PowerPoint slides prepared by S. K. Mitra 8-69

© The McGraw-Hill Companies, Inc., 2007

Tunable Lowpass and Highpass

Digital Filters
• Figure below shows the composite magnitude responses of
the two filters for two different values of α

Original PowerPoint slides prepared by S. K. Mitra 8-70

© The McGraw-Hill Companies, Inc., 2007

35
 2008/6/7

Tunable Bandpass and Bandstop

Digital Filters
• The 2nd-order bandpass transfer function

and the 2nd-order bandstop transfer function

also form a doubly-complementary pair

• Thus, they can be expressed in the form
HBP(z) = 1/2[1 − A2(z)]
HBS(z) = 1/2[1 + A2(z)]
• where
Original PowerPoint slides prepared by S. K. Mitra 8-71
© The McGraw-Hill Companies, Inc., 2007

Tunable Bandpass and Bandstop

Digital Filters
• A realization of HBP(z) and HBS(z) based on the allpass-
based decomposition is shown below

• The 2nd-order allpass filter is realized using a cascaded

single-multiplier lattice structure
• In the following structure,
structure the multiplier β controls the center
frequency and the multiplier α controls the 3-dB bandwidth

Original PowerPoint slides prepared by S. K. Mitra 8-72

© The McGraw-Hill Companies, Inc., 2007

36
 2008/6/7

Tunable Bandpass and Bandstop

Digital Filters
• Figure below illustrates the parametric tuning property of the
overall structure

Original PowerPoint slides prepared by S. K. Mitra 8-73

© The McGraw-Hill Companies, Inc., 2007

IIR Tapped Cascaded Lattice

Structures
• Consider the cascaded lattice structure derived earlier for
the realization of an allpass transfer function

• A typical lattice two-pair here is as shown below

• Its input-output relations are given by

Wm(z) = Wm+1(z) − kmz−1Sm(z)
S by S.(z)
Original PowerPoint slides prepared m+1
= kmWm (z) + z−1Sm(z)
K. Mitra 8-74
© The McGraw-Hill Companies, Inc., 2007

37
 2008/6/7

IIR Tapped Cascaded Lattice

Structures
• From the input-output relations we derive the chain matrix
description of the two-pair:

• The chain matrix description of the cascaded lattice

structure is therefore

• From the above equation we arrive at

Original PowerPoint slides prepared by S. K. Mitra 8-75

© The McGraw-Hill Companies, Inc., 2007

IIR Tapped Cascaded Lattice

Structures
• Using the relation S1(z) = W1(z) and the relations

• The transfer function W1(z)/X1(z) is thus an all-pole function

with the same denominator as that of the 3rd-order allpass
function A3(z)

Original PowerPoint slides prepared by S. K. Mitra 8-76

© The McGraw-Hill Companies, Inc., 2007

38
 2008/6/7

IIR Tapped Cascaded Lattice

Structures
Gray-Markel Method
• A two-stepp method to realize an Mth-order arbitraryy IIR
transfer function
H(z) = PM(z)/DM(z)
• Step 1: An intermediate allpass transfer function Am(z) =
z−MDM(z−1)/DM(z) is realized in the form of a cascaded lattice
structure
• Step 2: A set oof independent
depe de t variables
a ab es aare
e su
summed
ed witht
appropriate weights to yield the desired numerator PM(z)

Original PowerPoint slides prepared by S. K. Mitra 8-77

© The McGraw-Hill Companies, Inc., 2007

IIR Tapped Cascaded Lattice

Structures
• To illustrate the method, consider the realization of a 3rd-
order transfer function

• In the first step, we form a 3rd-order allpass transfer function

A3(z) = Y1(z)/X1(z) = z−3D3(z−1)/D3(z)
• Realization of A3(z) has been illustrated earlier resulting in
the structure shown below

Original PowerPoint slides prepared by S. K. Mitra 8-78

© The McGraw-Hill Companies, Inc., 2007

39
 2008/6/7

IIR Tapped Cascaded Lattice

Structures
• Objective: Sum the independent signal variables Y1, S1, S2,
and S3 with weights {αi} as shown below to realize the
desired numerator P3(z)

• To this end, we first analyze the cascaded lattice structure

realizing and determine the transfer functions S1(z)/ X1(z),
S2(z)/ X1(z), and S3(z)/ X1(z)

Original PowerPoint slides prepared by S. K. Mitra 8-79

© The McGraw-Hill Companies, Inc., 2007

IIR Tapped Cascaded Lattice

Structures
• We have already shown

• From the figure it follows that

and hence

• The following relation are shown previously

• From the last equation we get

Original PowerPoint slides prepared by S. K. Mitra 8-80
© The McGraw-Hill Companies, Inc., 2007

40
 2008/6/7

IIR Tapped Cascaded Lattice

Structures
• Note: The numerator of is precisely the numerator of the
allpass transfer function
• We now form

• Substituting the expressions for the various transfer

functions in the above equation we arrive at

Original PowerPoint slides prepared by S. K. Mitra 8-81

© The McGraw-Hill Companies, Inc., 2007

IIR Tapped Cascaded Lattice

Structures
• Comparing the numerator of Y0(z)/ X1(z) with the desired
numerator P3(z) and equating like powers of z−1 we obtain

• Solving the above equations we arrive at

Original PowerPoint slides prepared by S. K. Mitra 8-82

© The McGraw-Hill Companies, Inc., 2007

41
 2008/6/7

IIR Tapped Cascaded Lattice

Structures
• Example - Consider

• The corresponding intermediate allpass transfer function is

given by

• The allpass transfer function was realized earlier in the

cascaded lattice form as shown below

• where
Original PowerPoint slides prepared by S. K. Mitra 8-83
© The McGraw-Hill Companies, Inc., 2007

IIR Tapped Cascaded Lattice

Structures
• Other pertinent coefficients are:

• Substituting these coefficients in

• The final realization is as shown below

Original PowerPoint slides prepared by S. K. Mitra 8-84

© The McGraw-Hill Companies, Inc., 2007

42
 2008/6/7

FIR Cascaded Lattice Structures

• An arbitrary Nth-order FIR transfer function of the form

can be realized as a cascaded lattice structure as shown

below

• From
F figure,
fi iti follows
f ll that
h
Xm(z) = Xm−1(z) + kmz−1Ym−1(z)
Ym(z) = kmXm−1(z) + z−1Ym−1(z)

Original PowerPoint slides prepared by S. K. Mitra 8-85

© The McGraw-Hill Companies, Inc., 2007

FIR Cascaded Lattice Structures

• In matrix form the above equations can be written as

where m = 1,2,...,N
• Denote

• Then it follows from the input-output relations of the m-th

two-pair
i that
h
Hm(z) = Hm−1(z) + kmz−1Gm−1(z)
Gm(z) = kmHm−1(z) + z−1Gm−1(z)

Original PowerPoint slides prepared by S. K. Mitra 8-86

© The McGraw-Hill Companies, Inc., 2007

43
 2008/6/7

FIR Cascaded Lattice Structures

• From the previous equation we observe
H1((z)) = 1+ k1z−1, G1((z)) = k1 + z−1
where we have used the facts
H0(z) = X0(z)/ X0(z) = 1
G0(z) = Y0(z)/ X0(z) = X0(z)/ X0(z) = 1
• It follows from the above that
G1(z) = z−1(zk1+ 1) = z−1H1(z−1)
⇒ G1(z) is the mirror-image of H1(z)

Original PowerPoint slides prepared by S. K. Mitra 8-87

© The McGraw-Hill Companies, Inc., 2007

FIR Cascaded Lattice Structures

• From the input-output relations of the m-th two-pair we
obtain for m = 2:
H2(z) = H1(z) + k2z−1G1(z)
G2(z) = k2H1(z) + z−1G1(z)
• Since H1(z) and G1(z) are 1st-order polynomials, H2(z) and
G2(z) are 2nd-order polynomials
• Substituting G1(z) = z−1H1(z−1) in the two above equations
we get
H2(z) = H1(z) + k2z−2H1(z−1)
G2(z) = k2H1(z) + z−2H1(z−1)
• We can write G2(z) = z−2[k2z2H1(z) + H1(z−1)] = z−2H2(z−1)
⇒ G2(z) is the mirror-image of H2(z) 8-88
Original PowerPoint slides prepared by S. K. Mitra
© The McGraw-Hill Companies, Inc., 2007

44
 2008/6/7

FIR Cascaded Lattice Structures

• In the general case, from the input-output relations of the m-
th two-pair we obtain
Hm(z) = Hm−1(z) + kmz−1Gm−1(z)
Gm(z) = kmHm−1(z) + z−1Gm−1(z)
• It can be easily shown by induction that
• Substituting G1(z) = z−1H1(z−1) in the two above equations
we get
Gm(z) = z−mHm(z−1),
) m = 1, 1 2
2,... N
N−1
1, N
⇒ Gm(z) is the mirror-image of Hm(z)

Original PowerPoint slides prepared by S. K. Mitra 8-89

© The McGraw-Hill Companies, Inc., 2007

FIR Cascaded Lattice Structures

• To develop the synthesis algorithm, we express Hm−1(z) and
Gm−1(z) in terms of Hm(z) and Gm(z) for m = N, N−1,...,2,1
arriving at

• Substituting the expressions for

and

in the first equation we get

Original PowerPoint slides prepared by S. K. Mitra 8-90

© The McGraw-Hill Companies, Inc., 2007

45
 2008/6/7

FIR Cascaded Lattice Structures

• If we choose kN = pN, then HN−1(z) reduces to an FIR
transfer function of order N−1 and can be written in the form

• where
• Continuing the above recursion algorithm, all multiplier
coefficients of the cascaded lattice structure can be
computed
• Example - Consider
H4(z) = 1 + 1.2z−1 + 1.12z−2 + 0.12z−3 − 0.08z−4
• From the above, we observe k4 = p4 = −0.08

Original PowerPoint slides prepared by S. K. Mitra 8-91

© The McGraw-Hill Companies, Inc., 2007

FIR Cascaded Lattice Structures

• Using

we determine the coefficients of H3(z)

p3’= 0.2173913, p2’= 1.2173913, p1’= 1.2173913
• As a result
H3(z) = 1 + 1.2173913z−1 + 1.2173913z−2 + 0.2173913z−3
• Thus, k3 = p3’= 0.2173913
• Using

we determine the coefficients of H2(z)

p2”= 1.0, p1”= 1.0
Original PowerPoint slides prepared by S. K. Mitra 8-92
© The McGraw-Hill Companies, Inc., 2007

46
 2008/6/7

FIR Cascaded Lattice Structures

• As a result, H2(z) = 1 + z−1 + z−2
• From the above, we get k2 = p2”= 1
• The final recursion yields the last multiplier coefficient
k2 = p2”/(1+ k2) = 0.5
• The complete realization is shown below

k1 = 0.5, k2 =1, k3 = 0.2173913, k4 = −0.08

Original PowerPoint slides prepared by S. K. Mitra 8-93

© The McGraw-Hill Companies, Inc., 2007

47