Documente Academic
Documente Profesional
Documente Cultură
to accompany
Sanjit K. Mitra
Prepared by
Chowdary Adsumilli, John Berger, Marco Carli,
Hsin-Han Ho, Rajeev Gandhi, Martin Gawecki, Chin Kaye Koh,
Luca Lucchese, Mylene Queiroz de Farias, and Travis Smith
Chapter 8 Part 1
8.1
=
by H(z)
8.2
8.3
z1
. Thus The
The overall transfer function H(z) is given by H(z) =
1+ (1.5 + K)z1 + 0.5z2
transfer function is stable if 1.5 + K < 1.5 which is satisfied if 3 < K < 0.
G(z)
From the results of Problem 8.3, wehave H(z) =
which can be solved
1+
G(z)C(z)
H(z)
G(z)
yielding C(z) =
. Substituting the expressions for G(z) and H(z) in this
G(z)H(z)
expression we get
C(z) =
Pole-zero plots of G(z),C(z), and H(z) obtained using zplane can be easily
obtained.
8.4 The structure with internal variables is shown below. Analysis of this structure yields
_
1
d1
X(z)
V(z)
U(z)
_
z 1
_
z 1
_
z 1
_
z 1
Y(z)
d2
_1
U(z) = KX(z) + d2 z1V (z), V (z) = U(z) d1z1V (z), Y (z) = d1z1V (z) z2V (z) d1V (z).
Eliminating the internal variables we arrive at
Y (z) K(d2 + d1z1 + z2 )
H(z) =
=
.
X(z)
1++d1z1 d2 z2
(b) H(e j 0 ) =
H(z)H(z1 ) =
Kz2 D(z1 )
. Now,
D(z)
= K 2 . This implies
1
D(z)
D(z )
2
2
H(e j ) = H(z)H(z1 )
j = K , or in other words the transfer function has a
z=e
+
_
z 1
V(z)
Y(z)
k2
k1
_
z 1
U(z)
+
1
2
X(z)
V (z) = 2X(z) + U(z), U(z) = k1Y (z) + 1 X(z), Y (z) = z1V (z) + k 2z1U(z).
Eliminating the internal variables we arrive at
Y (z) [(1+ k 2 )1 + 2 ]z1
H(z) =
=
. For stability we require k1 (1+ k 2 ) < 1.
X(z) 1+ k1 (1+ k 2 )z1
8.6 An equivalent representation of the structure of Figure P8.4 with internal variables is shown
X(z)
_
0 (z 1_1)
Y(z)
_
1 (z 1_1)
_
2 (z 1_1)
_1
W(z)
Y (z)
1
=
where
X(z) D(z)
D(z) = [1 ( 0 + 2 ) 01 + 01 2 ] + [( 0 + 2 ) + 2 01 3 01 2 ]z1
+ [ 01 + 3 01 2 ]z2 01 2 z3 .
8.7
An
equivalent
representation
of
the
structure
of Figure P8.4 with internal variables is shown
1
z
X(z)
V(z)
_1
0
Y(z)
1
2z
1 z
_1
_1
1 _ 2z
U(z)
1 _ 1 z
_1
3 z
_1
1 _ 3z
W(z)
V (z) = X(z) + T2 (z)U(z), W (z) = T1 (z)V (z), U(z) = W (z) + T3 (z)V (z),
Y (z) = 0 X(z) + W (z).
Y (z)
1
Eliminating the internal variables we arrive at H(z) =
=
where
X(z) D(z)
8.8 The structure with internal variables is shown below. Analysis of this structure yields
1
1 z1
k1
k1
1
k1
k k2
W
(z)
+
k
Y
(z)
W (z) 1 1
Y (z). Substituting
(4): Y (z) =
=
2
1
1
1
2
(1 z )
1 z 1 z
1 z
k1
k k2
[X(z) + k1Y (z)] 1 1
Y (z)
Eq. (1) in Eq. (4) we then get Y (z) =
1 2
(1 z )
1 z
k1
k1 k1 + k 2 k2 z1
Y (z), or,
=
X(z)
(1 z1 ) 2
1 z1
1 z1
k (k + k k z1 )
k1
1+ 1 1 21 22
Y (z) =
X(z). Hence,
(1 z )
(1 z1 ) 2
Y (z)
k1
H(z) =
=
.
X(z)
[1+ k1 (k1 + k2 )] (2 + k1k 2 )z1 + z2
8.10
=
. Likewise, the input-output relation of the channel
1 X 2 (z)
Y2 (z) H 21 (z)
V1 (z) 1
G12 (z) Y1 (z)
separation circuit of Figure P8.7(b) is given by
=
.
1 Y2 (z)
V2 (z) G21 (z)
Hence, the overall system is characterized by
V1 (z) 1
G12 (z) 1
H12 (z) X1 (z)
1 H 21 (z)
1 X 2 (z)
V2 (z) G21 (z)
1
1
12 (z) = H
21
V2 (z) = (1 H12 (z)G
(z), and G 21(z) = H12
(z),
21 (z)) X 2 (z). Alternately, if G
(z) 1
H12 (z)H
H (z)H 21 (z) 1
21
(z), and V2 (z) = 12
then V1 (z) =
X 2
X1 (z).
H (z)
H (z)
21
12
8.12
(a) Figure P8.9(a) with internal adder output variables is shown below:
W1(z)
X(z)
Y (z)
W 2 (z)
1
k1
k2
_
z 1
_
z 1
W 3 (z)
(z1 + k1 )
W1 (z).
equation given above we obtain W 3 (z) = (z1 + k1 )W 2 (z) =
1 k2
Substituting the expression for W 3 (z)
in the first equation given above we arrive at
z1 (z1 + k1 )
z1 (z1 + k1 ) + (1 k 2 )1
Y (z) =
W1 (z) + 1W1 (z) =
W1 (z) .
1 k2
1 k2
(z1 + k1 )
W1 (z) which leads to
1 k2
(1 k 2 )X(z)
W1 (z) =
. Substituting this expression for W1 (z) in the last
1
1 k 2 2 z 2 k1
expression for Y(z) we finally arrive at the expression for input-output relation of
the structure of Figure P8.9(a)
z1 (z1 + k1 ) + (1 k 2 )1
Y (z) =
X(z).
1 k2 2 z1 2 k1
W1 (z) = X(z) + 2
It can be seen that there are two delay-free loops in the structure: one going
through the multipliers k1 and 2 , and the other one going through the multiplier
k2 .
To develop the equivalent realization without any delay-free loop we first remove
k1
k2
Y (z)
X1(z)
Y1(z)
X 2(z)
Y 2(z)
We then remove the output variable Y(z) and the two inputs to its adder as shown
below:
X(z)
k1
2
k2
X 2(z)
Y1(z)
Y 2(z)
As can be seen from the above structure, there are two delay-free loops with loop
gains 2 k1 and k2 . The corresponding expression for the determinant is
= 1 2 k1 k 2 .
X(z)
k2
2
k1 /k 2
X 2(z)
Y1(z)
Y 2(z)
k1
0
(1 k 2 )
Y1 (z) = t 21 t 22 t 23 =
Y2 (z) t 31 t 32 t 33
1
0
2
X(z)
X1 (z) .
X 2 (z)
A realization of the above structure without any delay-free loop is thus as shown
below:
X(z)
t11
t 21
Y (z)
X1 (z)
Y1 (z)
t 23
t33
X2 (z)
t13
t 31
Y 2(z)
(b) Figure P8.9(b) is redrawn below with the dashed block containing the two
delays.
7
X(z)
Y1 (z)
X1 (z)
Y (z)
X(z)
Y1 (z)
Y (z)
X1 (z)
The only delay-free loop goes through the multipliers and . The corresponding
loop-gain is and the determinant is = 1 loopgain = 1 .
The transfer matrix of the above two-pair is given by
=
=
.
Y1 (z) t 21 t 22 X1 (z) 1 X1 (z)
A realization of the structure of Figure P8.9(b) without any delay-free loop is thus
as shown below:
X(z)
1/
Y1 (z)
X1 (z) 1/
Y (z)
8.13
_
z 1
_
z 1
x[n]
1.2
_
z 1
_ 0.064
0.36
_
z 1
_
z 1
_
z 1
_ 0.0384
0.001
y[n]
The transposed form of the above structure yields another direct form realization as
indicated below:
z1
z1
z1
z1
_ 0.064
_ 0.0384
0.001
z1
z1
y[n]
1.2
0.36
x[n]
(b) A realization in the form of cascade of six first-order sections is shown below:
x[n]
_
z 1
_
z 1
0.4
0.4
_ 0.2
_ 0.2
0.4
0.4
_
z 1
_
z 1
_
z 1
_
z 1
y[n]
_
z 1
_
z 1
0.8
_
z 1
_
z 1
0.16
0.16
0.8
_
z 1
_
z 1
_ 0.12
_ 0.12
y[n]
(d) A realization in the form of cascade of two third-order sections is shown below:
_
z 1
_
z 1
x[n]
1.2
_
z 1
0.064
0.48
_
z 1
_
z 1
_ 0.12
0.016
_
z 1
y[n]
(e) A realization in the form of cascade of two first-order sections and two secondorder sections is shown below:
x[n]
_
z 1
_
z 1
0.8
0.16
_
z 1
_
z 1
0.16
0.8
_
z 1
_
z 1
_ 0.12
_ 0.12
y[n]
_4
x[n]
a + ez
_
b + fz 4
_
c + gz 4
y[n]
_
z 1
_
z 1
_
z 1
_
d + hz 4
_
z 4
h
d
8.15
+
_
z 1
g
c
y[n]
_
z 1
f
b
+
_
z 1
H(z) = E 0 (z 3 ) + z1 E1 (z 3 ) + z2 E 2 (z 3 ) where,
E 0 (z) = a + dz1 + gz2 , E1 (z) = b + ez1 + hz2 , E 2 (z) = c + fz1 . A direct 3band polyphase realization of H(z) is shown below which requires 17 unit delays
and hence is a non-canonic structure.
10
x[n]
_
_
a + dz 3 + gz 6
_
_
b + ez 3 + hz 6
y[n]
_
z 1
_
z 1
_
c + fz 3
x[n]
_
z 3
_
z 1
h
b
y[n]
e
_
z 1
f
c
_2
_4
_6
x[n]
a + cz
+ ez + gz
y[n]
_
z 1
_
_
_
b + dz 2 + fz 4+ hz 6
11
_
z 1
g
_
z 2
x[n]
y[n]
_
z 2
_
z 2
8.17
_
z 1
x[n]
_
z 1
_
z 1
+
_1
+
_1
_
z 1
_1
_
z 1
_1
_
z 1
h[1]
h[0]
_
z 1
_
z 1
h[2]
h[3]
y[n]
8.18
_
z 1
_
z 1
x[n]
_
z 1
+
_1
_1
_1
_
z 1
h[1]
h[0]
+
_1
_
z 1
_
z 1
_
z 1
_1
_1
z
h[3]
h[2]
_
z 1
h[4]
y[n]
8.19 G(z) = zN / 2 H(z). A canonic realization of both G(z) and H(z) is shown below for
N = 6.
12
8.20 Without any loss of generality, assume M = 5 which means N = 11. In this case, the
transfer function is given by
Tr (x) = 2x Tr1 (x) Tr2 (x), r 2 with T0 (x) = 1 and T1 (x) = x. Hence,
T2 (x) = 2x T1 (x) T0 (x) = 2x 2 1,
2
2
2
2
2
1 2
1 3 1
z+z 1
z+z
z+z
z+z
= h[5] + 2h[4]
+ 2h[3] 2 2 1 + 2h[2] 4 2 3 2
2
1 4 1 2
1 5
z+z 1 3 z+z 1
z+z
z+z
z+z
+2h[1] 8
8 2 +1 + 2h[0] 16 2 20 2 + 5 2
z+z 1 n
= a[n]
, where a[0] = h[5] 2h[3] + 2h[1], a[1] = 2h[4] 6h[2] +10h[0],
2
n=0
a[2] = 4h[3] 16h[1], a[3] = 8h[2] 40h[1], a[4] = 16h[1], and a[5] = 32h[0]. A
5
1+z 2 2 3
z+z 1 n
1+z 2 4
5
5
n=0
1+z 2 3 2
1+z 2 4 1
1+z 2 5
+a[3]
z + a[4] 2 z + a[5] 2 is shown below:
2
13
remaining 2 factors, and the numerator of the third stage isthe remaining
factor.
Similarly, that the denominator of the first stage can be one of the 3 factors, D1 (z),
D2 (z), and D3 (z). Likewise, the denominator of the second stage can be one of the
remaining 2 factors, and the denominator of the third stage is the remaining factor.
Hence, there are (3!) 2 = 36 different types of cascade realizations.
If the zeros of P(z) and D(z) are all real, then P(z) has 6 real zeros and D(z) has 6
real zeros. In this case, then there are (6!) 2 = 518400 different types of cascade
realizations.
K
Pi (z)
8.22 H(z) =
numerator of the first stage can be chosen in ways, the
. Here, the
1
Di (z)
i=1
K1
numerator of the second stage can be chosen in ways, and so on, until there is
1
only one possible choice for the numerator of the K th stage. Likewise,
the
K
denominator of the first stage can be chosen in ways, the denominator of the
1
K1
second stage can be chosen in ways,
and so on, until there is only one possible
1
choice for the denominator of the K th
Hence, the total number of possible
stage.
2
2
K K1 K2 2 2 2 1 2
= (K!) 2 .
cascade realizations are equal to
1
1
1
1 1
_
z D
The second-order direct form II structure with each delay replaced by the above ringing
delay is shown below:
14
_
z D
_
z D
_d
+
+
p2
_d
0.6
x[n]
_
z 1
+
+
y[n]
_
z 1
_ 0.9
_ 0.18
x[n]
8.25
_
z 1
y[n]
_
z 1
0.8
_ 0.15
_ 2.2
y[n]
_
z 1
_
z 1
0.81
_ 5.1
4.5
_
z 1
_ 2.9
15
x[n]
y[n]
+
_
z 1
_ 2.2
_
z 1
4.5
0.81
_1
_ 2.9 z _
5.1
z1
_4
_ 0.5
0.6
z1
0.3
z1
_1
1.5
x[n]
_
z 1
_
z 1
1.8
0.3
_4
_
z 1
_1
_ 0.5
1.5
1
z
_ 4.9
z1
3
1.5
_ 0.8
z1
4.2
z1
0.6
16
+
_
z 1
_
z 1
4.2
_1
_
z 1
z
1.5
_ 4.9
0.6
_ 0.8
z1
z1
0.475
0.75
_ 2.917
_ 0.25
1/12
x[n]
_ 1.9
8.27
_ 0.25
+
_
z 1
y[n]
_
z 1
_
z 1 _
2.917
1/12
0.75
p0 + p1z1 + p2 z2
1+ d1z1 + d2 z2
1 1 zM
.
(a) M-point moving average filter: H MA (z) =
M 1 z1
Choose a 0 = 1, p 0 = 1/ M, p1 = 0, p 2 = 0, d1= 1, d 2 = 0.
(b) First difference differentiator:
H FD (z) = 1+ z1 .
Choose a 0 = 0, p 0 = 1, p1 = 1, p 2 = 0, d1= 0, d 2 = 0.
17
.
1
1
Choose a 0 = 0, p 0 = 1, p1 = 0, p 2 = 0, d1= 1, d 2 = 0.
.
1
1
(1
)z
Choose a 0 = 0, p 0 = , p1 = 0, p 2 = 0, d1= (1 ), d 2 = 0.
1 z1
.
(f) DC blocker: H DC (z) =
1 z1
Choose a 0 = 0, p 0 = 1, p1 = 1, p 2 = 0, d1= , d 2 = 0.
0.5 + 0.5z1
.
(g) Trapezoidal integrator: HTI (z) =
1 z1
+ z1
.
(h) First-order allpass delay network: A1 (z) =
1+ z 1
+ z 1 + z2
(i) Second-order allpass delay network: A2 (z) =
.
1+ z 1 + z 2
Choose a 0 = 0, p 0 = , p1 = , p 2 = 1, d1= , d 2 = .
1
3
(j) Simpson integrator: H SI (z) =
4
3
1
3
+ z1 + z2
.
1 z2
Choose a 0 = 0, p 0 = 1/3, p1 = 4 /3, p 2 = 1/3, d1= 0, d 2 = 1.
1
0.6z1 0.2 + z1
2
0.4 +1.76z1 1.2z2
=
.
8.28 (a) H(z) =
1+ 0.25z1 1+ 0.3z1 1+ 0.25z1 1+ 0.8z1 + 0.2125z2 + 0.0187z3
(b)
y[n] = 0.4 x[n] +1.76x[n 1] 1.2x[n 2] 0.8y[n 1] 0.2125y[n 2] 0.0187y[n 3].
(c) A cascade realization of
is shown below:
18
0.2
z1
1
z
z1
_ 0.25
_ 0.6
_ 0.25
_ 0.3
_ 676.8
+
_ 0.3
_
z 1
548
_1
z
_ 0.25
129.2
+
+
_
z 1
_ 0.5
_
z 1
_ 0.0625
(e) The inverse z transform of the partial-fraction of H(z) given in Part (d) yields
h[n] = 548(0.25) n [n] +129.2(n +1)(0.25) n [n] 676.8(0.3) n [n].
8.29
A cascade
form given by
realization of based on the factored
1
1
is shown below:
H(z) =
1
2
1 0.4z1
1+ 0.8z + 0.5z
19
The above structure uses 4 unit delays, 6 multipliers, and 5 two-input adders, whereas the
structure of Figure 8.19 uses 3 unit delays, 6 multipliers, and 5 two-input adders.
8.30
(a) H1 (z) =
_ 0.0435
_1
z
_ 4.3452
0.1569
_1
z
_ 0.5
2.2866
_
z 1
0.3452
H1 (z) = 0.9 +
0.8217z1
+
1
1+ 4.3452z
II realization shown below:
0.0392z1
0.2725z1
+
which leads to the Parallel Form
1+ 0.5z1 1 0.3452z1
0.9
_
z 1
_ 4.3452
_
z 1
_ 0.8217
0.0392
_ 0.5
_
z 1
0.2725
0.3452
20
(b) H 2 (z) =
_1.6971
_1
z
_ 4.9
_
z 1
0.8049
1.5823
_1
z
_ 0.8
_
z 1
0.6951
_
z 1 0.8049
_1
z
_ 4.9
8,3159
_
z 1 0.6951
_1
z
_ 0.8
_ 1.2659
21
(c) H 3 (z) =
0.0921
0.3319
_1
z
0.75
4.9093
_1
0.2178
_
z 1
_ 2.2917
H 3 (z) = 5.3333 +
0.2489z1
+
1
0.2178z1 11.2506z2
1 0.75z
1+ 2.2917z2
5.3333
_
z 1
0.2489
0.75
_
z 1
_1
0.2178
z
_ 2.2917
8.31
_ 11.2506
1+ 3z1 1+ 2.1z1
1
.
whose partial-fraction expansion using
1
1
1+ z
1 0.5z 1 z1
0.7333 12.1333
12.4
+
. The inverse z-transform of
residuez is given by S(z) =
1
1
1+ z
1 0.5z
1 z1
S(z) then yields the step response:
8.32
It can be seen from Figure P8.14 that the top branch is the Direct Form II realization of
1+ 2.3z1
the transfer function
and the bottom branch is the Direct Form II t
1 1.2z1 0.9z2
realization of the same transfer function. Hence the transfer function of the overall
2 + 4.6z1
structure is given by H(z) =
which is a second-order function.
Thus a
1 1.2z1 0.9z2
canonic realization of H(z) is either the top branch or the bottom branch with a scaling
multiplier with a coefficient 2 at the input.
1 0.5z1
1
, X(z) =
1 0.3z
1 0.3z
1 0.3z
1 0.7z1
Y (z) (1 0.5z1 )(1 0.7z1 ) 1 1.2z1 + 0.35z2
H(z) =
=
=
.
X(z)
1 0.3z1
1 0.3z1
0.5z1
. Thus,
1.2
0.35
0.3
(d) A partial-fraction expansion of H(z) in z1 using residuez is given by
0.8889
H(z) = 0.1111z1 1.1667 +
whose realization yields the Parallel Form I
1 0.3z1
structure shown below.
23
_ 1.1667
0.8889
0.3
0.1111
1 0.4z
0.4z
=
. Therefore, Y (z) = H(z)X(z)
1 0.2z1 1 0.2z1 1 0.2z1
(1 0.5z1 )(1 0.7z1 ) 1 0.4z1
0.3z
1
0.2z
0.8889
7.5
+
residuez results in Y (z) = 5.6111z1 2.3333
an inverse
1
1 0.3z
1 0.2z1
-transform of which yields
y[n] = 2.333 [n] 5.6111 [n 1]. 0.8889(0.3) n [n] + 7.5(0.2) n [n].
(f) X(z) =
5z1
2z1 + 3z2
which is seen to
1 0.2z
1+ 0.6z1 0.4z2
be a Parallel Form II realization of the transfer function H(z). A partial-fraction
expansion of H(z) using residue yields
4z1
2z1 + 3z2
H(z) =
+
. Comparing this expansion with the transfer
1 0.2z1 1+ 0.6z1 + 0.4z2
function of the structure of Figure P8.15 we conclude that the multiplier coefficient of
value 5 in the top path is wrong and should be replaced with the value 4, the multiplier
coefficient of value 2 in the forward path of the bottom branch is wrong and should be
replaced with the value 2, and the multiplier coefficient of value 0.4 in the feedback path
of the bottom branch is wrong and should be replaced with the value 0.4 .
8.35
+
1
A + 3z1
which is seen to
1 Cz
1 0.9z
1 Bz 2
be a Parallel Form I realization of the transfer function H(z). A partial-fraction expansion
4
2 + 3z1
+
. Comparing this
of H(z) using residuez yields H(z) =
1+ 0.6z1 1 0.9z1 + 0.7z2
expansion with the transfer function of the structure of Figure P8.16 we conclude
A = 2, B = 0.7, C = 0.6.
+
1
24
8.36 (a) The difference equation corresponding to the transfer function H(z) =
is given by
1 1z1 + 2 z2
y[n] + 1 y[n 1] 2 y[n 2] = (1+ 1 + 2 )(x[n] + 2x[n
1] + x[n 2]), which can be
rewritten as y[n] = (x[n] + 2x[n 1] + x[n 2]) + 1 (y[n 1] x[n] 2x[n 1] x[n 2])
2 (y[n 2] x[n] 2x[n 1] x[n 2]). Denoting w[n] = x[n] + 2x[n 1] + x[n 2], the
difference equation representation becomes
y[n] = w[n] + 1 (y[n 1] w[n]) 2 (y[n 2] w[n]). A realization of H(z) based on the
Y (z)
X(z)
Finally, by delay sharing the above structure reduces to the canonic realization shown
below:
Y (z)
X(z)
(1 2 )(1 z2 )
is given by
1 1z1 + 2 z2
which can be rewritten as
y[n] 1 y[n 1] + 2 y[n 2] = (1 2 )(x[n] x[n 2])
y[n] = 1 y[n 1] 2 y[n 2] + x[n] 2 x[n] x[n 2] + 2 x[n 2]
25
A realization of H(z) based on the last two equations is as shown on next page.
Finally, by delay sharing the above structure reduces to a canonic realization as shown
below:
8.37
The 4 multiplier canonic realizations of H a (z) and Hb (z) are shown below:
26
X1
1
+
+
_1
z
3
Y1
X1
Y1
_
z 1
4
2
_1
+
_
z 1
(a)
(b)
Y1 C + D H N 1 (z)
=
,
8.38 (a) From the structure of Figure P8.174 it follows that H N (z) =
X1 A + B H N 1 (z)
C A H N (z)
. Substituting the expression for H N (z) we
from which we get H N 1 (z) =
B H N (z) D
N
C1+ di zi A pi
i=1
i=0
then arrive at H N 1 (z) =
N
N
B pi D1+ di zi
i=1
i=0
.
(Bp 0 D) + (Bp1 Dd1 )z1 + + (Bp N 1 DdN 1 )zN +1 + (Bp N DdN )zN
Substituting
the values A = 1,B = dN z1,C = p0 , and D = p N z1, we get H N 1 (z)
( p 0 d1 p1 ) + + ( p 0 d N 1 p N 1 )zN +2 + ( p 0 d N p N )zN +1
(dN p0 pN ) + (dN p1 pN d1 )z1 + + (dN pN 1 pN dN 1 )zN +1
p'0 + p1'z1 + + p'N 2 zN +2 + p'N 1zN +1
N +2
N +1
'
'
1+ d1'z1 + + dN
+ dN
2 z
1z
p d pN dk
, 1 k N 1.
and dk' = k N
p0 dN p N
+(dN pN pN dN)zN 1
p d
pk+1
, 0 k N 1,
where p'k = k k+1
p0 dN p N
C
(b) From the chain parameters, we obtain for the first stage t11= = p 0 ,
A
AD BC
1
B
t12 =
= ( pN p0 dN )z1, t 21= = 1, t 22 = = dN z1 . The corresponding
A
A
A
input-output relations of the two-pair are given by
Y1 = p 0 X1 + ( pN p 0 dN )z1 X 2 = p 0 (X1 dN z1 X 2 ) + pN z1 X 2 , Y2 = X1 dN z1 X 2 .
Substituting the second equation into the first we rewrite it as Y1 = p0Y2 + pN z1 X 2 . A
realization of the two-pair based on the last two equations is therefore as indicated below:
27
(c) Except for the first stage, all other stages require 2 multipliers. Hence, the total number
of multipliers needed to implement an N th order transfer function H N (z) is 2N +1. The
total number of 2-input adders required is 2N, while the overall realization is canonic
requiring N delays.
0.5 +1.3667z1 + z2
0.3252 + z1
.
G
(z)
=
.
get G2 (z) =
Repeating
the
procedure
we
get
1
1+ 5.2z1 + 22.5z2
1 2.08z1
From
G 3(z),G 2 (z), and G1(z), we arrive at the cascaded lattice realization shown below:
7.5
_ 22.5
2.08
_ 0.5
0.3252
_ 1.8
2 +11.4z2 + 7.84z4
0.5995 +1.1097z1 0.0383z2 + z3
3.3286 +1.3395z1 + z2
1.6514 + z1
. From
get G2 (z) =
and G1(z) =
1+ 3.7014z1 + 2.2982z2
1 0.0937z1
G 4 (z),G 3(z),G 2 (z), and G1(z), we arrive at the cascaded lattice realization shown below:
_ 7.84
3
_ 0.3
_ 0.5995
_ 2.2982
0.0937
3.3286
1.6514
28
1.2401+ z1
1.7188
_ 11.39
5.3333
0.0518
0.5197
_ 1.2401
_ 0.1583
29