chapter11 sm[1]
TRANSCRIPT
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 1/43
Chapter 11
11.1 Analysis yields{ } ],[][][,]1[]1[][ 3122211 nwnwnwnwn x nw α β
{ } ],[][][,]1[]1[][534422213
nwnwnwnwnwnw αα β
{ } ].[][][,]1[]1[][ 01534235 n x nwn ynwnwnw ααα β In matrix form the above equations can be written as
1 1 1 1
2 2
3 3 2 2 2 2
4 4
5 5 2 3 3 3
[ ] 0 0 0 0 0 0 [ ] 0 0 0 0 0
[ ] 1 0 1 0 0 0 [ ] 0 0 0 0 0 0
[ ] 0 0 0 0 0 0 [ ] 0 0 0 0
[ ] 0 0 1 0 1 0 [ ] 0 0 0 0 0 0
[ ] 0 0 0 0 0 0 [ ] 0 0 0 0
[ ] 1 0 0 0 0 0 [ ] 0 0 0 0 0
w n w n
w n w n
w n w n
w n w n
w n w n
y n y n
α β
α β α β
α β α β
⎡ ⎤ ⎡ ⎤⎡ ⎤⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥
= +⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎣ ⎦ ⎣ ⎦⎣ ⎦
11
2
3
4
5
0
[ 1][ 1]
0[ 1]
0[ 1]
0[ 1]
0[ 1]
[ ]0 [ 1]
x nw n
w n
w n
w n
w n
x n y n
β
α
−− ⎡ ⎤⎡ ⎤⎡ ⎤⎢ ⎢ ⎥⎢ ⎥−⎢ ⎢ ⎥⎢ ⎥⎢ ⎢ ⎥⎢ ⎥− + ⎢ ⎢ ⎥⎢ ⎥− ⎢ ⎢ ⎥⎢ ⎥⎢ ⎢ ⎥⎢ ⎥−⎢ ⎢ ⎥⎢ ⎥− ⎢ ⎣ ⎦⎣ ⎦ ⎣ ⎦
Here the F matrix is given by: .
000001000000010100000000000101 000000
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎣
⎡
=F
Since the F matrix contains nonzero entries above the main diagonal, the above set ofequations are not computable.
11.2 A computable set of equations are given by:{ },]1[]1[][ 2111 nwn x nw α β { },]1[]1[][ 422213 nwnwnw αα β
{ },]1[]1[][ 534235 nwnwnw αα β ],[][][ 312 nwnwnw ][][][ 01 n x nwn y α ,][][][ 534 nwnwnw .
In matrix form the above equations can be written as
1 1 1 1
3 3 2 2 2 2
5 5 2 3 3 3
2 2
4 4
[ ] [ ]0 0 0 0 0 0 0 0 0 0 0
[ ] [ ]0 0 0 0 0 0 0 0 0 0
[ ] [ ]0 0 0 0 0 0 0 0 0 0
[ ] [ ]1 0 1 0 0 0 0 0 0 0 0 0
[ ] [ ]1 0 0 0 0 0 0 0 0 0 0 0
[ ] [ ]0 0 1 0 1 0 0
w n w n
w n w n
w n w n
w n w n
y n y n
w n w n
α β
α β α β
α β α β
⎡ ⎤ ⎡ ⎤⎡ ⎤⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥
= +⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎣ ⎦⎣ ⎦ ⎣ ⎦
1 1
3
5
2
0
4
[ 1] [ 1]
[ 1] 0
[ 1] 0
[ 1] 0
[ 1] [ ]
[ 1]0 0 0 0 0 0
w n x n
w n
w n
w n
y n x n
w n
β
α
− −⎡ ⎤⎡ ⎤ ⎡⎢ ⎥
⎤⎢ ⎥ ⎢−
⎢ ⎥⎥
⎢ ⎥ ⎢⎢ ⎥
⎥⎢ ⎥ ⎢− +⎢ ⎥
⎥⎢ ⎥ ⎢−⎢ ⎥
⎥⎢ ⎥ ⎢
⎢ ⎥⎥
⎢ ⎥ ⎢−⎢ ⎥
⎥⎢ ⎥ ⎢−⎢ ⎥
⎥⎣ ⎦ ⎣⎣ ⎦ ⎦
Here the F matrix is given by: .
010100000001000101000000000000000000
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎣
⎡
=F
Not for sale 388
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 2/43
Since the F matrix has no nonzero entries above the main diagonal, the new set of equationsare computable.
11.3 Analysis yields { } ],[][][,]1[]1[]1[][ 1263422111 n x nwnwnwnwnwnw ααα β
{ } ],[][][,]1[]1[][ 324634223 nwnwnwnwnwnw αα β { },]1[][ 6335 nwnw α β
],[][][ 546 nwnwnw ].[][][][][ 0634221 n x nwnwnwn y αααα In matrix form the above set of equations are given by
1 1 1 1 1 2
2 2
3 3
4 4
5 5
6 6
1 2 3
0 0 0 0 0 0 0[ ] [ ] 0 0 0
1 0 0 0 0 0 0[ ] [ ]
0 0 0 0 0 0 0[ ] [ ]
0 1 1 0 0 0 0[ ] [ ]0 0 0 0 0 0 0[ ] [ ]
0 0 0 1 1 0 0[ ] [ ]
0 0 0 0[ ] [ ]
w n w n
w n w n
w n w n
w n w n
w n w n
w n w n
a a a y n y n
β α β α β ⎡ ⎤⎡ ⎤ ⎡ ⎤⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥= +⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥⎢ ⎥ ⎢ ⎥
⎣ ⎦ ⎣ ⎦⎣ ⎦
1 3 1
2
2 2 2 3 3
4
3 3 5
6
0
00 [ 1]
[ ]0 0 0 0 0 0 0 [ 1]
00 0 0 0 0 [ 1]
00 0 0 0 0 0 0 [ 1]00 0 0 0 0 0 [ 1]
00 0 0 0 0 0 0 [ 1]
[ ]0 0 0 0 0 0 0 [ 1]
w n
x nw n
w n
w n
w n
w n
x n y n
α
β α β α
β α
α
− ⎡ ⎤⎡ ⎤⎡⎢ ⎥⎢ ⎥⎢ −⎢ ⎥⎢ ⎥⎢⎢ ⎥⎢ ⎥⎢ −⎢ ⎥⎢ ⎥⎢ +− ⎢ ⎥⎢ ⎥⎢⎢ ⎥⎢ ⎥⎢ −⎢ ⎥⎢ ⎥⎢ − ⎢ ⎥⎢ ⎥⎢⎢ ⎥⎢ ⎥⎢ −⎣ ⎦⎣ ⎣ ⎦
⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦
Here the F matrix is given by: .
0000001100000000000000110000000000000010000000
321 ⎥
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢
⎣
⎡
=
ααα
F
Since the diagonal of the F matrix has all zeros, and no nonzero entries above the maindiagonal, the new set of equations are computable.
11.4
x [n] y[n]w [n]1 w [n]2 w [n]3 w [n]4
w [n]5
α 0
1β_ 1 z
1α 11β
_ 1 z
1
1 1
α 12β_ 1 z
α 22β_ 1 z
α 23β_ 1 z
α 33β_ 1 z
1 Reduced signal flow graph obtained by removing the branches going out of the input nodeand the delay branches is:
y[n]w [n]1w [n]2 w [n]3 w [n]4
w [n]5
1 1
1 1
1
Not for sale 389
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 3/43
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 4/43
.
⎥
⎥⎥⎥⎥⎥
⎦
⎤
⎢
⎢⎢⎢⎢⎢
⎣
⎡
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎣
⎡
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎣
⎡
=
⎥
⎥⎥⎥⎥⎥
⎦
⎤
⎢
⎢⎢⎢⎢⎢
⎣
⎡
][
][][][][][
00000
0000000000000000000100000010
0
00
][00
][
][][][][][
32321
3
21
32321
n y
nsnsnwnwnw
k
k k
n x
n y
nsnsnwnwnw
4 4 4 4 34 4 4 4 21F
⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢
⎣
⎡
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎣
⎡
]1[]1[]1[]1[]1[]1[
010000001000000001000000000000000
32321
32
1
n ynsnsnwnwnw
k k
k
4 4 4 4 4 34 4 4 4 4 21G
As the diagonal elements of F -matrix are all zeros, there are no delay-free loops. However,
the above set of equations are not computable as there are non-zero elements above thediagonal of F .
(b) The reduced signal flow-graph representation of Figure P11.3 is shown below:
From the above flow-graph we observe that the set composed of nodes with only outgoing branches is The set of nodes with incoming branches from and
outgoing branches is The set of nodes with incoming branches fromand and outgoing branches is
]}.[{ 31 nwN 1N
]}.[{ 22 nwN 1N
2N ]}.[{ 13 nwN Finally, the set of nodes with only
incoming branches from , and , and no outgoing branches
Therefore, one possible ordered set of equations that is
computable is given by
1N 2N 3N
]}.[],[],[{ 324 n ynsnsN
⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢
⎣
⎡
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎣
⎡
=
⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢
⎣
⎡
][][][][][][
000000000000000000010000001000000
][][][][][][
12123
32
1
12123
n ynsnsnwnwnw
k k
k
n ynsnsnwnwnw
4 4 4 4 34 4 4 4 21F
.
⎥⎥⎥⎥
⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎢⎢
⎣
⎡
⎥
⎥⎥⎥
⎥
⎦
⎤
⎢
⎢⎢⎢
⎢
⎣
⎡
]1[]1[]1[]1[
]1[]1[
01000000100000010000000
0000000000
121
23
32
3
n ynsnsnw
nwnw
k
k k
4 4 4 4 4 34 4 4 4 4 21G
Note: All elements on the diagonal and above diagonal of F are zeros.
Not for sale 391
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 5/43
11.7 .5.131
)(21
22
110
= z z
z p z p p z H From Eq. (11.18) we get,
Hence,.
0.5
0.4
2.3
5.1
3
1
2.36.50.7
02.36.5
002.3
2
1
0
⎥
⎥⎥
⎦
⎤
⎢
⎢⎢
⎣
⎡
⎥
⎥⎥
⎦
⎤
⎢
⎢⎢
⎣
⎡
⎥
⎥⎥
⎦
⎤
⎢
⎢⎢
⎣
⎡
=
⎥
⎥⎥
⎦
⎤
⎢
⎢⎢
⎣
⎡
p
p
p
.
5.131
542.3)(
21
21
=
z z
z z z H
11.8 From Eqn (11.15),
0
1
12
2
2 0 0
12 2 0
4 2 2
0 8 4 2
0 12 8 4
p
p
d p
d
⎡ ⎤ ⎡ ⎤⎢ ⎥ ⎢ ⎥− ⎡ ⎤⎢ ⎥ ⎢ ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥= −
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢ ⎥− ⎣ ⎦⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥−⎣ ⎦ ⎣ ⎦
and from Eqn (11.20),
. Finally, from Eqn (11.21),
Hence,
.
5.3
25.0
12
8
48
24 1
2
1⎥
⎦
⎤⎢
⎣
⎡ =
⎦
⎤⎢
⎣
⎡
⎦
⎤⎢
⎣
⎡
⎦
⎤⎢
⎣
⎡
d
d
.
5.11
5.2
2
5.3
25.0
1
224
022
002
2
1
0
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
p
p
p
.5.325.01
5.115.22)(
21
21
= z z
z z z H
11.9 From Eqn. (11.15)
0
1
2
13
2
3
2 0 0 0
4 2 0 0
18 4 2 0
8 8 4 2
0 12 8 8 40 16 12 8 8
0 8 16 12 8
p
p
p
d p
d d
⎡ ⎤ ⎡ ⎤⎢ ⎥ ⎢ ⎥− ⎡ ⎤⎢ ⎥ ⎢ ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥−
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢ ⎥= − −⎢ ⎥ ⎢ ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥− − ⎢ ⎥⎢ ⎥ ⎢ ⎥⎢ ⎥−⎢ ⎥ ⎢ ⎥⎣ ⎦
⎢ ⎥ ⎢ ⎥− −⎣ ⎦ ⎣ ⎦
. Next, from Eqn. (11.20),
Finally, from Eqn. (11.21),
Hence,
.
4.7
8.4
4.0
8
16
12
81216
8812
488 1
3
2
1
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
=⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
d
d
d
.
4.02.3
2.3
0.2
4.78.4
4.0
1
24880248
0024
0002
3
2
1
0
⎥⎥
⎥⎥
⎦
⎤
⎢⎢
⎢⎢
⎣
⎡
=⎥⎥
⎥⎥
⎦
⎤
⎢⎢
⎢⎢
⎣
⎡
⎥⎥
⎥⎥
⎦
⎤
⎢⎢
⎢⎢
⎣
⎡
=⎥⎥
⎥⎥
⎦
⎤
⎢⎢
⎢⎢
⎣
⎡
p p
p
p
.4.78.44.01
4.02.32.32)(
321
321
= z z z
z z z z H
Not for sale 392
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 6/43
11.10 . Hence,
0
1
2
3
2 0 0 0 1 2
4 2 0 0 0.6 5.2
4 4 2 0 0.2 6.8
6 6 4 2 1.8 5.6
p
p
p
p
⎡ ⎤ ⎡ ⎤ ⎡ ⎤⎢ ⎥ ⎢ ⎥ ⎢ ⎥− −⎢ ⎥ ⎢ ⎥ ⎢ ⎥= =⎢ ⎥ ⎢ ⎥ ⎢ ⎥−⎢ ⎥ ⎢ ⎥ ⎢ ⎥− − − −⎣ ⎦ ⎣ ⎦⎣ ⎦
⎡ ⎤⎢ ⎥−⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦
( ) 1 22 5.2 6.8 5.6 P z z z z − −= − + − 3− .
11.11
0
2 23 31
232
233
5 13 34
2 0 0 0 0 1 2
2 2 0 0 0
4 2 2 0 0 0 2
6 4 2 2 0 2 4
8 6 4 2 2 3
p
p
p
p
p
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦
⎢ ⎥ ⎢ ⎥ ⎢ ⎥− −⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥= =−⎢ ⎥ ⎢ ⎥ ⎢ ⎥− −⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥− − −⎣ ⎦ ⎣ ⎦⎣ ⎦
. { } { }2 2 2 13 3 3 32, , 2 ,4 , 3i p = − −
11.12 The –th sample of an –point DFT is given by Thus, the
computation of requires
k N .][][ 1
0
=
N n
nk N W n x k X
][k X N complex multiplications and 1 N complex additions. Now, each complex multiplication, in turn, requires 4 real multiplications and 2 realadditions. Likewise, each complex addition requires 2 real additions. As a result, thecomplex multiplications needed to compute require a total of real multiplicationsand a total of real additions. Therefore, each sample of the –point DFT involves
real multiplications and
N ][k X N 4
22 N N N 4 24 N real additions. Hence, the computation of all DFT
samples thus requires real multiplications and24 N N N )24( real additions.
11.13 Let the two complex numbers be b ja α and .d jc β Thus, ))(( d jcb ja αβ which requires 4 real multiplications and 2 real additions. Consider
the product and which require 3 real multiplications and 2 realadditions. The imaginary part of can be formed from
),()( bcad jbd ac
,),)(( acd cba ,bd αβ bd acd cba ))(( ,bcad which now requires 2 real additions. Likewise, the real part of can be formed from
requiring an additional real additions. Hence, the complex multiplication can becomputed using 3 real multiplications and 5 real additions.
αβ bd ac
11.14 Recall, , / 2 s jceW N j N π where ) / 2cos( N c π and ). / 2sin( N s π Thus,
Now,.122 =sc ( )}Im{}Re{)( r r r N r j s jcW Ψ+Ψ+=Ψ⋅=Ψ +1
.}Re{}Im{}Im{}Re{ r r r r sc jsc ΨΨΨΨ Thus, }Im{}Re{}Re{ 1 r r r sc ΨΨΨ ⋅ and }.Re{}Im{}Im{ 1 r r r sc ΨΨΨ ⋅
Figure P11.4 with internal node label is shown below. Its analysis yields
c 1s
_____ c 1s
_____
s
Re{ ψ }r
Im{ ψ }r
Re{ ψ }r +1
Im{ ψ }r +1
U
Not for sale 393
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 7/43
(1): },Im{1
}Re{ r r sc
U ΨΨ
(2): ,}Im{}Im{ 1 sU r r ΨΨ and
(3): }.Im{1
}Re{ 11 r r sc
U ΨΨ Substituting Eqs. (1) in Eq. (2) we get
(4): }.Re{}Im{}Re{}Im{1
}Im{}Im{ 1 r r r r r r scsc
s ΨΨΨΨΨΨ ⋅⎠
⎞
⎜⎝
⎛
Next,substituting Eqs. (1) and (4) in Eq. (3) we get (5):
(5): }Re{}Im{1
}Re{}Im{1
}Re{ 1 r r r r r sc
scs
cΨΨΨΨΨ =
}.Re{Im}Re{}Re{}Im{}Re{}Im{1 22
r r r r r r csccss
cs
cΨΨΨΨΨΨ ⋅==
It thus follows that the structure of Figure P11.4 implements the multiplication of a complexsignal with the twiddle factor using only 3 real multiplications.
.
In the case of multiplication by complex twiddle factor we have, jscW N
−=−1
( )}Im{}Re{)( r r r N r j s jcW Ψ+Ψ−=Ψ⋅=Ψ −+
11
( ) ( ).}Re{}Im{}Im{}Re{ r r r r sc j sc Ψ⋅−Ψ⋅+Ψ⋅+Ψ⋅= Thus, here}Im{}Re{}Re{ r r r sc Ψ⋅+Ψ⋅=Ψ +1 and }.Re{}Im{}Im{ r r r sc Ψ⋅−Ψ⋅=Ψ +1 It follows
then that the real and imaginary parts are simply obtained by reversing the sign of of thereal and imaginary parts derived in the case of multiplication by As a result, thecorresponding structure is obtained by cascading an inverter to the real multipliers in FigureP11.4 as indicated below:
s. N W
c 1s
_____ c 1s
_____s
Re{ ψ }r
Im{ ψ }r
Re{ ψ }r +1
Im{ ψ }r +1
U _1 _
1_1
11.15 The center frequency bin ,)(: N kF
k f k T c = where # N of bins, and T F is the sampling
frequency. Inverting we have .)( ⎥⎦
⎥⎢⎣
⎢=T F
N f f k Therefore, the absolute difference from one of
the given four tones (150 Hz, 375 Hz, 620 Hz, and 850 Hz) to the center of its bin is given by
.),( ⎥⎦⎥⎢
⎣⎢
T
T F N f
N F f f N dist It follows from this equation that the distance goes to zero if
f F N f
N F
T
T =⎦
⎥⎢⎣
⎢ or
T F N f
is an integer.
Not for sale 394
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 8/43
The total distance is reduced to zero ifT
i
F
N f is an integer for .4,,1 Ki The minimum value
of for which this true is 500. N However, the total distance can be small, but nonzero, for significantly smaller values of . N
11.16 .1
1)(
1 zW z H
k N
k Hence, ),()(1
1
1
)()( 2 /
1
2 /
1 zV z zV
zW
z
zW
z X zY N
k N
N
k N
=
where .1
1)(
1 zW zV
k N
Or, in other words, ].[][][2
N nvnvn y
Consider Then:1k .1
1)(
11 zW zV
N
This implies,
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
=
LL ,,,,,,1][][)1
2(
2 / 21 N
N N
N N N n
N W W W W nW nv µ
{ },,,1,,,,1 121 LL N N N W W W since ,,1 1
)12
(2 /
N
N
N N
N W W W and
son on. Thus, { }.,,,1,,0,0,0][ 212
LL = N N N
W W nv Hence,
.,0,0,0,,,,,1][][][)1
2(
212
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
=
LL
N
N N N N
W W W nvnvn y
Now, consider :2
N k = .
1
1)(
12 / zW
zV N
N
This implies,
.,,,,,,1][][ 2)1
2(
222)2 / (
⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
=⋅
LL
N N
N
N N
N N
N
N
N n N
N W W W W nW nv µ
Thus, { }.,,,1,0,,0,0,0][ 2 / 2
LL N N
N N
N W W nv =
Now, ),1()1(,1,1,)1( 2 / 2)1
2(
22 / 22 N N N
N N
N
N
N N
N N
N W W W W etc. Hence,
.,,)1(1,)1(1,,1,1,1,1][1
1
2 /
2 /
2
2⎪⎭
⎪
⎬
⎫
⎪⎩
⎪
⎨
⎧
= LLL N
N
n N n
N n y Therefore, if
is even,
2 / N
,,2,2,2,,1,1,1,1][12 /
2⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
LL N
n N nn y and if is odd,2 / N
Not for sale 395
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 9/43
,,0,0,0,,1,1,1,1][12 /
2⎪⎭
⎪⎬
⎫
⎪⎩
⎪⎨
⎧
LL N
n N nn y
11.17
11.18
11.19 ∑=
= 1
0
][][ N
n
nk N W n x k X
∑
=
=
=
1
0
)1(11
1
01
1
01
111
11
11 ]1[]1[][
r N
n
k r nkr N
r N
n
k nkr N
r N
n
nkr N W r nr x W nr x W nr x K
Not for sale 396
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 10/43
∑ ∑−
=−
−
=⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜
⎝
⎛
+=1
0
1
01
1
1
11
r
i
ki N
r N
r N
n
nkr N W W inr x
DFT point)/(
][ .
Thus, if the -point DFT has been calculated, we need at the first stage an additionalmultiplications to compute one sample of the –point DFT and, as a result,
additional multiplications are required to compute all
) / ( 1r N )1( 1 r N ][k X
N r )1( 1 N samples of the N –pointDFT. Decomposing it further, it follows that additional N r )1( 2 multiplications are neededat the second stage, and so on. Therefore, the total number of multiply (add) operations
.)1()1()1(121 N r N r N r N r
i i ⎟⎠
⎞⎜⎝
⎛ ∑ = νν
νK
11.20 An examination of the flow-graph of the 8-point DIT FFT algorithm shown in Figure 11.24
reveals that in the first stage all twiddle factors are and in the second stage thetwiddle factors are either or Hence, there are no twiddle factors with
nonunity magnitudes in the first two stages. In all succeeding stages, one of the twiddle
factors is and another one is The number of twiddle factors with
nonzero magnitudes in the -th stage, is Hence, the total number of twiddlefactors with nonzero magnitudes in an N -point radix-2 FFT algorithm is
where
10 = N W
10 = N W ./ jW N N =4
10 = N W ./ jW N N =4
i ,3i .22 1 −−i
,),()( 3222R3
1 ≥ν−ν−⎟⎟⎟
⎠
⎞
⎜⎜⎜
⎝
⎛ =ν ∑
ν
=
−
i
i .2ν= N
11.21 Direct computation of M samples of an N -point DFT requires multiplications,
whereas, the Radix-2 FFT algorithm requires
2 M
N N
22log multiplications. In order for a N -
point radix-2 FFT algorithm to be computationally more efficient than a direct computation
of M samples of an N -point DFT, the following inequality must hold: .log 22 ⎥⎦
⎥⎢⎣
⎢> N M N
a) N = 512 , M = 49 , b) N = 1024 , M = 72 , c) N = 2048 , M = 107
11.22 ).()()()( 32
231
130 z X z z X z z X z X Thus, the –point DFT can be expressed as
Hence, the structural
interpretation of the first stage of the radix-3 DFT is as indicated below:
N
].[][][][ 3 / 22
3 / 13 / 0 N k
N N k
N N k X W k X W k X k X
Not for sale 397
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 11/43
3
3
3
z
z
N _3
-pointDFT
N _3
-pointDFT
N _3
-pointDFT
+ X [< k > ]1 N /3
+ x [n] x [n]0
x [n]1
x [n]22k
N W
k N W
X [< k > ]2 N /3
X [< k > ]0 N /3 X [k ]
11.23 k k k
n
nk W x W x W x W n x k X 69
39
09
8
09 ]6[]3[]0[][][ ⋅
=∑
k k k k k k W x W x W x W x W x W x 89
59
29
79
499 ]8[]5[]2[]7[]4[]1[
k k k k k k k W W x W x W x W x W x W x 969
39
09
69
39
09 ]7[]4[]1[]6[]3[]0[ ⋅
k k k k W W x W x W x 29
69
39
09 ]8[]5[]2[ ⋅ ,][][][ 2
93293130k k W k GW k Gk G where
,]6[]3[]0[][ 2
330
330k k k
W x W x W x k G ⋅
,]7[]4[]1[][ 2
330
331k k k
W x W x W x k G ⋅
are 3-point DFTs. A flow-graph representation
of this radix-3 DIT FFT computation scheme is shown below,
,]8[]5[]2[][ 233
0332
k k k W x W x W x k G ⋅
where the twiddle factors for computing the DFT samples are indicated below for a typicalDFT sample:
Not for sale 398
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 12/43
In the above diagram, the 3-point DFT computation is carried out as indicated below:
11.24 k k k k k
n
nk W x W x W x W x W x W n x k X 1215
915
615
315
015
14
015 ]12[]9[]6[]3[]0[][][ ⋅
=∑
k k k k k W x W x W x W x W x 1315
1015
715
41515 ]13[]10[]7[]4[]1[
k k k k k W x W x W x W x W x 14151115815515215 ]14[]11[]8[]5[]2[ where,][][][ 2
1552155150k k W k GW k Gk G
,]12[]9[]6[]3[]0[][ 45
35
255
0550
k k k k k W x W x W x W x W x k G ⋅
,]13[]10[]7[]4[]1[][ 45
35
255
0551
k k k k k W x W x W x W x W x k G ⋅ and
.]14[]11[]8[]5[]2[][ 45
35
255
0552
k k k k k W x W x W x W x W x k G ⋅
A flow-graph representation of this mixed-radix DIT FFT computation scheme is shown below:
Not for sale 399
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 13/43
11.25
11.26 Now, by definition, ]}.[Re{]}[Im{][ n X jn X nq Its -point DFT is given by N
Thus,.][][1
0∑=
= N
n
nk N W nqk Q
Not for sale 400
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 14/43
(1): ∑=
⎥⎦
⎤⎢⎣
⎡⎟⎠
⎞⎜⎝
⎛ ⎠
⎞⎜⎝
⎛=1
0
2sin][Re{
2cos][Im{]}[Re{
N
m N mk
n X N
mk n X k Q
π π ,
(2): ∑=
⎥⎦
⎤⎢⎣
⎡⎟⎠
⎞⎜⎝
⎛ ⎠
⎞⎜⎝
⎛ 1
0
2cos][Re{
2sin][Im{]}[Im{
N
m N
mk n X
N mk
n X k Q π π
.
From the definition of the inverse DFT we observe .][1
][1
0∑
=
= N
m
mk N W m X
N k x Hence,
(3): ,2
sin]}[Im{2
cos]}[Re{1
]}[Re{1
0∑
=⎥⎦
⎤⎢⎣
⎡⎟⎠
⎞⎜⎝
⎛ ⎠
⎞⎜⎝
⎛= N
m N mk
m X N
mk m X
N k x
π π
(4): .2
sin]}[Re{2
cos]}[Im{1
]}[Im{1
0∑
=⎥⎦
⎤⎢⎣
⎡⎟⎠
⎞⎜⎝
⎛ ⎠
⎞⎜⎝
⎛= N
m N mk
m X N
mk m X
N k x
π π
Comparing Eqs. (2) and (3) we get ,]}[Im{1
]}[Re{nk
k Q N
n x =
and comparing Eqs. (1)
and (4) we get .]}[Re{1
]}[Im{ nk k Q N
n x =
11.27 Therefore,
Thus,
⎩ ≠==
.0if ],[,0if ],0[][][
nn N X n X n X nr N
∑
=
=
=
1
1
1
1
1
0
][]0[][]0[][][ N
n
nk N
N
n
nk N
N
n
nk N W n N X X W nr r W nr k R
].[][][]0[][]0[1
0
1
1
1
1
)( k x N W n X W n X X W n X X N
n
nk N
N
n
nk N
N
n
k n N
N ⋅
∑
=
=
=
.][][ 1nk N
k Rn x =
11.28 Let denote the result of convolving a length- L sequence x [n] with a length- N sequence h[n]. The length of y[n] is then L + N – 1. Here L = 16 and N = 9, hence length of y[n] is 24.
][n y
Method #1: Direct linear convolution - For a length- L sequence x [n] and a length- N sequenceh[n],
# of real mult. = .135)1916(92)1(2
9
11 =∑=n
N
n n N L N n
Method # 2: Linear convolution via circular convolution - Since y[n] is of length 24, to getthe correct result we need to pad both sequences with zeros to increase their lengths to 24
before carrying out the circular convolution.# of real mult. = 24 ×24 = 576.
Not for sale 401
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 15/43
Method #3: Linear convolution via radix-2 FFT - The process involves computing the 16- point FFT G of the length-16 complex sequence][k ],[][][ nh jn x ng e where is
obtained by zero-padding h[n] to length 16. Then, the 16-point DFTs, and of
][nhe
][k X ],[k H e][n x and , respectively, are recovered from . Finally, the IDFT of the product
yields y[n].
][nhe ][k G
][][][ k H k X k Y e
Now, the first stage of the 16-point radix-2 FFT requires 0 complex multiplications, thesecond stage requires 0 complex multiplications, the third stage requires 4 complexmultiplications, and the last stage requires 6 complex multiplications, resulting in a total of10 complex multiplications .# of complex mult. to implement = 10][k G# of complex mult. to recover and from = 0][k X ][k H e ][k G# of complex mult. to form = 16][k Y # of complex mult. to form the IDFT of = 10][k Y Hence, the total number of complex mult. = 36A direct implementation of a complex multiplication requires 4 real multiplications resultingin a total of 4 × 36 = 144 real multiplications for Method #3. However, if a complex multiplycan be implemented using 3 real multiplies (see Problem 11.13 ), in which case Method #3requires a total of 3 × 36 = 108 real multiplications.
11.29 Let y[n] denote the result of convolving a length- L sequence x [n] with a length- N sequenceh[n]. The length of y[n] is then L + N – 1. Here, L = 16 and N = 10, hence length of y[n] is25.
Method #1: Direct linear convolution - For a length- L sequence x [n] and a length- N sequenceh[n],
# of real mult. = .160)11016(102)1(210
11
=∑=n
N
n
n N L N n
Method # 2: Linear convolution via circular convolution - Since y[n] is of length 24, to getthe correct result we need to pad both sequences with zeros to increase their lengths to 24
before carrying out the circular convolution.# of real mult. = 25 ×25 = 625.
Method #3: Linear convolution via radix-2 FFT - The process involves computing the 16-
point FFT G of the length-16 complex sequence][k ],[][][ nh jn x ng e where isobtained by zero-padding h[n] to length 16. . Then, the 16-point DFTs, and of
][nhe][k X ],[k H e
][n x and , respectively, are recovered from . Finally, the IDFT of the product
yields y[n].
][nhe ][k G
][][][ k H k X k Y e
Now, the first stage of the 16-point radix-2 FFT requires 0 complex multiplications, thesecond stage requires 0 complex multiplications, the third stage requires 4 complex
Not for sale 402
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 16/43
multiplications, and the last stage requires 6 complex multiplications resulting in a total of 10complex multiplications.# of complex mult. to implement = 10][k G# of complex mult. to recover and from = 0][k X ][k H e ][k G# of complex mult. to form = 16][k Y # of complex mult. to form the IDFT of = 10][k Y
Hence, the total number of complex mult. = 36
A direct implementation of a complex multiplication requires 4 real multiplications resultingin a total of 4 × 36 = 144 real multiplications for Method #3. However, if a complex multiplycan be implemented using 3 real multiplies (see Problem 11.13 ), in which case Method #3requires a total of 3 × 36 = 108 real multiplications.
11.30 (a) Since the impulse response of the filter is of length 72, the transform length N should begreater than 72. If L denotes the number of input samples used for convolution, then L = N – 71. So for every L samples of the input sequence, an N -point DFT is computed andmultiplied with an N -point DFT of the impulse response sequence h[n] (which needs to becomputed only once), and finally an N -point inverse of the product sequence is evaluated.Hence, the total number M R of complex multiplications required (assuming N is a power-
of-2) is given by .log)log( 222712048 N N N N
N
N M ⎥⎤
⎢⎡=
R
It should be noted that in developing the above expression, multiplications due to twiddlefactors of values ±1 and ± j have not been excluded. The values of M R for different valuesof N are as follows:
1) For N = 128, M R = 37312, 2) For N = 256, M R = 28672, 3) For N = 512, M R =27904, 4) For N = 1024, M R = 38912.
Hence, N = 512 is the appropriate choice for the transform length requiring 27904 complexmultiplications or equivalently, 27904 × 3 = 83712 real multiplications.Since the first stage of the FFT calculation process requires only multiplications by ±1, thetotal number of complex multiplications for N = 128 is actually
,27648log)log(222271
2048 =⎥⎤
⎢⎡=
N N
N M N N N N R or equivalently, 27648 × 3 = 82944
real multiplications.
(b) For direct convolution, # of real multiplications =
147456)1722048(722)1(272
11
=∑=n
N
n
n N L N n .
11.31 (a)
Not for sale 403
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 17/43
,
0001000
0000100
00000100000001
0001000
0000100
0000010
0000001
78
68
58
48
38
28
18
08
8
⎥⎥⎥⎥⎥⎥
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢
⎢⎢⎢⎢⎢
⎣
⎡
=
W
W
W W
W
W
W
W
V ,
0100000
0010000
01000000010000
0000010
0000001
0000010
0000001
28
08
28
08
28
08
28
08
4
⎥⎥⎥⎥⎥⎥
⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢
⎢⎢⎢⎢⎢
⎣
⎡
=
W
W
W W
W
W
W
W
V
,
1000000
1000000
00100000010000
0000100
0000100
0000001
0000001
48
08
48
08
48
08
48
08
2
⎥⎥⎥⎥⎥
⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎢⎢⎢⎢⎢⎢
⎣
⎡
=
W
W
W W
W
W
W
W
V .
1000000000001000 001000000000001001000000000001000001000000000001
⎥⎥⎥
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎢⎢⎢⎢
⎣
⎡
=E
As can be seen from the above, multiplication by each matrix ,3,2,1, =k k V requires at most8 complex multiplications.
(b) The transpose of the matrices given in Part (a) are as follows:
,
000000
000000
000000
00000010001000 01000100
0010001000010001
38
38
28
28
18
18
08
088
⎥⎥⎥⎥⎥⎥⎥⎥
⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢⎢
⎢
⎣
⎡
=
W W
W W
W W
W W t V ,
000000
0000001010000001010000000000000000
0000101000000101
28
28
08
08
28
28
0808
4
⎥⎥⎥⎥⎥⎥⎥⎥
⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢⎢
⎢
⎣
⎡
=
W W
W W
W W W W
t V
,
00000011000000000000 00110000
0000000000110000000000000011
48
08
48
08
48
08
48
08
2
⎥⎥⎥⎥⎥
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢
⎢⎢⎢⎢
⎣
⎡
=
W W
W W
W W
W W
t V .
1000000000001000001000000000001001000000000001000001000000000001
⎥⎥⎥
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎢⎢⎢⎢
⎣
⎡
=EE t
It is easy to show that the flow-graph representation of is precisely the 8-
point DIF FFT algorithm of Figure 11.28 .
t t t t 8428 VVVED =
Not for sale 404
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 18/43
11.32 .10,][][][]2[2
1
2 /
2
12
0
21
0
2 ∑
=
=
=
N N
N n
n N
N
n
n N
N
n
n N W n x W n x W n x X ll lll Replacing n by
2 N
n in the right-most sum we get
,][][][][]2[
12
02 / 2
12
0
22
12
0
2 ∑
=
=
=⎟⎠
⎞⎜⎝
⎛
N
n
n N
N
N
n
n N
N
N
n
n N W n x n x W n x W n x X llll .10
2 N
l
Likewise,
,][][][][]14[1
43
)14(
14
3
2
)14(
12
4
)14(
14
0
)14( ∑
=
=
=
=
N
N n
n N
N
N n
n N
N
N n
n N
N
n
n N W n x W n x W n x W n x X lllll
where .104
N l Replacing byn
4 N
n in the second sum, byn2
N n in the third
sum, and byn4
3 N n in the fourth sum, we get
4 /
14
0
4
14
0
4 ]4
[][]14[ N N
N N
N
n
n N
n N
N
n
n N
n N W W W W
N n x W W n x X llll ∑
=
=
.][][ 4 / 33
14
0
44
32 / 2
14
0
42
N N
N N
N
n
n N
n N
N N N
N N
N
n
n N
n N
N W W W W n x W W W W n x llll ∑
=
=
Now, and Therefore,,,1 4 / 32 jW W W W N N
N N
N N
N N lll ,12 / N
N W .4 / 3 jW N N
.10,][][][][]14[44 /
14
04
342
⎭⎩⎟⎠
⎞⎜⎝
⎛ ⎠
⎞⎜⎝
⎛ ∑
=
N n N
n N
N
n
N N N W W n x n x jn x n x X ll l
Similarly, 4 / )34(
14
0
)34(4
14
0
)34( ][][]34[ N N
N
n
n N
N
N
n
n N W W n x W n x X
=
=
∑ llll
4 / 3)34(14
0
)34(4
32 / )34(14
0
)34(2
][][ N N
N
n
n N
N N N
N
n
n N
N W W n x W W n x
=
=
∑ llll
4 / 3
14
0
344
14
0
34 ][][ N N
N N
N
n
n N
n N
N
N
n
n N
n N W W W W n x W W n x lll ∑
=
=
Not for sale 405
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 19/43
4 / 93
14
0
344
4 / 62
14
0
342
][][ N N
N N
N
n
n N
n N
N N N
N N
N
n
n N
n N
N W W W W n x W W W W n x llll ∑
=
=
.10,][][][][44 /
3
1
4
04
342
⎭⎩⎟⎠⎞⎜
⎝⎛
⎠⎞⎜
⎝⎛ ∑
=
N n N
n N
N
n
N N N W W n x n x jn x n x ll
The butterfly here is as shown below which is seen to require two complex multiplications.
11.33 From the flow-graph of the 8-point split-radix FFT algorithm given below it can be seenthat the total number of complex multiplications required is 2. On the other hand, the totalnumber of complex multiplications required for a standard DIF FFT algorithm is also 2.
11.34 If multiplications by are ignored, the flow-graph shown below requires 8 complexmultiplications = 24 real multiplications. A radix-2 DIF 16-point FFT algorithm, on theother hand requires 10 complex multiplications = 30 real multiplications.
1, ± j
Not for sale 406
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 20/43
11.35 (a) N = 12. Choose and41 = N .32 = N Thus, and
The corresponding index mappings are indicated below:
⎩⎨⎧
≤≤≤≤+=
,,,
20304
21
21 nnnnn
⎩
⎨⎧
≤≤
≤≤+=.
,,20
3032
121
k
k k k k
0 1 2 3n1n2 x [0] x [1] x [2] x [3] x [4] x [5] x [6] x [7] x [8] x [9] x [10] x [11]
012
0 1 2 3k 1k 2 X [0] X [3] X [6] X [1] X [4] X [7] X [10] X [2] X [5] X [8] X [11]
012
X [9]
(b) N = 15. Choose and31 = N .52 = N Thus, and
The corresponding index mappings are indicated below:
⎩⎨⎧
≤≤≤≤+=
,,,
40203
21
21 n
nnnn
⎩⎨⎧
≤≤≤≤+=
.,,
40205
21
21 k
k k k k
0 1 2n1n2 x [0] x [1] x [2]
x [3] x [4] x [6]
x [5] x [7] x [8]
x [9] x [10] x [11]
0123
4 x [12] x [13] x [14]
0 1 2k 1k 2 X [0] X [5] X [10]
X [1] X [6] X [2]
X [11] X [7] X [12]
X [3] X [8] X [13]
0123
4 X [4] X [9] X [14]
Not for sale 407
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 21/43
(c) N = 21. Choose and71 = N .32 = N Thus, and
The corresponding index mappings are indicated below:
⎩⎨⎧
≤≤≤≤+=
,,,
20607
21
21 nnnnn
⎩⎨⎧
≤≤≤≤+=
.,,
20603
21
21 k k k k k
0 1 2 3n1n2 x [0] x [1] x [2] x [3] x [4] x [5] x [6]
x [7] x [8] x [9] x [10] x [11]012
4 5 6
x [12] x [13] x [14] x [15] x [16] x [17] x [18] x [19] x [20]
0 1 2 3k 1k 2 X [0] X [3] X [6] X [9] X [12] X [15] X [18]
X [1] X [4] X [7] X [10] X [13]012
4 5 6
X [16] X [19] X [2] X [5] X [8] X [11] X [14] X [17] X [20]
(d) N = 35. Choose and71 = N .52 = N Thus, and
The corresponding index mappings are indicated below:
⎩⎨⎧
≤≤≤≤+=
,,,
40607
21
21 n
nnnn
⎩⎨⎧
≤≤≤≤+=
.,,
40605
21
21 k
k k k k
0 1 2 3n1n2 x [0] x [1] x [2] x [3] x [4] x [5] x [6]
x [7] x [8] x [9] x [10] x [11]012
4 5 6
x [12] x [13] x [14] x [15] x [16] x [17] x [18] x [19] x [20]
3 x [21] x [22] x [23] x [24] x [25] x [26] x [27]
4 x [28] x [29] x [30] x [31] x [32] x [33] x [34]
0 1 2 3k 1k 2 X [0] X [5] X [10] X [15] X [20] X [25] X [30]
X [1] X [6] X [11] X [16] X [21]012
4 5 6
X [26] X [31] X [2] X [7] X [12] X [17] X [22] X [27] X [32]
3 X [3] X [8] X [13] X [18] X [23] X [28] X [33]
4 X [4] X [9] X [14] X [19] X [24] X [29] X [34]
11.36 (a) N = 12. Choose and41 = N .32 = N
Thus,
,,, 93343 41 ==== −C B A
.444 31 == − D
⎩⎨⎧
≤≤≤≤+=
,,,
203043
21
1221 n
nnnn
⎩⎨⎧
≤≤≤≤+=
.,,
203049
21
1221 k
k k k k
The corresponding index mappings are indicated below:
0 1 2 3n1n2 x [0] x [3] x [6] x [9] x [4] x [7] x [10] x [1] x [8] x [11] x [2] x [5]
012
0 1 2 3k 1k 2 X [0] X [9] X [6] X [4] X [1] X [10] X [7] X [8] X [5] X [2] X [11]
012
X [3]
(b) N = 15. Choose and31 = N .52 = N ,,, 105535 31 ==== −C B A .633 5
1 == − D
Not for sale 408
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 22/43
Thus,⎩⎨⎧
≤≤≤≤+=
,,,
402035
21
1521 nnnnn
⎩⎨⎧
≤≤≤≤+=
.,,
4020610
21
1521 k k k k k
The corresponding index mappings are indicated below:
0 1 2n1n2
x [0] x [5] x [10] x [3] x [8] x [6]
x [13] x [11] x [1]
x [9] x [14] x [4]
0123
4 x [12] x [2] x [7]
0 1 2k 1k 2
X [0] X [10] X [5] X [6] X [1] X [12]
X [11] X [7] X [2]
X [3] X [13] X [8]
0123
4 X [9] X [4] X [14] (c) N = 21. Choose and71 = N .32 = N
Thus,
The corresponding index mappings are indicated
below:
,,, 15533373 41 =×==== −C B A
.71777 31 =×== − D
⎩⎨⎧
≤≤≤≤+=
,,,
206073
21
2121 nnnnn
⎩
⎨⎧
≤≤
≤≤+=.
,,20
607152
12121
k
k k k k
0 1 2 3n1n2 x [0] x [3] x [6] x [9] x [12] x [15] x [18]
x [7] x [10] x [13] x [16] x [19]012
4 5 6
x [1] x [4] x [14] x [17] x [20] x [2] x [5] x [8] x [11]
0 1 2 3k 1k 2 X [0] X [15] X [9] X [3] X [18] X [12] X [6]
X [7] X [1] X [16] X [10] X [4]012
4 5 6
X [19] X [13] X [14] X [8] X [2] X [17] X [11] X [5] X [20]
(d) N = 35. Choose and71 = N .52 = N
Thus,
The corresponding index mappings are indicated
below:
,,, 15355575 71 =×==== −C B A
.213777 51 =×== − D
⎩⎨⎧
≤≤≤≤+=
,,,
406075
21
3521 n
nnnn
⎩⎨⎧ ≤≤ ≤≤+= .
,,40602115
213521 k
k k k k
0 1 2 3n1n2 x [0] x [5] x [10] x [15] x [20] x [25] x [30]
x [7] x [12] x [17] x [22] x [27]012
4 5 6
x [32] x [2] x [14] x [19] x [24] x [29] x [34] x [4] x [9]
3 x [21] x [26] x [31] x [1] x [6] x [11] x [16]
4 x [28] x [33] x [3] x [8] x [13] x [18] x [23]
0 1 2 3k 1k 2 X [0] X [15] X [30] X [10] X [25] X [5] X [20]
X [21] X [1] X [16] X [31] X [11]012
4 5 6
X [26] X [6] X [7] X [22] X [2] X [17] X [32] X [12] X [27]
3 X [28] X [8] X [23] X [3] X [18] X [33] X [13]
4 X [14] X [29] X [9] X [24] X [4] X [19] X [34]
Not for sale 409
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 23/43
11.37 N = 12. Choose and41 = N .32 = N
Thus,
,,, 93343 41 ==== −C B A .444 3
1 == − D
⎩
⎨⎧
≤≤≤≤+=
,,,
203043
2
11221 n
nnnn
⎩
⎨⎧
≤≤≤≤+=
.,,
203049
2
11221 k
k k k k
The corresponding index mappings are indicated below:
0 1 2 3n1n2 x [0] x [3] x [6] x [9] x [4] x [7] x [10] x [1] x [8] x [11] x [2] x [5]
012
0 1 2 3k 1k 2 X [0] X [9] X [6] X [4] X [1] X [10] X [7] X [8] X [5] X [2] X [11]
012
X [3]
Alternately,⎩⎨⎧
≤≤≤≤+=
,,,
203049
21
1221 nnnnk
⎩⎨⎧
≤≤≤≤+=
.,,
203043
21
1221 k k k k k
The corresponding index mappings are indicated below:
0 1 2 3k 1k 2 X [0] X [3] X [6] X [9] X [4] X [7] X [10] X [1] X [8] X [11] X [2] X [5]
012
0 1 2 3n1n2 x [0] x [9] x [6] x [4] x [1] x [10] x [7] x [8] x [5] x [2] x [11]
012
x [3]
Hence, and],[][ k Y k X 22 = .,,,],)([][ 51012612 12 K=++=+ k k Y k X
11.38 (a) N = 6. Choose and21 = N .32 = N
Thus,
,,, 33323 21 ==== −C B A .422 3
1 == − D
⎩⎨⎧ ≤≤≤≤
+= ,,
, 2010
23 21621 n
nnnn ⎩⎨⎧ ≤≤
≤≤+= .
,, 20
1043 2
1621 k k
k k k
The corresponding index mappings are indicated below:
0 1n1n2 x [0] x [3]
x [2] x [5]
x [4] x [1]
012
0 1n1k 2G[0,0] G[1,0]
G[0,1] G[1,1]
G[0,2] G[1,2]
012
0 1k 1k 2 X [0] X [3]
X [4] X [1]
X [2] X [5]
012
3-ptDFT
3-ptDFT
2-ptDFT
2-ptDFT
2-ptDFT
G[0,0]
G[1,0]
G[0,1]
G[1,1]
G[0,2]
G[1,2]
x [0]
x [1]
x [2]
x [3]
x [4]
x [5]
X [0]
X [1]
X [2]
X [3]
X [4]
X [5]
Not for sale 410
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 24/43
(b) N = 10. Choose and21 = N .52 = N
Thus,
,,, 55525 21 ==== −C B A .622 5
1 == − D
⎩⎨
⎧
≤≤≤≤
+= ,,
, 2040
25 21
1021 nn
nnn ⎩⎨
⎧
≤≤≤≤
+= .,
, 2040
65 21
1021 k k
k k k The corresponding index mappings are indicated below:
0 1n1n2 x [0] x [5] x [2] x [7]
01
x [4] x [9]2
0 1n1k 2G[0,0] G[1,0]G[0,1] G[1,1]G[0,2] G[1,2]
012
0 1k 1k 2 X [0] X [5] X [6] X [1] X [2] X [7]
012
x [6] x [1]3 x [8] x [3]4
G[0,3] G[1,3]3G[0,4] G[1,4]4
X [8] X [3]3 X [4] X [9]4
2-ptDFT
2-ptDFT
2-ptDFT
G[0,0]G[1,0]
x [0]
x [1]
x [2]
x [3]
x [4]
x [5]
X [0]
X [1]
X [2]
X [3]
X [4]
X [5]
2-ptDFT
2-ptDFT
5-ptDFT
5-ptDFT
x [7]
x [9]
x [6]
x [8]
X [6]
X [8]
X [7]
X [9]
G[0,1]
G[1,1]
G[0,2]
G[1,2]
G[1,3]
G[1,4]
G[0,3]
G[0,4]
(c) N = 12. Choose and31 = N .42 = N
Thus,
,,, 44434 31 ==== −C B A .333 4
1 == − D
⎩⎨⎧
≤≤≤≤+=
,,,
302034
21
1221 nnnnn
⎩⎨⎧
≤≤≤≤+=
.,,
302034
21
1221 k k k k k
The corresponding index mappings are indicated below:
0 1 2k 1n2G[0,0] G [1,0] G [2,0]G[0,1] G [1,1] G [2,1]G[0,2] G [1,2] G [2,2]
012
0 1 2n1n2 x [0] x [4] x [8] x [3] x [7] x [11] x [6] x [10] x [2]
012
x [9] x [1] x [5]3 G [0,3] G [1,3] G [2,3]3
0 1 2k 1k 2 X [0] X [4] X [8] X [3] X [7] X [11] X [6] X [10] X [2]
012
X [9] X [1] X [5]3
Not for sale 411
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 25/43
G[0,0]
G[1,0] x [0]
x [1]
x [2]
x [3]
x [4]
x [5]
X [0]
X [1]
X [2]
X [3]
X [4]
X [5]
4-ptDFT
x [7]
x [9]
x [6]
x [8] X [6]
X [8]
X [7]
X [9]
3-ptDFT
3-ptDFT
3-ptDFT
4-pt
DFT
4-ptDFT
x [10]
x [11]
X [10]
X [11]
3-ptDFT
G[0,1]
G[1,1]
G[0,2]
G[1,2]
G[1,3]
G[2,1]
G[0,3]
G[2,0]
G[2,2]
G[2,3]
(d) N = 15. Choose and51 = N .32 = N
Thus,
,,, 103353 51 ==== −C B A
.655 31 == − D
⎩⎨⎧
≤≤≤≤+=
,,,
204035
21
1021 nnnnn
⎩⎨⎧
≤≤≤≤+=
.,,
2040610
21
1021 k k k k k
The corresponding index mappings are indicated below:
0 1n1n2
x [0] x [5] x [2]
x [7]
01
x [4]
x [9]
2
0 1n1k 2
G[0,0] G[1,0]G[0,1] G[1,1]G[0,2] G[1,2]
012
0 1k 1k 2 X [0]
X [5]
X [6] X [1]
X [2] X [7]
012
x [6]
x [1]
3
x [8] x [3]
G[3,1]G[2,1]
3
G[2,2] G[3,2]
4
X [8]
X [3]
3
X [4] X [9]
2
x [12] x [11] x [14]
x [10] x [13]
2
G[2,0] G[3,0]
2 4
X [10] X [12]
X [11] X [14] X [13]
Not for sale 412
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 26/43
G[0,0]
G[1,0]
x [0]
x [1]
x [2]
x [3]
x [4]
x [5]
X [0]
X [1]
X [2]
X [3]
X [4]
X [5]
x [7]
x [9]
x [6]
x [8]
X [6]
X [8]
X [7]
X [9]
G[0,1]
G[1,1]
G[0,2]
G[1,2]
G[1,3]
G[1,4]
G[0,3]
G[0,4]
5-ptDFT
5-ptDFT
5-ptDFT
3-ptDFT
3-ptDFT
3-ptDFT
3-ptDFT
3-ptDFT x [10]
x [13]
x [11]
x [14]
x [12]
X [14]
X [11]
X [12]
X [13]
X [10]
G[2,4]
G[2,3]
G[2,2]
G[2,1]G[2,0]
11.39 Note that 1536 = 256 × 6. Now an -point DFT , with divisible by 6, can be computed
as follows:
where
N N
][][][][][ 6 / 22
6 / 1
1
06 / 0 N
k N N
k N
N
n N
nk N k X W k X W k X W n x k X ∑
=],[][][ 6 / 5
56 / 4
46 / 3
3 N
k N N
k N N
k N k X W k X W k X W
.50,]6[][
16
06 / 6 / ≤∑
=lll
N
r
rk N N W r x k X For ,1536 N we thus get
where
][][][][][ 25633
153625622
1536256115362560 k X W k X W k X W k X k X k k k
],[][ 25655
153625644
1536 k X W k X W k k .50,]6[][5111
0256256 ≤∑
=lll
r
rk W r x k X
X [k ]
768k W
6 z
x [n] DFT256-point
z
z
z
6
6
6
6
6
z
DFT256-point
DFT256-point
DFT256-point
DFT256-point
DFT256-point
x [n]000
x [n]001
x [n]010
x [n]011
x [n]100
x [n]101
768k W
768k W
768k W
1536k W
k ] X [ 256000
k ] X [ 256001
k ] X [ 256010
k ] X [ 256100
k ] X [ 256101
k ] X [ 256011
Not for sale 413
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 27/43
Now an N -point FFT algorithm requires N N
22log complex multiplications and
complex additions. Hence, an
N N 2log
6 N -point FFT algorithm requires )6 / (log 212
N N complex
multiplications and )6 / (log 26
N N complex additions. In addition, we need complex
multiplications and complex additions to compute the N -point DFT X[k]. Hence, for N =
1536, the evaluation of using six 256-point FFT modules requires
N 5
N 5
][k X N N N 5)6 / (log 212
complex multiplications and870415365)256(log128 2 = N N N 5)6 / (log 26
972815365)256(log256 2 = complex additions. It should be noted that a directcomputation of the 3072-point DFT would require 6431296 complex multiplications and2357760 complex additions.
11.40 (a) # of zero-valued samples to be added is 512 – 498 = 14.
(b) Direct computation of a 512-point DFT of a length-498 sequence requires (498) 2 =248004 complex multiplications and 497 ×498 = 247506 complex additions .(c) A 512-point Cooley-Tukey type FFT algorithm requires 2304)512(log256 2 = complexmultiplications and 46308)512(log512 2 = complex additions.
11.41 Hence, Since.ll α= z .lll α=φ−θ− oo j j
oo eeV A α is real, we have ,/, α== 11 oo V A
., 00 =φ=θ oo 11.42 (a) or)()()( z X z H zY = ).]1[]0[)(]1[]0[(]2[]1[]0[ 1121 z x x zhh z y z y y Now,
]),1[]0[])(1[]0[()1()1(]2[]1[]0[)1()( 0 x x hh X H y y yY zY ],0[]0[)()(]0[)()( 1 x h X H yY zY =
}).1{]0[])(1[]0[{)1()1(]2[]1[]0[)1()( 2 x x hh X H y y yY zY
From Eqs. (6.114) and (6.115), we arrive at
),()(
)()(
)()(
)()(
)()( 2
22
21
11
10
00
0 zY z I z I
zY z I z I
zY z I
z I zY where ),1)(1()( 1
21
10 z z z z z I
),1)(1()( 12
101
z z z z z I ).1)(1()( 11
102
z z z z z I Therefore,
),1()1)(1(
)1)(1()(
)(112
11
021
01
1
2
1
100
0 = z z z z z z
z z z z z I
z I
),1()1)(1(
)1)(1()()( 2
112
110
12
10
11
1 = z z z z z
z z z z
z I z I
and
).1()1)(1(
)1)(1()(
)( 1121
121
120
11
10
02
2 = z z z z z z
z z z z
z I z I
Hence,
Not for sale 414
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 28/43
)()1()()1()()1()( 211
21
12
011
21
zY z z zY z zY z z zY
222
1102
1122
102
11 )()()()()()(
⎟⎠
⎞⎜⎝
⎛ ⎠
⎞⎜⎝
⎛ z zY zY zY z zY zY zY
1
2
1
2
1 ])1[]0[])(1[]0[(])1[]0[])(1[]0[(]0[]0[ ⎟
⎠
⎞⎜
⎝
⎛ z x x hh x x hh x h
221
21 ])1[]0[])(1[]0[(]0[]0[])1[]0[])(1[]0[(
⎟⎠
⎞⎜⎝
⎛ z x x hh x h x x hh
.]1[]1[])0[]1[]1[]0[(]0[]0[ 21 z x h z x h x h x h
Ignoring the multiplications by ,21 computation of the coefficients of require the values
of and , which can be evaluated using only 3 multiplications.
)( zY
)( 2 zY ),(),( 10 zY zY
(b) or)()()( z X z H zY = ]2[]1[]0[]2[]1[]0[]4[]3[]2[]1[]0[ 21214321 z x z x x zh zhh z y z y z y z y y
Now,
]),2[4]1[2]0[])(2[4]1[2]0[()()(21
0 x x x hhhY zY
]),2[]1[]0[])(2[]1[]0[()1()( 1 x x x hhhY zY ],0[]0[)()( 2 x hY zY = ]),2[]1[]0[])(2[]1[]0[()1()( 3 x x x hhhY zY
]).2[4]1[2]0[])(2[4]1[2]0[()()(21
4 x x x hhhY zY
From Eqs. (6.114) and (6.115), we arrive at
),()(
)()(
)()(
)()(
)()(
)()(
)()(
)()( 4
44
43
33
32
22
21
11
10
00
0 zY z I
z I zY
z I
z I zY
z I
z I zY
z I
z I zY
z I
z I zY where
),1)(1)(1)(1()( 141312110 z z z z z z z z z I
),1)(1)(1)(1()( 14
13
12
101
z z z z z z z z z I
),1)(1)(1)(1()( 14
13
11
102
z z z z z z z z z I
),1)(1)(1)(1()( 14
12
11
103
z z z z z z z z z I
).1)(1)(1)(1()( 13
12
11
104
z z z z z z z z z I Therefore,
),1)(1()1)(1)(1)(1(
)1)(1)(1)(1()(
)( 21211
121
104
103
102
101
14
13
12
11
00
0 = z z z z z z z z z z z
z z z z z z z z z I z I
),1)(1()1)(1)(1)(1()1)(1)(1)(1(
)()( 2
4111
321
141
131
121
10
14
13
12
10
111 = z z z
z z z z z z z z z z z z z z z z
z I z I
),1)(1()1)(1)(1)(1(
)1)(1)(1)(1()(
)( 2412
124
123
121
120
14
13
11
10
22
2 = z z z z z z z z z z
z z z z z z z z
z I z I
),1)(1()1)(1)(1)(1(
)1)(1)(1)(1()(
)( 24111
32
134
132
131
130
14
12
11
10
33
3 = z z z z z z z z z z z
z z z z z z z z
z I
z I
Not for sale 415
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 29/43
).1)(1()1)(1)(1)(1(
)1)(1)(1)(1()(
)( 21211
121
143
142
141
140
13
12
11
10
44
4 = z z z z z z z z z z z
z z z z z z z z
z I z I
Hence,
)())())( 14
413
4121(
32
04
4132
211(
121 zY z z z z zY z z z z zY
)())()1( 3441
341
21(32
2441
245
zY z z z z zY z z
)() 44
2131
211(
121 zY z z z z
1412
133
213
2012
12 )()()()()(
⎟⎠
⎞⎜⎝
⎛ z zY zY zY zY zY
2424
133
224
513
2024
1 )()()()()( ⎟⎠
⎞⎜⎝
⎛ z zY zY zY zY zY
3412
136
116
1012
1 )()()()( ⎟⎠
⎞⎜⎝
⎛ z zY zY zY zY
.)()()()()( 4424
1
36
1
24
1
16
1
024
1 ⎟
⎠
⎞⎜
⎝
⎛ z zY zY zY zY zY
Substituting the expressions for and in the aboveequation, we then arrive at the expressions for the coefficients in terms of thecoefficients and Thus,
),(),(),(),( 3210 zY zY zY zY ),( 4 zY ]}[{ n y
]}[{ nh ]}.[{ n x ],0[]0[)(]0[ 2 x h zY y =
],0[]1[]1[]0[)()()()(]1[ 4121
332
132
0121
x h x h zY zY zY zY y
],0[]2[]1[]1[]2[]0[)()()()()(]2[ 4241
332
245
132
0241
x h x h x h zY zY zY zY zY y
],1[]2[]2[]1[)()()()(]3[ 4121
361
161
0121
x h x h zY zY zY zY y
].2[]2[)()()()()(]4[ 424136
124116
10241 x h zY zY zY zY zY y =
Hence, ignoring the multiplications by ,,,,,61
41
45
32
121 and ,
241 computation of the
coefficients of require the values of and which can be evaluated using only 5 multiplications.
)( zY ),(),(),(),( 3210 zY zY zY zY ),( 4 zY
11.43 or)()()( z X z H zY = 1121 ]1[]0[]1[]0[]2[]1[]0[ z x x zhh z y z y y
.]2[[]2[])0[]1[]1[]0[(]0[]0[ 21 z x h z x h x h x h Hence, ],0[]0[]0[ x h y = ].1[]1[]2[],0[]1[]1[]0[]1[ x h y x h x h y = Now,
].1[]0[]1[]1[]0[]1[]1[]0[]0[])1[]0[])(1[]0[( y x h x h x h x h x x hh = As a result,evaluation of requires the computation of 3 products, , and
. In addition, it requires 4 additions,)()( z X z H ]1[]1[],0[]0[ x h x h
])1[]0[])(1[]0[( x x hh ],1[]0[],1[]0[ x x hh and]1[]1[]0[]0[])1[]0[])(1[]0[( x h x h x x hh .
11.43 Let the two length- sequences be denoted by and Denote the sequence]}[{ nh ]}.[{ n x N
Not for sale 416
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 30/43
generated by the linear convolution of and][nh ][n x as Let and].[n y ∑ =
= 10
][)( N
nn znh z H
.][)( 1
0
=
= N n
n zn x z X Rewrite and in the form and
where
)( z H )( z X ),()()( 12 /
0 z H z z H z H N
),()()( 12 /
0 z X z z X z X N ,][)( 1)2 / (
00 ∑ =
= N n
n znh z H
,][)( 1)2 / (
0 21 ∑ =
N n
n N znh z H ,][)(
1)2 / (00 ∑
= = N
nn zn x z X
.][)( 1)2 / (
0 21 ∑ =
N n
n N zn x z X Therefore, we can write
)()()()()( 12 /
012 /
0 z X z z X z H z z H zY N N
)()()()()()()()( 1101102 /
00 z X z H z z X z H z X z H z z X z H N N
where),()()( 212 /
0 zY z zY z zY N N ),()()( 000 z X z H zY =
and),()()()()( 01101 z X z H z X z H zY ).()()( 112 z X z H zY = Note that and are
products of two polynomials of degree
)(0 zY )(1 zY
2 N
each, and hence, require2
2 ⎟⎠⎞⎜⎝⎛ N
multiplications each.
Now, we can write ).()()()()()()( 1010101 zY zY z X z X z H z H zY Since,
is a product of two polynomials of degree)()()()( 1010 z X z X z H z H 2
N each, it can be
computed using2
2⎟⎠
⎞⎜⎝
⎛ N multiplications. As a result, )()()( z X z H zY = can be computed using
2
23 ⎟⎠
⎞⎜⎝
⎛ N multiplications instead of multiplications. If is a power-of-2,2 N N 2
N is even, and
the same procedure can be applied to compute and reducing further the
number of multiplications. This process can be continued until the sequences beingconvolved are of length 1 each.
),(),( 10 zY zY ),(2 zY
Let )( N R denote the total number of multiplications required to compute the linearconvolution of two –length sequences. Then, in the method outlined above, we have N
)2 / (3)( N N R R ⋅ with .1)1( =R A solution of this equation is given by .3)( 2log N N =R
11.45 The dynamic range of a signed –bit integer is given by B η )12()12( )1()1( B B η
which for is given by32 B ).12()12( 3131 η
(a) For and the value of a 32-bit floating-point number is given by
Hence, the value of the largest number is and the value of the
smallest number is The dynamic range is therefore
6 E ,25 M
).(2)1( 31 M E s η ,232≈.232 .22 32×
(b) For 7 E and the value of a 32-bit floating-point number is given by,24 M
).(2)1( 63 M E s η Hence, the value of the largest number is and the value of the,264≈
Not for sale 417
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 31/43
smallest number is The dynamic range is therefore.264 .22 64× (c) For and the value of a 32-bit floating-point number is given by
Hence, the value of the largest number is and the value of the
smallest number is The dynamic range is therefore
8 E ,23 M
).(2)1( 127 M E s η ,2128≈
.2128
.22 128
× Hence, the dynamic range in a floating-point representation is much larger than that in a fixed-
point representation with the same wordlength.
11.46 A 32-bit floating-point number in the IEEE Format has E = 8 and M = 23. Also, theexponent E is coded in a biased form as 127 E with certain conventions for special casessuch as E = 0, 255, and M = 0 (See page 637 of text).
Now, a positive 32-bit floating-point number represented in the “normalized” form have anexponent in the range 0 < E < 255, and is of the form ).1(2)1( 127 M E s
∆ η Hence, the
smallest positive number that can be represented will have E = 1, and
and has therefore a value given by
,000022
4 4 34 4 21 Kbits
M =
.1018.12 38126 × For the largest positive number, E = 254, and Thus, here.1111
224 4 34 4 21 K
bits
M = 22)11111(2)1( 127
22
1270 ×4 4 34 4 21 K
bits
∆η
.104.3 38× Note: For representing numbers less than ,2 126 IEEE format uses the “de-normalized”
form where E = 0, and ).0(2)1( 126 M s∆
η In this case, the smallest positive number
that can be represented is given by 149
22
1260 2)10000(2)1( ≈4 4 34 4 21
K
bits
∆η
.104013.1 45× 11.47 For a two’s-complement binary fraction the decimal equivalent for
is simply For
,21 baaas K∆
0s .21
ibi ia = ,1s the decimal equivalent is given by
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛
=
bb
i
iia 22)1(
1
=
=
∑ )21(22211
bbb
i
ii
b
i
i a bib
iia
= 22
1
Hence, the decimal equivalent of is given by.211
ib
iia
= baaas K21∆
.21
ib
iias
=
11.48 For a ones’-complement binary fraction the decimal equivalent for,21 baaas K∆
Not for sale 418
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 32/43
0s is simply For.21
ibi ia = ,1s the decimal equivalent is given by
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛ ∑
=
b
i
iia
1
2)1( =
=
∑ )21(2211
bb
i
ii
b
i
i a ib
iia
= 2
1
Hence, the decimal equivalent of is given by.211
ib
iia
= baaas K21∆
.2)21(1
ib
ii
b as
= ∑
11.49 .01111153125.0,10110068750.0,11001078125.0 103102101 ∆∆∆ =ηηη
.01111110110011001021 ∆∆∆ =ηη
Dropping thee overflow bit and adding )( 21 ηη to :3η
,9375.0111100011111011110)( 10321 =∆∆∆nηη where we have dropped the carry bit in the MSB location. Note that the final sum is correctinspite of the overflow .
11.49 The transformation ω β αω ˆcoscos is equivalent to ,22
ˆˆ
⎟⎠
⎞
⎜⎝
⎛ ωωωω β α
j j j j eeee
which by analytic continuation can be expressed as .2ˆˆ
2
11
⎟⎠
⎞
⎜⎝
⎛ z z z z β α Now, let
be a Type 1 linear-phase FIR transfer function of degree As indicated in Eq. (8.128),
can be expressed as
)( z H
.2 M
)( z H ∑ = ⎟⎠
⎞⎜⎝
⎛ = M n
n M z zna z z H
01
2][)( with a frequency response
given by with,)](cos[)(0=
= M n
n jM j naee H ωωω ∑ = M n
nna H 0
)](cos[)( ωω(
denoting
the amplitude function or the zero-phase frequency response. The amplitude function or thezero-phase frequency response of the transformed filter obtained by applying the mapping
is therefore given byω β αω ˆcoscos .)ˆcos]([)ˆ(0= M
nnna H ω β αω
( Or, equivalently,
the transfer function of the transformed filter is given by
∑ =
⎟⎠
⎞
⎜⎝
⎛ M n
n M z z
na z z H 0
1
2ˆˆ
][)ˆ( β α A convenient way to realize is to consider the
realization of the parent transfer function in the form of a Taylor structure as outlined inProblem 8.17 which is obtained by expressing in the form
)ˆ( z H
)( z H )( z H
.2
1][)(
0
2∑ =
⎟⎠
⎞
⎜⎝
⎛ = M n
nn M z
zna z H Similarly, the transformed filter can be realized by
Not for sale 419
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 33/43
replacing each block2
1 2 z in the Taylor structure by the block .
2
ˆ1ˆ
21
z z β α
Now, for a lowpass-to-lowpass transformation we can impose the condition.)()ˆ(
00ˆ ==ωω
ωω H H ((
This condition is met if 1 β α and .10 <α In this case, the
transformation reduces to .ˆcos)1(cos ωααω From the plot of the mapping given below, itfollows that as is varied between 0 and 1,α .ˆ cc ωω <
-1 -0.5 0 0.5 1-1
-0.5
0
0.5
1
cos ω
c o s ω
^
On the other hand if is desired along with a lowpass-to-lowpass transformation, we
can impose the condition
.ˆ cc ωω >.)()ˆ(
ˆ π ωπ ωωω == H H
(( This condition is met if and
The corresponding transformation is now given by Fromthe plot of the mapping given below, it follows that as is varied between 0 and 1,
α β 1
.11 ≤α .ˆcos)1(cos ωααω α .ˆ cc ωω >
-1 -0.5 0 0.5 1-1
-0.5
0
0.5
1
cos ω
c o s ω
^
11.51 (a) .][)( ∑−
=
−=1
0
N
n
n z n x z X ,)()(
][)()( z D z P
z
z n x z X z X
N
n
n
z
z z (
(
(
((
(
( =⎟⎟
⎠
⎞⎜⎜
⎝
⎛
α−+α−== ∑
−
= −
−
α−+α−= −
−−1
01
1
1 11
11
where ,)()]([][)( nn N N
n
N
n
n z z n x z n p z P 111
0
11
01 −−−−
=
−−
=
− +α−α−== ∑∑ (((( and
.)(][)( 111
01 −−−−
=α−== ∑ N n
N
n z z nd z D (((
Not for sale 420
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 34/43
(b) ,][][
)()(
)(][/
/k Dk P
z D z P
z X k X N k j
N k j
e z e z (
(
((
((
(( ===
ππ
== 2
2 where N k je z z P k P /)(][ π== 2(((
is
the –point DFT of the sequence and N ][n p N k je z z Dk D /)(][ π== 2(((
is the –point DFT
of the sequence
N
].[nd
(c) Let and[ ]t N p p p ][][][ 110 −= LP [ ].][][][ t N x x x 110 −= LX Without anyloss of generality, assume in which case4= N
( )][][][][][)( 3210 323
0 x x x x z n p z P
n
n α−α+α−== ∑=
−((
( 1222 332212103 −α+α+α−α++α−+ z x x x x (][][)(][)(][
( ) 2222 33211203 −α−α+−α+α−α+ z x x x x (][][)(][)(][
( ) .][][][][ 323 3210 −α+α−α+α−+ z x x x x ( Equating like powers of 1− z ( we can write
or It can be shown
that the elements
XQP ⋅= .
][][][][
)(][][][][
⎢⎢⎢⎢
⎣
⎡
αα−α+α−α
α+α−α−
=⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
23213
1
3210
23
22
2
p p p p
)(
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥⎥
⎦
⎤
α−α−α+
αα+α−α−α
3210
1321
322
2232
x x x x
,,,, 30 ≤≤ sr q sr of the 4 ×4 matrix Q can be determined as follows:
(i) The first row is given by ,)(, s
sq α−=0
(ii) The first column is given by ,)()!(!
!)(,
r r r r r r
C q α−−
=α−=333
0 and
(iii)
The remaining elements can be obtained using the recurrence relation.,,,, sr sr sr sr qqqq 1111 −−−− α+α−=
In the general case, we only change the computation of the elements of the first column
using the relation .)()!(!
)!()(,
r r r
N r r N r
N C q α−
−−−=α−= −111
0
11.52 The highpass transfer function can be expressed as [ ],)()()( 1021 z A z A z H herew
1
1
03038.01
3038.0)( =
z
z z A and
21
21
1639.04012.01
4012.0639.0)( =
z z
z z z A . The tunable highpass
transfer function is obtained by applying the lowpass-to-lowpass transformation of Eq.
(11.115) where the tuning parameter is given by .]2 / )ˆ6.0sin[(]2 / )ˆ6.0sin[(
p
p
ωπ ωπ α
= The tunable
highpass transfer function is then given by [ ],)(ˆ)(ˆ),( 1021 z A z A z H α where from Eqn.
Not for sale 421
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 35/43
(11.119)1
1
0]9077.03038.0[1
]9077.03038.0[)(ˆ
= z
z z A
α
α and from Eqn. (11.123)
.]1448.0639.0[]439.34012.0[1
]439.34012.0[]1448.0639.0[)(ˆ
21
211 =
z z
z z z A
αα
αα
11.53 The lowpass transfer function can be expressed as [ ],)()()( 1021
z A z A z H where
1
1
01584.01
1584.0)( =
z
z z A and .
3554.04191.01
4191.03554.0)(
21
21
1 = z z
z z z A The tunable bandpass
transfer function is obtained by applying the lowpass-to-bandpass transformation of Eq.(11.124) where the tuning parameter is given by oω β cos with denoting the centerfrequency of the bandpass filter. The tunable bandpass transfer function is then given by
oω
[ ],)(ˆ)(ˆ),( 1021
z A z A z H α where⎟⎠
⎞
⎜⎝
⎛
⎟⎠
⎞
⎜⎝
⎛
= 1
1
1
1
11
11
0
1584.01
1584.0
)(ˆ
z
z
z
z
z
z
z A
β
β
β
β
and
.
3554.04191.01
4191.03554.0
)(ˆ2
12
11
2
12
11
1
1
1
1
1
1
1
1
1
⎟⎠
⎞
⎜⎝
⎛ ⎟⎠
⎞
⎜⎝
⎛
⎟⎠
⎞
⎜⎝
⎛ ⎟⎠
⎞
⎜⎝
⎛ =
z
z
z
z
z
z
z
z
z z
z z
z A
β
β
β
β
β
β
β
β
M11.1 (a) Numerator coefficients =[0.0528 0.0797 0.1295 0.1295 0.0797 0.0528] Denominator coefficients =[1.0000 -1.8107 2.4947 -1.8801 0.9537 -0.2336]
0 5 10 15 20 25 30-0.2
0
0.2
0.4
0.6
Time index n
m p
u e
Impulse response samples
(b) Numerator coefficients =[0.0084 -0.0335 0.0502 -0.0335 0.0084]Denominator coefficients =
Not for sale 422
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 36/43
[1.0000 2.3741 2.7057 1.5917 0.4103]
0 5 10 15 20 25 30-0.4
-0.2
0
0.2
0.4
Time index n
m p u
e
Impulse response samples
(c) Numerator coefficients = [0.0003 0 -0.0019 0 0.0057 0-0.0095 0 0.0095 0 -0.0057 0 0.0019 0 -0.0003]Denominator coefficients = [1.0000 1.7451 4.9282 6.11959.8134 9.2245 10.4323 7.5154 6.4091 3.4595
2.2601 0.8470 0.4167 0.0856 0.0299]
0 5 10 15 20 25 30
-0.4
-0.2
0
0.2
0.4
Time index n
m p
u e
Impulse response samples
M11.2 The modified MATLAB program is given below:
n = 0:60; w = input('Normalized angular frequency vector = ');
num = input('Numerator coefficients = ');den = input('Denominator coefficients = ');x1 = cos(w(1)*pi*n); x2 = cos(w(2)*pi*n);x = x1+x2;% Generate the output sequence by filtering the inputy = filter(num,den,x);% Plot the input and the output sequencesfigure(1)stem(n,x);xlabel('Time index n'); ylabel('Amplitude');title('Input sequence');figure(2)stem(n,y);xlabel('Time index n'); ylabel('Amplitude');
Not for sale 423
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 37/43
title('Output sequence');
The plots generated by the above program for the filter of Example 9.14 for an inputcomposed of a sum of two sinusoidal sequences of angular frequencies, 0.3 π and 0.6 π, aregiven below:
0 10 20 30 40 50 60-2
-1
0
1
2
Time index n
A m p
l i t u d e
Input sequence
0 10 20 30 40 50 60
-1
-0.5
0
0.5
1
Time index n
m p
u e
Output sequence
The blocking of the high-frequency signal by the lowpass filter can be demonstrated byreplacing the statement stem(n,x); with stem(n,x2); and the statementy=filter(num,den,x); with y=filter(num,den,x2); . The plots of the high-frequency input signal and the corresponding output are shown below:
0 10 20 30 40 50 60-1
-0.5
0
0.5
1
Time index n
m p
u e
Input sequence
0 10 20 30 40 50 60
-0.2
-0.1
0
0.1
0.2
0.3
Time index n
m p
u e
Output sequence
M11.3 The plots generated by the program of Exercise M11.2 for the filter of Example 9.15for an input composed of a sum of two sinusoidal sequences of angular frequencies, 0.3 π and0.6π, are given below:
0 10 20 30 40 50 60-2
-1
0
1
2
Time index n
A m p
l i t u d e
Input sequence
0 10 20 30 40 50 60
-0.6
-0.4
-0.2
0
0.2
0.4
Time index n
m p
u e
Output sequence
Not for sale 424
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 38/43
The blocking of the low-frequency signal by the highpass filter can be demonstrated byreplacing the statement stem(n,x); with stem(n,x1); and the statementy=filter(num,den,x); with y=filter(num,den,x1); . The plots of the low-frequency input signal and the corresponding output are shown below:
0 10 20 30 40 50 60-1
-0.5
0
0.5
1
Time index n
m p
u e
Input sequence
0 10 20 30 40 50 60
-0.2
-0.1
0
0.1
0.2
Time index n
m p
u e
Output sequence
M11.4 % The factors for the transfer function of% the lowpass filter of Example 9.14 are% num1 = [0.0528 0.0528 0];% den1 = [1.0000 -0.4909 0];% num2 = [1.0000 0.6040 1.0000];% den2 = [1.0000 -0.7624 0.5390];% num3 = [1.0000 -0.0949 1.0000];% den3 = [1.0000 -0.5574 0.8828];N = input( ‘Total number of sections = ‘ );for k = 1:N;
num(k,:) = input( 'Numerator factor = ' );den(k,:) = input( 'Denominator factor = ' );
endn = 0:60;
w = input( 'Normalized angular frequency vector = ' );x1 = cos*w(1)*pi*n); x2 = cos*w(2)*pi*n);x = x1 + x2;% Plot the input sequencefigure(1)stem(n,x);xlabel( 'Time index n' ); ylabel( 'Amplitude' );title( 'Input sequence' );si = [0 0];for k = 1:N;
y(k,:) = filter(num(k,:),den(k,:),x,si);x = y(k,:);
end% Plot the output sequencefigure(2)stem(n,y(N,:));xlabel( 'Time index n' ); ylabel( 'Amplitude' );title( 'Output sequence' );
Not for sale 425
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 39/43
The plots generated by the above program for the filter of Example 9.14 for an inputcomposed of a sum of two sinusoidal sequences of angular frequencies, 0.3 π and 0.6 π, aregiven below:
0 10 20 30 40 50 60-2
-1
0
1
2
Time index n
A m p
l i t u d e
Input sequence
0 10 20 30 40 50 60
-1
-0.5
0
0.5
1
Time index n
m p
u e
Output sequence
M11.5 % The factors for the transfer function of
% the highpass filter of Example 9.15 are% num1 = [0.0084 -0.0167 0.0084];% den1 = [1.0000 1.3101 0.5151];% num2 = [1.0000 -2.0000 1.0000];% den2 = [1.0000 1.0640 0.7966];
The plots generated by the program of Exercise M11.4 for the filter of Example 9.15 for aninput composed of a sum of two sinusoidal sequences of angular frequencies, 0.3 π and 0.6 π,are given below:
0 10 20 30 40 50 60-2
-1
0
1
2
Time index n
A m p
l i t u d e
Input sequence
0 10 20 30 40 50 60
-0.6
-0.4
-0.2
0
0.2
0.4
Time index n
m p
u e
Output sequence
M11.6 To apply the function direct2 to filter a sum of two sinusoidal sequences, we replacethe statement y = filter(num,den,x,si); in the MATLAB program given in thesolution of Exercise M11.2 with the statement y = direct2(num,den,x,si); . The
plots generated by running the modified program for the data given in this problem are given below:
Not for sale 426
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 40/43
0 10 20 30 40 50 60
-2
-1
0
1
2
Time index n
A m p
l i t u d e
Input sequence
0 10 20 30 40 50 60
-1
-0.5
0
0.5
1
Time index n
m p
u e
Output sequence
M11.7 The MATLAB program that can be used to compute all DFT samples using the functiongoertzel and the function fft is as follows:
clearN = input( 'Desired DFT length = ' ) ;x = input( 'Input sequence = ' );for j = 1:N
Y(j)=goertzel(x,j);enddisp( 'DFT samples computed using goertzel are ' ) ;disp(Y)disp( 'DFT samples computed using fft are ' ) ;X = fft(x); disp(X);
Results obtained for two input sequences of lengths 8 and 12, respectively, are given below:
Desired DFT length = 8Input sequence = [1 2 3 4 4 3 2 1]
DFT samples computed using goertzel areColumns 1 through 420.0000 -5.8284-2.4142i -0.0000-0.0000i -0.1716-0.4142iColumns 5 through 80-0.0000i -0.1716+0.4142i 0.0000-0.0000i -5.8284+2.4142i
DFT samples computed using fft areColumns 1 through 420.0000 -5.8284-2.4142i 0 -0.1716-0.4142iColumns 5 through 80 -0.1716+0.4142i 0 -5.828 +2.4142i
Desired DFT length = 12Input sequence = [1 2 4 8 10 12 7 3 -4 5 0 6]
DFT samples computed using goertzel areColumns 1 through 454.0000 -13.0622-21.0885i 1.5000+19.9186i -4.0000-2.0000iColumns 5 through 8
Not for sale 427
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 41/43
4.5000+2.5981i -0.9378+10.0885i -18.0000+0.0000i-0.9378-10.0885i
Columns 9 through 124.5000-2.5981i -4.0000+2.0000i 1.5000-19.9186i -
13.0622 +21.0885i
DFT samples computed using fft are54.0000 -13.0622-21.0885i 1.5000+19.9186i -4.0000-2.0000iColumns 5 through 8
4.5000+2.5981i -0.9378+10.0885i -18.0000+0.0000i-0.9378-10.0885i
Columns 9 through 124.5000-2.5981i -4.0000+2.0000i 1.5000-19.9186i -13.0622 +21.0885i
M11.8 The MATLAB program that can be used to verify the plots of Figure 11.43 is given below:
%Program_11_8.m[z,p,k] = ellip(5,0.5,40,0.4);a = conv([1 -p(1)],[1 -p(2)]);b =[1 -p(5)];c = conv([1 -p(3)],[1 -p(4)]);
w = 0:pi/255:pi;alpha = 0;an1 = a(2) + (a(2)*a(2) - 2*(1 + a(3)))*alpha;an2 = a(3) + (a(3) -1)*a(2)*alpha;g = b(2) - (1 - b(2)*b(2))*alpha;cn1 = c(2) + (c(2)*c(2) - 2*(1 + c(3)))*alpha;cn2 = c(3) + (c(3) -1)*c(2)*alpha;a = [1 an1 an2];b = [1 g]; c = [1 cn1 cn2];h1 = freqz(fliplr(a),a,w); h2 = freqz(fliplr(b),b,w);h3 = freqz(fliplr(c),c,w);ha = 0.5*(h1.*h2 + h3);ma = 20*log10(abs(ha));alpha = 0.1;an1 = a(2) + (a(2)*a(2) - 2*(1 + a(3)))*alpha;an2 = a(3) + (a(3) -1)*a(2)*alpha;g = b(2) - (1 - b(2)*b(2))*alpha;cn1 = c(2) + (c(2)*c(2) - 2*(1 + c(3)))*alpha;cn2 = c(3) + (c(3) -1)*c(2)*alpha;a = [1 an1 an2];b = [1 g]; c = [1 cn1 cn2];h1 = freqz(fliplr(a),a,w); h2 = freqz(fliplr(b),b,w);h3 = freqz(fliplr(c),c,w);hb = 0.5*(h1.*h2 + h3);mb = 20*log10(abs(hb));alpha = -0.25;an1 = a(2) + (a(2)*a(2) - 2*(1 + a(3)))*alpha;an2 = a(3) + (a(3) -1)*a(2)*alpha;g = b(2) - (1 - b(2)*b(2))*alpha;cn1 = c(2) + (c(2)*c(2) - 2*(1 + c(3)))*alpha;cn2 = c(3) + (c(3) -1)*c(2)*alpha;
Not for sale 428
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 42/43
a = [1 an1 an2];b = [1 g]; c = [1 cn1 cn2];h1 = freqz(fliplr(a),a,w); h2 = freqz(fliplr(b),b,w);h3 = freqz(fliplr(c),c,w);hc = 0.5*(h1.*h2 + h3);mc = 20*log10(abs(hc));plot(w/pi,ma,'r-',w/pi,mb,'b--',w/pi,mc,'g-.');axis([0 1 -80 5]);xlabel( 'Normalized frequency' );ylabel ('Gain, dB' );legend( 'b--' , '\alpha = 0.1 ' , 'r-' , '\alpha = 0 ' , 'g-.' , '\alpha = -0.25' );
M11.9 The MATLAB program that can be used to verify the plots of Figure 11.44 is given below:%Program_11_9.m
w = 0:pi/255:pi; wc2 = 0.31*pi;
f = [0 0.36 0.46 1];m = [1 1 0 0];b1 = remez(50, f, m);h1 = freqz(b1,1,w);
m1 = 20*log10(abs(h1));n = -25:-1;c = b1(1:25)./sin(0.41*pi*n);d = c.*sin(wc2*n);q = (b1(26)*wc2)/(0.4*pi);b2 = [d q fliplr(d)];h2 = freqz(b2,1,w);
m2 = 20*log10(abs(h2)); wc3 = 0.51*pi;
d = c.*sin(wc3*n);q = (b1(26)*wc3)/(0.4*pi);b3 = [d q fliplr(d)];h3 = freqz(b3,1,w);
m3 = 20*log10(abs(h3));plot(w/pi,m1,'r-',w/pi,m2,'b--',w/pi,m3,'g-.');axis([0 1 -80 5]);xlabel( 'Normalized frequency' );ylabel( 'Gain, dB' );legend('b--','\omega_c = 0.3\pi','r-','\omega_c = 0.41\pi','g-.','\omega_c = 0.51\pi')
M11.10 The MATLAB program to evaluate Eqs. (11.155) is given below:%Program_11_10.mx = 0:0.001:0.5;y = 3.140625*x + 0.0202636*x.^2 - 5.325196*x.^3 +0.5446778*x.^4 + 1.800293*x.^5;x1 = pi*x;z = sin(x1);plot(x,y);xlabel( 'Normalized angle, radians' );ylabel( 'Amplitude' );title( 'Approximate sine values' );grid;axis([0 0.5 0 1]);pauseplot(x,y-z);xlabel( 'Normalized angle, radians' );ylabel( 'Amplitude' );title( 'Error of approximation' );grid;
The plots generated by the above program are as indicated below:
Not for sale 429
8/12/2019 Chapter11 SM[1]
http://slidepdf.com/reader/full/chapter11-sm1 43/43
0 0.1 0.2 0.3 0.4 0.50
0.2
0.4
0.6
0.8
1
Normalized angle, radians
A m p
l i t u d e
Approximate sine values
0 0.1 0.2 0.3 0.4 0.5
-2
-1
01
2
3
4x 10 -5
Normalized angle, radians
A m p
l i t u d e
Error of approximation
M11.11 The MATLAB program to evaluate Eqs. (11.156) and (11.135) is given below:%Program_11_11.mk = 1;for x = 0:.01:1
op1 = 0.318253*x+0.00331*x^2-0.130908*x^3+0.068524*x^4-0.009159*x^5;op2 = 0.999866*x-0.3302995*x^3+0.180141*x^5-
0.085133*x^7+0.0208351*x^9;arctan1(k) = op1*180/pi;arctan2(k) = 180*op2/pi;actual(k) = atan(x)*180/pi;k = k+1;
endsubplot(211)x = 0:.01:1;plot(x,arctan2);ylabel( 'Angle, degrees' );xlabel( 'Tangent Values' );subplot(212)plot(x,actual-arctan2,'--');ylabel( 'Tangent Values' );xlabel( 'error, radians' );
The plots generated by the above program are as indicated below:
0 0.2 0.4 0.6 0.8 10
10
20
30
40
50
A n g
l e ,
d e g r e e s
Tangent Values 0 0.2 0.4 0.6 0.8 1
-1
-0.5
0
0.5
1x 10 -3
a n g e n
a u e s
error, radians