(N - A) James H. McClellan, Ronald W.schafer, Mark A. Yoder - DSP First Solution Manual-Pears Education (2015)

T
an his
th d w
o
or eir is p rk
w sa co ro is
ill le u vi pr
de o rse de ot
st f a s d s ec
ro n an o te
y y p d le d
th a a ly by
e rt ss fo U
in o e r
te f t ss th nite
gr hi in e
ity s w g us d S
of or stu e o tat
th k ( de f i es
e in nt ns co
w cl le tr p
or ud a uc y
k r
an ing rnin tors igh
d on g. in t la
is D w
no the iss tea s
Second Edition
t p W em ch
er or in ing
m ld a
itt W tio
ed id n
. e
W
eb
)
Solutions Manual for DSP First
J.H. McClellan, R.W. Schafer, M.A. Yoder
exist. No portion of this material may be reproduced, in any form or by any means, without permission in writing from the publisher.
© 2016 Pearson Education, Inc., Hoboken, NJ. All rights reserved. This material is protected under all copyright laws as they currently
Chapter 2
Sinusoids
2-1 Problems
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
1
CHAPTER 2. SINUSOIDS
P-2.1 DSP First 2e
9
6
3
x.t /
0
-3
-6
-9
-20 -16 -12 -8 -4 0 4 8 12 16 20
Time t (ms)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.2 DSP First 2e
In the plot the period can be measured, T = 12.5 ms ⇒ ω0 = 2π/(12.5 × 10−3 ) = 2π(80) rad.
Positive peak closest to t = 0 is at t 1 = 2.5 ms ⇒ ϕ = −2π(2.5 × 10−3 )/(12.5 × 10−3 ) = 2π/5 = −0.4π rad.
Amplitude is A = 8.
x(t) = 8 cos(160πt − 0.4π)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.3 DSP First 2e
(a) Plot of cos θ

1
0.5
cos
-0.5
-1
3 2 0 2 3
(rad)
(b) Plot of cos(20πt)

1
0.5
cos.20 t /
-0.5
s
w
ng
la
hi
)
n
ht
-1
eb
c
tio
ea
rig
W
a
t
py
in
-0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1
in
e
em
co
d
s
i
W
r
Time t (s)
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
(c) Plot of cos(2π/T0 + π/2)

of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
1
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
0.5
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
x.t /
ed
k
as
or
0
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
-0.5
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
-1
th
y
l
sa
ro
T0 3T0 =4 T0 =2 T0 =4 0 T0 =4 T0 =2 3T0 =4 T0
st
or
de
Time t (s)
ill
w

P-2.4 DSP First 2e
( jθ) 2 ( jθ) 3 ( jθ) 4 ( jθ) 5

e jθ = 1 + jθ + + + + +···
2! 3! 4! 5!
θ2 θ3 θ4 θ5
= 1 + jθ − −j + + j +···
2! 3! 4! 5!
θ 2 θ 4 θ3 θ5
! !
= 1− + − · · · +j θ − + +···
2! 4! 3! 5!
| {z } | {z }
cos θ sin θ
Thus, e jθ = cos θ + j sin θ
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.5 DSP First 2e
(a) Real part of complex exponential is cosine.
cos(θ 1 + θ 2 ) = < e j (θ1 +θ2 ) = < e jθ1 e jθ2

( ) ( )
= < {(cos θ 1 + j sin θ 1 )(cos θ 2 + j sin θ 2 )}

= < {(cos θ 1 cos θ 2 − sin θ 1 sin θ 2 ) + j (sin θ 1 cos θ 2 + cos θ 1 sin θ 2 )}
cos(θ 1 + θ 2 ) = cos θ 1 cos θ 2 − sin θ 1 sin θ 2
(b) Change the sign of θ 2 .

( ) ( )
cos(θ 1 − θ 2 ) = < e j (θ1 −θ2 ) = < e jθ1 e−jθ2
= < {(cos θ 1 + j sin θ 1 )(cos θ 2 − j sin θ 2 )}
= < {(cos θ 1 cos θ 2 + sin θ 1 sin θ 2 ) + j (sin θ 1 cos θ 2 − cos θ 1 sin θ 2 )}
cos(θ 1 − θ 2 ) = cos θ 1 cos θ 2 + sin θ 1 sin θ 2
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.6 DSP First 2e
n
(cos θ + j sin θ) n = e jθ = e jnθ = cos(nθ) + j sin(nθ)
n 100 100
3
5 + j 45 = e j0.927 = e j0.295167π
= e j29.5167π
1
= e j1.5167π
e j28π
*
= cos(1.5167) + j sin(1.5167)
= 0.0525 − j0.9986
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.7 DSP First 2e
(a) 3e jπ/3 + 4e−jπ/6 = 5e j0.12 = 4.9641 + j0.5981

√ √ 10
(b) ( 3 − j3) 10 = 12e−jπ/3 = 248, 832 | {z } = −124, 416 + j215, 494.83
e−j10π/3
e+ j2π /3
√ √ −1 √
(c) ( 3 − j3) −1 = 12e−jπ/3 = (1/ 12)e+jπ/3 = 0.2887e+jπ/3 = 0.14434 + j0.25
√ √ 1/3
(d) ( 3 − j3) 1/3 = 12e−jπ/3 e j2π` = (12) 1/6 e−jπ/9 e j2π`/3 for ` = 0, 1, 2.
There are 3 answers:
1.513e−jπ/9 = 1.422 − j0.5175
1.513e−j7π/9 = −1.159 − j0.9726
1.513e−j13π/9 = 1.513e+j5π/9 = −0.2627 + j1.49
( ) ( ) ( ) √
(e) < je−jπ/3 = < e jπ/2 e−jπ/3 = < e jπ/6 = cos(π/6) = 3/2 = 0.866
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.8 DSP First 2e
The variable zz defines z(t), and xx defines x(t) = <{z(t)}.

z(t) = 15e j (2π (7)(t+0.875)) ⇒ x(t) = 15 cos(2π(7)(t + 0.875))
The period of x(t) is 1/7 = 0.1429, so the time interval −0.15 ≤ t ≤ 0.15 is (0.3)(7) = 2.1 periods.
There will be positive peaks of the cosine wave at t = −0.1607 s and t = −0.0179 s.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.9 DSP First 2e
A=9
T = 8 × 10−3 s ⇒ ω0 = 2000π/8 = 250π rad/s
t 1 = −3 × 10−3 s ⇒ ϕ = −2π(−3/8) = 3π/4 rad
z(t) = 9e j (250πt+0.75π) , X = 9e j0.75π , and x(t) = 9 cos(250πt + 0.75π)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.10 DSP First 2e
(a) Add complex amps: 3e−j2π/3 + 1 = 2.646e−j1.761 ⇒ x(t) = 2.646 cos(ω0 t − 1.761)
(b) x(t) = <{z(t)} = <{2.646e−j1.761 e jω0 t }
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.11 DSP First 2e
Add complex amps: e−jπ + e jπ/3 + 2e−jπ/3 = |

e−jπ + e jπ/3
{z + e } +e
−jπ/3 −jπ/3
= e−jπ/3
=0
⇒ x(t) = cos(ωt − π/3)
Here is the Matlab plot of the vectors.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.12 DSP First 2e
Find angles satisfying −π < θ ≤ π; all others are obtained by adding integer multiples of 2π.
<{(1 + j)e jθ } =0
√
<{ 2e jπ/4 e jθ } =0
√
<{ 2e j (θ+π/4) } =0
√
2 cos(θ + π/4) =0
√
 π/2
  π/4
  (1 + j)/ 2
⇒ θ + π/4 =  ⇒ θ= jθ
=

⇒ e √
 −π/2  −3π/4  (−1 − j)/ 2
  
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.13 DSP First 2e
Three periods of the signal will be 3(1/250) = 12 ms.
(a) Plot si (t) = <{ j s(t)} = <{0.8e jπ/2 e jπ/4 e j500πt } = 0.8 cos(2π(250)t + 3π/4).
0.8
0.4
si .t /
-0.4
-0.8
-3 -2 -1 0 1 2 3 4 -6 -5 -4
Time t (ms)
(b) Plot q(t) = <{ dt

d
s(t)} = <{0.8e jπ/4 ( j500π)e j500πt } = <{400πe j3π/4 e j500πt } = 400π cos(500πt + 3π/4)
1200
800
400
q.t /
0
-400
s
w
ng
la
hi
-800
)
n
ht
eb
c
tio
ea
rig
W
a
-1200
t
py
in
in
e
em
co
d
-3 -2 -1 0 1 2 3 4 -6 -5 -4
s
i
W
r
ss
to
es
Time t (ms) rld

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.14 DSP First 2e
√
(a) If z1 (t) = 5e−jπ/3 e j7t then x 1 (t) = <{z1 (t)}.
√
(b) If z2 (t) = 5e jπ e j7t then x 2 (t) = <{z2 (t)}.
√ √ √
(c) If z(t) = z1 (t) + z2 (t) = 5e j7t e−jπ/3 + e jπ = 5e−j2π/3 e j7t , then x(t) = <{z(t)} = 5 cos(7t − 2π/3).
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.15 DSP First 2e
Need to add complex amps: 2e j5 + 8e j9 + 4e j0 = 3.051e j2.673
Here is the plot of vectors representing the complex amplitudes:
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.16 DSP First 2e
t1 (−2) 4π π
(a) ϕ = −2π = −2π = = ⇒ True.
T 8 8 2
t1 3 3π
(b) ϕ = −2π = −2π = − ⇒ False.
T 8 4
(c) In this case a multiple of 2π must be added.
t1 7 −7π −7π π
ϕ = −2π = −2π = → + 2π = ⇒ True.
T 8 4 4 4
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.17 DSP First 2e
(a) Need to plot five vectors: {1, e j2π/5, e j4π/5, e j6π/5, e j8π/5 }.
Note: one is NOT missing; these are the five “5th roots of unity.”
4
X
(b) The sum is zero: x(t) = cos(ωt + 25 πk) = 0.
k=0
If the upper limit were 3 instead of 4,

X3 4
X
then x(t) = cos(ωt + 25 πk) = x(t) = cos(ωt + 52 πk) − cos(ωt + 85 π) = − cos(ωt + 58 π)
k=0 k=0
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.18 DSP First 2e
(a) Inverse Euler formula:

ω = 8 rad/s, A = 9/2, ϕ = −2π/3
(b) 30-60-90 triangle:

ω = 9 rad/s, ϕ = 0, A = 8.66
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.19 DSP First 2e
(a) 9e j0.5 = 3Ae j (−2+ϕ) + 4
(b) 9e j0.5 = 3 |{z}

Ae jϕ e−j2 + 4
z
9e j0.5 − 4
(c) z = = (1/3)e j2 9e j0.5 − 4 = 3e j2.5 − (4/3)e j2 = 1.938e j2.836
3e−j2
(d) A = 1.938 and ϕ = 2.836
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.20 DSP First 2e
(a) Convert to complex amplitudes (phasors):
1 = A1 e jϕ1 + A2 e jϕ2
e−jπ/2 = 2A1 e jϕ1 + A2 e jϕ2
(b) Write complex amplitudes as z1 and z2 :
1 = z1 + z2
e −jπ/2
= 2z1 + z2
√
(c) z1 = e−jπ/2 − 1 = 2e−j3π/4 and z2 = 2 − e−jπ/2 = 2.236e j0.464
√ √
(d) A1 = 2, ϕ1 = −0.75π rad, and A2 = 2.236 = 5, ϕ2 = 0.148π = 0.464 rad
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.21 DSP First 2e
(a) Convert to complex amplitudes (phasors):
e−j1 = 4e−jπ/2 A1 e jϕ1 + A2 e jϕ2

e−jπ/2+j2 = 3e−jπ/2 A1 e jϕ1 + A2 e jϕ2
e−j1 = 4e−jπ/2 z1 + z2
e−jπ/2+j2 = 3e−jπ/2 z1 + z2
z1 = 1.2576 − j0.3690 = 1.311e−j0.285

z2 = 2.0163 + j4.1890 = 4.649e j1.122
A1 = 1.311, ϕ1 = −0.285 rad

A2 = 4.649, ϕ2 = 1.122 rad
(b) Should plot − j4z1 + z2 and − j3z1 + z2 . Here is the Matlab plot of the vectors.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.22 DSP First 2e
Convert to phasors (complex amps): Me jπ/3 = 5e jψ − 4

The lefthand side is a ray from the origin at the angle of π/3 rad, or 60◦ when M > 0; and at −2π/3 when M < 0.
The righthand side is the set {z : z = 5e jψ − 4} which is a circle of radius 5 centered at z = −4 + j0. Since the origin is inside
the circle, there must be two solutions.
For M > 0, ray at π/3: M = 5e j (ψ−π/3) − 4e−jπ/3 must be purely real

√
0 = ={5e j (ψ−π/3) − 4e−jπ/3 } = 5 sin(ψ − π/3) − 4(− 3/2)
√
⇒ sin(ψ − π/3) = −2 3/5 ⇒ ψ − π/3 = −0.7654 ⇒ ψ = 0.2818 (or 16.1458◦ )
Then solve for M via : ={Me jπ/3 = 5e jψ − 4}

√ √
⇒ M ( 3/2) = 5 sin ψ ⇒ M = (10/ 3) sin ψ ⇒ M = 1.6056
For M < 0, ray at −2π/3: M = 5e j (ψ+2π/3) − 4e j2π/3 must be purely real

√
0 = ={5e j (ψ+2π/3) − 4e j2π/3 } = 5 sin(ψ + 2π/3) − 4( 3/2)
√
⇒ sin(ψ + 2π/3) = 2 3/5 ⇒ ψ + 2π/3 = 0.7654 ⇒ ψ = −1.329 (or −76.146◦ )
s
w
ng
la
hi
)
n
ht
Then solve for M via : ={Me−j2π/3 = 5e jψ − 4}
eb
c
tio
ea
rig
W
a
t
py
in
√ √
in
e
em
co
d
⇒ M (− 3/2) = 5 sin ψ ⇒ M = (−10/ 3) sin ψ ⇒ M = 5.6056
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
Another way to obtain M follows:

e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
Me jπ/3 = 5e jψ − 4
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
⇒ Me jπ/3 + 4 = 5e jψ
k
k
es
so
or
or
is
w
s
w
ed
k
as
or
⇒ |Me jπ/3 + 4| 2 = |5e jψ | 2 = 25

id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
M 2 + 8M cos(π/3) + 16 = 25
g
ur
d
te
y
an
co
an
in
r
M 2 + 4M − 9 = 0 which has two roots: M = 5.6056 and M = 1.6056.

of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.23 DSP First 2e
(a) z(t − 0.24) = Ze j10π (t−0.24) = 7e j0.3π e j10πt e−j2.4π = |

7e−j2.1π
{z } e
j10πt = 7e−j0.1π e j10πt
| {z }
W
(b) z(t − t d ) = Ze j10π (t−td ) = 7e j0.3π e j10πt e−j10πtd must equal y(t) = Y e j10πt = 7e−j0.1π e j10πt
⇒ 7e j0.3π−j10πtd = 7e−j0.1π ⇒ 0.3π − 10πt d = −0.1π ⇒ t d = (0.4/10) = 0.04 s
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.24 DSP First 2e
(a) The frequency is the same for all terms, so ω̂0 = 0.22π rad in the expression for y[n].
(b) Perform phasor addition:
y[n] = 7e j (0.22π (n+1)−0.25π) − 14e j (0.22πn−0.25π) + 7e j (0.22π (n−1)−0.25π)

= 7e j (0.22πn−0.03π) − 14e j (0.22πn−0.25π) + 7e j (0.22πn−0.47π)

= 7e−j0.03π − 14e−j0.25π + 7e−j0.47π e j0.22πn
| {z }
Phasor Addition
= 3.213 e j0.75π j0.22πn
e ⇒ A = 3.213, ϕ = 0.75π rad
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.25 DSP First 2e
(a) d
dt z(t) = d
dt Ze j2πt = ( j2π) Z e j2πt ⇒ Q = ( j2π)(e jπ/4 ) = 2πe j3π/4
| {z }
Q
(b) Need a plot. Angle of Q is greater by π/2 rad.
(c) Compare the interchange of derivative and real part, which is always true.
( )
< dt d
z(t) = <{2πe j3π/4 e j2πt } = 2π cos(2πt + 3π/4)
d d ( ) d
<{z(t)} = < e jπ/4 e j2πt = cos(2πt + π/4) = (2π)(− sin(2πt + π/4)) = 2π cos(2πt + 3π/4)
dt dt dt
(d) Integrating a complex exponential over one period should give zero.
Z0.5 0.5
e jπ/4 e j2πt e jπ − e−jπ
e jπ/4 j2πt
e dt = = e jπ/4 =0
j2π −0.5 j2π
−0.5
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.26 DSP First 2e
Try x(t) = Ae jωt and solve for ω
d d2
x(t) = jω Ae jωt and x(t) = ( jω) 2 Ae jωt
dt dt 2 |{z}
−ω 2
Plug x(t) into differential equation
−ω2 Ae jωt = −289 Ae jωt
⇒ −ω2 = −289 ⇒ ω = ±17
Two solutions: x(t) = Ae j17t
or x(t) = Ae−j17t
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.27 DSP First 2e
d 2 v(t) d 2 v(t)
!
d d dv 1
(a) v(t) = −L i(t) = −L C = −LC 2
⇒ 2
=− v(t)
dt dt dt dt dt LC
1
(b) The frequency of oscillation will be ω0 = √
LC
d 2 v(t) 1
(c) Starting with v(t) = A cos(ω0 t + ϕ), we obtain 2
= − ω02 A cos(ω0 t + ϕ) = − v(t)
dt |{z} | {z } LC
1/LC v(t)
dt = 5Cω0 sin(ω0 t + π/3) = 5Cω0 cos(ω0 t + π/3 − π/2)

(d) v(t) = 5 cos(ω0 t + π/3) ⇒ i(t) = C dv
There is a 90◦ phase difference between the current and the voltage.
(e) This is true in general:

d 2i(t) d 2i(t)
!
d d di 1
i(t) = C v(t) = C −L = −LC 2
⇒ 2
=− i(t)
dt dt dt dt dt LC
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-2.28 DSP First 2e
In a mobile radio system a transmitting tower sends a sinusoidal signal, and a mobile user receives not one but two
copies of the transmitted signal: a direct-path transmission and a reflected-path signal (e.g., from a large building) as
depicted in the following figure.
y
TRANS- REFLECTOR
MITTER (0, dt ) (dr , dt )
Velocity = c
VEHICLE
0 x
The received signal is the sum of the two copies, and since they travel different distances they have different time delays,
i.e.,
s
w
ng
la
hi
r (t) = s(t − t 1 ) + s(t − t 2 )
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
The distance between the mobile user in the vehicle at x and the transmitting tower is always changing. Suppose that the
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
direct-path distance is
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
q
en
g
he
d 1 = x 2 + d t2 (meters)
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
where d t = 1000 meters, and where x is the position of the vehicle moving along the x-axis. Assume that the reflected-path
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
distance is
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
q
g
ur
d
te
d 2 = dr + (x − dr ) 2 + d t2 (meters)
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
where dr = 55 meters.
st
or
de
ill
w
(a) The amount of the delay (in seconds) can be computed for both propagation paths, by converting distance into time
delay by dividing by the speed of light (c = 3×108 m/s).
q
x 2 + d t2
p
x 2 + 106
t 1 = d 1 /c = = secs.
c 3×108
q q
dr + (x − dr ) + d t
2 2
55 + (x − 55) 2 + 106
t 2 = d 2 /c = = secs.
c 3×108
(b) When the transmitted signal is s(t) = cos(300π×106 t), the general formula for the received signal is:
r (t) = s(t − t 1 ) + s(t − t 2 ) = cos(300π×106 (t − t 1 )) + cos(300π×106 (t − t 2 ))

When x = 0 we can calculate t 1 and t 2 , and then perform a phasor addition to express r (t) as a sinusoid with a known
amplitude, phase, and frequency. When x = 0, the time delays are
p
02 + 106
t1 = = 3.3333×10−6 secs.
3×108
q
55 + (0 − 55) 2 + 106
t2 = = 3.5217×10−6 secs.
3×108
Thus we must perform the following addition:
r (t) = cos(300π×106 (t − 3.3333×10−6 )) + cos(300π×106 (t − 3.5217×10−6 ))

= cos(300π×106 t − 1000π)) + cos(300π×106 t − 1056.5113579π)
As a phasor addition, we carry out the following steps (since 1000π and 1056π are integer multiples of 2π):
R = 1e j0 + 1e j0.5113579π
= 1 + j0 + (−0.035674 + j0.99936)
= 0.9643 + j0.9994 = 1.389e j0.803 = 1.389e j0.256π = 1.389∠46.02◦
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
From the polar form of the phasor R, we can write r (t) as a sinusoid:
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
r (t) = 1.389 cos(300π×106 t + 0.256π) rld

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
(c) In order to find the locations where the signal strength is zero, we note that the phase angles of the two delayed
t
by
en
g
he
no
in
ud
d
ud
sinusoids must differ by an odd multiple of π in order to get cancellation. Thus,

rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
(2` + 1)π = ∆ϕ = −ωt 1 − (−ωt 2 )

w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
q
g
ur
d
(x − 55) 2 + 10655 +
p
te
y
an
+
co
x2 106
an
in
= −300π×10 . 6*
r
+/
e
of
−
ei
th
.
th
3×108 3×108
y
l
/
sa
ro
st
or
de
, -
ill
q !
w
p
= −π x + 10 − 55 − (x − 55) + 10
2 6 2 6
The general solution to this equation is difficult, involving a quartic. However, if we choose ` = 27 so that the left hand
side becomes 55π, then the 55π term on the right hand side will cancel, and we obtain an equation in which squaring
both sides will produce the answer.
p q
π x + 10 = −π (x − 55) 2 + 106
2 6
=⇒ x 2 + 106 = (x − 55) 2 + 106

=⇒ x 2 = x 2 − 110x + 552
=⇒ 110x = 552
!
55
=⇒ x = 55 = 27.5 meters
110

The general solution would be done in the following manner:

p q
−(2` + 1) = x 2 + 106 − 55 − (x − 55) 2 + 106
p q
⇒ 55 − (2` + 1) = x 2 + 106 − (x − 55) 2 + 106
p q
⇒ 552 − 110(2` + 1) + (2` + 1) 2 = x 2 + 106 − 2 x 2 + 106 (x − 55) 2 + 106 + (x − 55) 2 + 106
p q
⇒ 2 x + 10 (x − 55) 2 + 106 = −4` 2 + 216` + 109 − 552 + x 2 + 2×106 + (x − 55) 2
2 6
Squaring both sides would eliminate the square roots, but would produce a fourth-degree polynomial that would have
to be solved for the vehicle position x.
(d) Here is a Matlab script that will plot the signal strength versus vehicle position x, thus demonstrating that there are
numerous locations where no signal is received (note the null at x = 27.5).
xx = -100:0.05:100;
d1 = sqrt(xx.*xx + 1e6);
d2 = 55 + sqrt((xx-55).*(xx-55)+1e6);
omeg = 300e6*pi; c = 3e8;
phi1 = -omeg*d1/c;
s
phi2 = -omeg*d2/c; w
ng
la
hi
)
n
ht
eb
c
RR = 1*exp(j*phi1) + 1*exp(j*phi2);
tio
ea
rig
W
a
t
py
in
in
subplot(’Position’,[0.1,0.1,0.6,0.3]);
e
em
co
d
s
i
W
r
ss
to
es
hp = plot(xx,abs(RR)); grid on,

rld
c
i
at
.D
ru
o
St
.
st
xlabel(’Vehicle Position (x)’);

ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ylabel(’Signal Strength’);
ni
er
e
le
on
U
us
tp
t
by
en
title(’Multipath Problem in SP-First’);

g
he
no
in
ud
d
ud
rt
te
is
set(hp,’LineWidth’,2);
st
c
fo
cl
d
te
in
an
y
n
o
(
l
print -dpdf multipathResult.pdf

le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
Multipath Problem in SP−First

an
of
of
is
pr
Th
rt
y
se
rit
is
pa
2
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
1.5
Signal Strength
ill
w
0.5
0
−100 −80 −60 −40 −20 0 20 40 60 80 100
Vehicle Position (x)
Over the range −100 ≤ x ≤ 100 the nulls appear to be equally spaced 36.4 meters apart, but they are not uniform. A
plot over the range 0 ≤ x ≤ 1500 would demonstrate the non-uniformity.

Chapter 3
Spectrum
3-1 Problem Solutions
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
32
CHAPTER 3. SPECTRUM
P-3.1 DSP First 2e
(a) x(t) = 11 + 14 cos(100πt − π/3) + 8 cos(350πt − π/2)
(b) Since the gcd of 50 and 175 is 25, x(t) is periodic with period T0 = 1/25 = 0.04 s.
(c) Negative frequencies are implicit in the cosine terms. They are needed to give a real signal when combined with their
corresponding positive-frequency terms.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.2 DSP First 2e
(a) A single plot labeled with complex amplitudes is sufficient. The spectrum consists of the lines
{(400, 5e jπ/4 ), (−400, 5e−jπ/4 ), (600, 3.5e−jπ/3 ), (−600, 3.5e jπ/3 ), (800, 1.5), (−800, 1.5)}
where the frequencies are in Hz.
5e−jπ/4 5e jπ/4
3.5e jπ/3 3.5e−jπ/3
1.5 1.5
-
−800 −600 −400 0 400 600 800 f
(b) The signal x(t) is periodic with fundamental frequency 200 Hz or period 1/200 = 0.005 s since the gcd of {400, 600, 800}
is 200.
(c) The spectrum has the added components {(500, 2.5e jπ/2 ), (500, 2.5e−jπ/2 )}. Now we seek the gcd of {400, 500, 600, 800}
so the fundamental frequency changes to 100 Hz and the period is 0.01 s.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.3 DSP First 2e
(a) x(t) = 2 cos(2t) + 4 cos(4t − π/3) + 2 cos(6t + π/4)
(b) The spectrum is {(2, 1), (−2, 1), (4, 2e−jπ/3 ), (−4, 2e jπ/3 ), (6, e jπ/4 ), (−6, e−jπ/4 )}
The frequencies are all in rad/s.
2e jπ/3 2e−jπ/3
e−jπ/4 1 1 e jπ/4
ω
-
−6 −4 −2 0 2 4 6
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.4 DSP First 2e
(a) Determine a formula for x(t) as the real part of a sum of complex exponentials.
Use Euler’s formula for the sine function obtaining

!3
e j27πt − e−j27πt
sin (27πt) =
3
2j
1 j27πt3
= e − 3e j27πt2 e−j27πt + 3e j27πt e−j27πt2 − e−j27πt3
−8 j
3 1
= sin(27πt) − sin(81πt)
4 4
(b) What is the fundamental period for x(t)?
The fundamental frequency is 27/2 so the fundamental period is 2/27.
(c) Plot the spectrum for x(t).
s
The spectrum is {(27π, − j3/8), (−27π, j3/8), (81π, j/8), (−81π, − j/8)}, where the frequencies are in rad/s. w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
j3/8 − j3/8
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
− j/8 j/8
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
ω
-
g
in
an
y
−81π −27π
n
0 27π 81π
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.5 DSP First 2e
There are seven spectral components:

{(−3, 1/(1 − j3)), (−2, 1/(1 − j2)), (−1, 1/(1 − j)), (0, 1), (1, 1/(1 + j)), (2, 1/(1 + j2)), (3, 1/(1 + j3))},
where the frequencies are all in rad/s.
Putting all the complex numbers in polar form gives the following plot:
1
0.707e j0.25π 0.707e−j0.25π
0.447e j0.3524π 0.447e−j0.3524π
0.316e j0.398π 0.316e−j0.398π
ω
-
−3 −2 −1 0 1 2 3
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.6 DSP First 2e
(a) In this case we need to find the gcd of 36 and 84, which is 12. Thus, the fundamental frequency is ω0 = 1.2π rad/s.
(b) The fundamental period is T0 = 2π/ω0 = 1/0.6 = 5/3 s.
(c) The DC value is −7.
(d) The ak coefficients are nonzero for k = 0, ±3, ±7.

Here is the list of the nonzero Fourier series coefficients in a table.
k −7 −3 0 3 7
ak 3e−jπ/4 4e jπ/3 7e jπ 4e−jπ/3 3e jπ/4
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.7 DSP First 2e
(a) The phasor representation is z(t) = Ae j2π ( fc − f∆ )t + Be j2π ( fc + f∆ )t
(b)
z(t) = e j2π fc t ( Ae−j2π f∆ t + Be j2π f∆ t )

= e j2π fc t ( A cos(2π f ∆ t) − j A sin(2π f ∆ t) + B cos(2π f ∆ t) − j B sin(2π f ∆ t))
= e j2π fc t [( A + B) cos(2π f ∆ t) − j ( A − B) sin(2π f ∆ t)]
Therefore, the real part is
x(t) = <{z(t)} = ( A + B) cos(2π f ∆ t) cos(2π f c t) + ( A − B) sin(2π f ∆ t) sin(2π f c t)
so C = A + B and D = A − B. If A = B = 1, C = 2 and D = 0, so using the trigonometric identity cos α cos β =

2 cos(α − β) + 2 cos(α + β), it follows that
1 1
x(t) = 2 cos(2π f ∆ t) cos(2π f c t) = 2[ 21 cos(2π( f c − f ∆ )t) + 21 cos(2π( f c + f ∆ )t)]
(c) The values are A = 1 and B = −1. In this case,

s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
e j2π f∆ t − e−j2π f∆ t e j2π fc t − e−j2π fc t
! ! t
py
in
in
e
x(t) = 2 em
co
d
s
i
W
r
ss
to
es
2j 2j rld
c
i
at
.D
ru
o
St
.
st
ed
ng
j2π( fc + f∆ )t
− e−j2π ( fc − f∆ )t + e−j2π ( fc + f∆ )t

= −2 e
1 j2π ( fc − f∆ )t
in
d
−e itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
The spectrum is {(− f c − f ∆, −0.5), (− f c + f ∆, 0.5), ( f c − f ∆, 0.5), ( f c + f ∆, −0.5)}, and the plot is
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
−0.5 0.5 0.5 −0.5

le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
−( f c + f ∆ ) ( f c + f∆)
-
− fc −( f c − f ∆ ) 0 ( f c − f∆) fc f
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.8 DSP First 2e
(a) Using Euler’s relation we get
x(t) = 10 + 10e jπ/4 e j2π (100)t + 10e−jπ/4 e−j2π(100)t + 5e j2π (250)t + 5e−j2π (250)t
The gcd of 100 and 250 is 50 so f 0 = 50 and therefore N = 5. The nonzero Fourier coefficients are, therefore, a−5 = 5,
a−2 = 10e−jπ/4 , a0 = 10, a2 = 10e jπ/4 , and a5 = 5.
(b) The signal is periodic because all the frequencies are multiples of 50 Hz. Therefore, the fundamental period is
T0 = 1/50 = 0.02s.
(c) Here is the spectrum plot of this signal versus f in Hz.
10e−jπ/4 10 10e jπ/4
5 5
s
w
ng
la
hi
)
n
-
ht
eb
c
f
tio
−250 −100 0 100 ea 250
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.9 DSP First 2e
(a) Use phasors to show that x(t) can be expressed in the form
x(t) = A1 cos(ω1 t + ϕ1 ) + A2 cos(ω2 t + ϕ2 ) + A3 cos(ω3 t + ϕ3 )
where ω1 < ω2 < ω3 ; i.e., find values of the parameters A1, A2, A3, ϕ1, ϕ2, ϕ3, ω1, ω2, ω3 .
Using Euler’s relation we get

f g
x(t) = 12 − 3.5 je−jπ/3 e jπt + 3.5 je jπ/3 e−jπt 0.5e j13πt + .5e−j13πt
= 12 cos(13πt) − 1.75 je−jπ/3 (e j14πt + e−j12πt ) + 1.75 je jπ/3 (e j12πt + e−j14πt )
= 12 cos(13πt) + 1.75e−jπ/2 e−jπ/3 (e j14πt + e−j12πt ) + 1.75e jπ/2 e jπ/3 (e j12πt + e−j14πt )
= 12 cos(13πt) + 1.75e−j5π/6 e j14πt + 1.75e−j5π/6 e−j12πt + 1.75e j5π/6 e j12πt + 1.75e j5π/6 e−j14πt
= 12 cos(13πt) + 3.5 cos(14πt − 5π/6) + 3.5 cos(12πt + 5π/6)
The requested parameters are easily picked off from this equation.
(b) Sketch the two-sided spectrum of this signal on a frequency axis. Be sure to label important features of the plot. Label
s
w
ng
your plot in terms of the numerical values of Ai , ϕi , and ωi .
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
12 12
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
3.5e j5π/6 3.5e−j5π/6 3.5e j5π/6 3.5e−j5π/6

te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
ω
-
of
of
is
−14π −13π −12π 0 12π 13π 14π

pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.10 DSP First 2e
(a) Assume without limitation that ω2 − ω1 > 0. For periodicity with period T0 we require that ω0 = 2π/T0 . This means
that k 1 ω0 = ω2 − ω1 and k 2 ω0 = ω2 + ω1 , where k 1 and k 2 are integers and k 2 > k 1 .
(b) Part (a) gives two equations for ω1 and ω2 . If we solve them in terms of ω0 we get ω1 = (k2 − k 1 )ω0 /2 and
ω2 = (k 2 + k 1 )ω0 /2, so the main condition is that both ω1 and ω2 are integer multiples of ω0 /2. This is the most
general condition.
Therefore, the relationship between ω2 and ω1 is
k2 + k1
ω2 = ω1
k2 − k1
if x(t + T0 ) = x(t). Thus, ω2 could be an integer multiple of ω1 if k 2 − k 1 divides into k 2 + k 1 with no remainder, but
that is not necessary for periodicity of x(t).
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.11 DSP First 2e
(a) The gcd of 40, 60, 120 is 20 so ω0 = 20π and the fundamental period is T0 = 2π/ω0 = 0.1s. The finite Fourier
series has components indexed by 0, ±2, ±3, ±6 so N = 6. The coefficients are a0 = 2, a±2 = 2e∓jπ/5 , a±3 = 1.5e∓jπ/2 ,
a±6 = 2e∓π/3
(b) The spectrum is

{(−120π, 2e jπ/3 ), (−60π, 1.5e jπ/2 ), (−40π, 2e jπ/5 ), . . .
(0, 2), (40π, 2e−jπ/5 ), (60π, 1.5e−jπ/2 ), (120π, 2e−jπ/3 )}
2e jπ/3 2e jπ/5 2 2e−jπ/5 2e−jπ/3

1.5e jπ/2 1.5e−jπ/2
ω
-
−120π −60π −40π 0 40π 60π 120π
s
w
ng
la
hi
)
n
ht
(c) Now the fundamental frequency is 10π rad/s because the gcd of 20, 40, 50, and 120 is 10. Therefore, the period is
eb
c
tio
ea
rig
W
a
t
T0 = 2π/10π = 1/5 = 0.2s. The spectrum is the same as in part (b) except there are two additional components at
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
±50π rad/s: (−50π, 5e jπ/6 ) and (50π, 5e−jπ/6 ). rld

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.12 DSP First 2e
(a) Make a table of the frequencies of the tones of the octave beginning with middle C, assuming that the A above middle
C is tuned to 440 Hz.
Note name C C# D E[ E F F#
Note number 40 41 42 43 44 45 46
Frequency 262 277 294 311 330 349 370
Note name F# G G# A B[ B C
Note number 46 47 48 49 50 51 52
Frequency 370 392 415 440 466 494 523
(b) The formula for the frequency f as a function of note number n is
f = 440 · 2(n−49)/12
(c) The spectrum would have the form:

{(−440, a3∗ ), (−370, a2∗ ), (−294, a1∗ ), (294, a1 ), (370, a2 ), (440, a3 )}
To sound like a musical chord, the coefficients should have similar magnitudes, but the phases could be arbitrarily
chosen. A chord from a real instrument would have overtones (higher harmonics) of each individual note.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.13 DSP First 2e
(a) The frequency of the DC component is by definition 0. The waveform is periodic with period 25 ms so the frequency
is 1/0.025 = 40 Hz.
(b) The DC level is (20 − 10)/2 = 5, the amplitude of the cosine is (20 + 10)/2 = 15, and the cosine is delayed by 0.005 s,
so
x(t) = 5 + 15 cos(2π(40)(t − .005)) = 5 + 15 cos(80πt − 0.4π)
(c) x(t) = 5 + 7.5e j (80πt−0.4π) + 7.5e−j (80πt−0.4π) = 7.5e j0.4π e−j80πt + 5 + 7.5e−j0.4π e j80πt
(d) Plot of the two-sided spectrum of the signal x(t).

7.5e j0.4π 7.5e−j0.4π
5
ω
-
−80π 0 80π
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.14 DSP First 2e
(a) Using symmetry we obtain

√ √
X−1 = 2 − j 2 = 2e−jπ/4 X2 = 8e jπ/3 ω1 = 70π ω2 = −100π
(b) x(t) = 20 + 4 cos(70πt + π/4) + 16 cos(100πt + π/3)
(c) The gcd of 70 and 100 is 10, so the fundamental frequency of the signal is f 0 = 5 Hz and the fundamental period is
T0 = 1/5 = 0.2 s.
(d) Note that the −20 ≤ 4 cos(70πt + π/4) +16 cos(100πt + π/3) ≤ 20 since the individual terms satisfy −4 ≤ 4 cos(70πt +
π/4) ≤ 4 and −16 ≤ 16 cos(100πt + π/3) ≤ 16. The value ±20 would be attained only if the phases of the two cosines
are such that 4 cos(70π(t − t 0 )) + 16 cos(100π(t − t 0 )).
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.15 DSP First 2e
(a) x(t) = cos2 (7πt − 0.1π)

We need to express x(t) in terms of complex exponentials
2
x(t) = 0.5e j (7πt−0.1π) + 0.5e−j (7πt−0.1π) = 0.25e j0.2π e−j14πt + 0.5 + 0.25e−j0.2π e j14πt
0.5
0.25e j0.2π 0.25e−j0.2π
ω
-
−14π 0 14π
(b) y(t) = cos2 (7πt − 0.1π) cos(77πt + 0.1π)

x(t) = 0.25e j0.2π e−j14πt + 0.5 + 0.25e−j0.2π e j14πt 0.5e j0.1π e j77πt + 0.5e−j0.1π e−j77πt
s
w
ng
la
= 0.125e j0.1π e−j91πt + 0.25e−j0.1π e−j77πt + 0.125e−j0.3π e−j63πt
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
+ 0.125e j0.3π e j63πt + 0.25e j0.1π e j77πt + 0.125e−j0.1π e j91πt
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
0.25e−j0.1π 0.25e j0.1π

t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
0.125e j0.1π 0.125e−j0.3π 0.125e j0.3π 0.25e−j0.1π

es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
ω
an
-
in
−91π −77π −63π 0 63π 77π 91π

r
e
of
ei
th
th
y
l
sa
ro
Note: ω axis is
st
or
de
ill
not to scale.
w

CHAPTER 3. SPECTRUM
P-3.16 DSP First 2e
(a) Plotting spectrum of v(t) versus f in kHz,
0.5e−j0.3π 0.5e j0.3π
-
−3 0 3 f (kHz)
(b) The spectrum for x(t) versus f in kHz.
0.75 0.75
0.25e−j0.3π 0.25e j0.3pi 0.25e−j0.3π 0.25e j0.3π
s
w
ng
-
−683 −680 −677 0 677 680 683 f (kHz)
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
Note: f axis is not to scale. rld

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.17 DSP First 2e
(a) The gcd of 105 and 180 is 15, so the given frequencies are the 7th and 12th harmonics of f 0 = 15 Hz.
(b) x(t) = 22 cos(2π(105)t − 0.4π) + 14 cos(2π(180)t − 0.6π)
(c) Simplify the numerical values for the complex amplitudes, i.e., phases should be in [−π, π].
x 2 (t) = 22 cos(2π(105)(t − 0.05) − 0.4π) + 14 cos(2π(180)(t − 0.05) − 0.6π)

= 22 cos(2π(105)t − 10.5π − 0.4π) + 14 cos(2π(180)t − 18π − 0.6π)
= 22 cos(2π(105)t − 10π − 0.9π) + 14 cos(2π(180)t − 18π − 0.6π)
= 22 cos(2π(105)t − 0.9π) + 14 cos(2π(180)t − 0.6π)
Note that even multiples of 2π rad can be dropped from the equation. Thus, the spectrum is:
{(−180, 7e j0.6π ), (−105, 11e j0.9π ), (105, 11e−j0.9π ), (180, 7e−j0.6π )} where the frequencies are in hertz. Therefore the
plot of the spectrum looks just like Fig. ?? except the phase is different at frequencies ±105.
(d) The effect of this operation is simply to increase all the frequencies by 105 Hz, or, in other words, to shift the spectrum
of x(t) to the right by 105 Hz.
Therefore the spectrum line at −105 Hz will move to f = 0. The DC component is therefore, the value of the spectrum
s
w
ng
la
at f = −105, or 11e j0.9π .
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.18 DSP First 2e
(a) The spectrum of y(t) is the spectrum of x(t) with an added DC component of size 8.
11e j0.4π 11e−j0.4π

8
7e j0.6π 7e−j0.6π
-
−180 −105 0 105 180 f
(b) The spectrum of z(t) is the same as that of x(t) with the addition of components of size 9e±j0.8π at frequencies ±40
Hz.
11e j0.4π 11e−j0.4π

9e−j0.8π 9e j0.8π
7e j0.6π 7e−j0.6π
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
-
co
d
−180 −105 −40 0 40 105 180 f
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
(c) The fundamental frequency is the gcd of 40, 105, 180, which is 5 Hz.
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
(d) The derivative operation multiplies each spectrum component by j2π f , where f is the frequency of the complex
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
exponential component. So we get

as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
− j2π(105)11e j0.4π j2π(105)11e−j0.4π

se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
− j2π(180)7e j0.6π j2π(180)7e−j0.6π

th
th
y
l
sa
ro
st
or
de
ill
w
-
−180 −105 0 105 180 f

CHAPTER 3. SPECTRUM
P-3.19 DSP First 2e
(a)
X1 = 8e−jπ/3 and ω1 = −10π
(b) Here is the plot of the spectrum of x(t).
8e jπ/3 8e−jπ/3
ω
-
−10π 0 10π
(c) The symmetry implies that ωb = 20π and B = +4 j. Furthermore, symmetry requires that ω a = 0. To find A, ωc ,
and ϕ we can write y(t) as y(t) = 0.5x(t)e jϕ e jωc t + 0.5x(t)e−jϕ e−jωc t , which shows that the spectrum of y(t) will
consist of the sum of scaled copies of the spectrum of x(t) shifted right (up) by ωc and left (down) by ωc . In or-
der to have only three components we must choose ωc = 10π so that two of the shifted spectrum lines over lap at ω = 0.
s
w
ng
la
4e jπ/3 e jϕ + 4e−jπ/3 e−jϕ
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
4e jπ/3 e−jϕ 4e−jπ/3 e jϕ
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ω
ud
rt
-
te
is
−20π
st
0 20π
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
jϕ = 4e−jπ/2 so ϕ = −π/6. Finally, note that the DC value can be written as A = 8 cos(π/3 + ϕ) =
w
Now, 4e−jπ/3 e √
ed
k
as
or
√
id
hi
th
w
d
ov
t
an
8 cos(π/6) = 8 3/2 = 4 3.
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.20 DSP First 2e
(a) The gcd of 40 and 90 is 10, so f 0 = 10 Hz.
(b) The fundamental period is T0 = 1/ f 0 = 1/10 = 0.1 s.
(c) From the plot, the DC value is 0.5.
(d) With f 0 = 10, the harmonics are k = 0, ±4, ±9.
k −9 −4 0 4 9
ak 0.4e−j2 0.6e j1.4 0.5 0.6e−j1.4 0.4e j2
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.21 DSP First 2e
dψ
(a) The instantaneous frequency is ωi (t) = dt = 2αt + β, so ω1 = ωi (0) = β and ω2 = ωi (T2 ) = 2αT2 + β.
(b) The instantaneous frequency versus time is ωi (t) = 80t + 27
(c) Here is the plot of the instantaneous frequency (in Hz) versus time over the range 0 ≤ t ≤ 1 sec.
ωi (t)
f i (t) = 2π
6
107
2π

27
2π
-
0 1 t
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.22 DSP First 2e
(a) The general form for the chirp signal is x(t) = cos(αt 2 + βt + ϕ). The instantaneous frequency of this signal is
ωi (t) = 2αt + β. From this we observe that ω1 = 2π f 1 = 2π(4800) = ωi (0) = β. To obtain α, we note that ω2 =
2π(800) = ωi (2) = 2α(2) + β = 4α + 9600π so α = −2000π. Therefore, the signal is
x(t) = cos(−2000πt 2 + 9600πt + ϕ)
where ϕ is an arbitrary phase constant.
(b) The instantaneous frequency is ωi = 800πt + 500π, so ω1 = ωi (0) = 500π and ω2 = ωi (3) = 800π(3) + 500π =
2900π rad/s.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.23 DSP First 2e
(a) The instantaneous frequency is ωi (t) = 2αt + β. Substituting the given parameters gives α = 4π and β = 2π, so the
signal with the given parameters is x(t) = cos(4πt 2 + 2πt + ϕ).
(b–f) The solution to this problem is given in the following figure. Note that the times at which f i (t) is equal to 4 Hz and 8
Hz are indicated with dashed lines. Careful scrutiny of the plots confirms that the waveform of the chirp signal does
match the waveforms of the 4 Hz and 8 Hz constant-frequency sinusoids at the corresponding two times.
Instantaneous Frequency of Chirp Signal
frequency in Hz
10
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
4 Hz Signal
1
0.5
0
-0.5
-1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Chirp Signal
1
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
0
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
-1
at
.D
ru
o
St
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
.
st
ed
ng
W
in
d
itt
ni
8 Hz Signal
te
of
th
m
ar
ni
1
er
e
le
on
U
us
tp
t
by
en
0.5
g
he
no
in
ud
d
ud
rt
te
is
0
st
c
fo
cl
d
te
in
an
y
n
o
-0.5
(
l
le
si
pr
k
es
so
or
or
is
-1
w
s
w
ed
k
as
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

or
e
id
hi
th
w
time in seconds
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.24 DSP First 2e
1
(a) f 1 (t) = √ (c) f 3 (t) = e2t /π
2π (t)
(b) f 2 (t) = t/π (d) f 4 (t) = − sin(2πt)
Solution to Problem 3.24

0.8
frequency in Hz
0.6
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4
frequency in Hz
1
0.5
0
-0.5
-1
-3 -2 -1 0 1 2 3
frequency in Hz
150
s
w
ng
la
100
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
50
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
0
ru
o
St
.
st
ed
ng
W
0 0.5 1 1.5 2 2.5 3
in
d
itt
ni
e
te
of
th
m
ar
ni
frequency in Hz
1
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
0.5
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
0
g
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
-0.5
is
w
s
w
ed
k
as
or
e
id
hi
th
w
-1
d
ov
t
an
of
of
is
0 0.5 1 1.5 2 2.5 3

pr
Th
rt
y
se
rit
is
time in seconds
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.25 DSP First 2e
∞
X
(a) Let x(t) be given by the Fourier series x(t) = ak e j (2π/T0 )kt . Then it follows that
k=−∞
∞
X :1 X∞

x(0) = e j
ak (2π/T
0 )k (0)
= ak .
k=−∞ k=−∞
(b) Let f 3 = − f 2 = f 0 and f 4 = − f 1 = 3 f 0 so that from the spectrum we can write
x(t) = 12 cos(2π f 0 t + π/4) + 4 cos(6π f 0 t + 3π/4)

√ √ √
Therefore x(0) = 12 cos(π/4) + 4 cos(3π/4) = 6 2 − 2 2 = 4 2. Now if we add the coefficients of the Fourier series
we get
√
a1 + a2 + a3 + a4 = 2e−j3π/4 + 6e−jπ/4 + 6e jπ/4 + 2e j3π/4 = 12 cos(π/4) + 4 cos(3π/4) = 4 2
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.26 DSP First 2e
The equations corresponding to the spectra are:
The matches are (a) (3) x 1 (t) = 4 cos(4πt + π) + 4 cos(6πt + π/2)

(b) (1) x 2 (t) = 2 cos(4πt + π/4) + 4 cos(6πt − 0.333π)
(c) (2)
x 3 (t) = −3 + 2 cos(4πt + π/4)
(d) (5)
(e) (4) x 4 (t) = −2 + 4 cos(4πt + π)
x 5 (t) = 4 cos(2πt + π) + 4 cos(4πt + π)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.27 DSP First 2e
(a) x(t) = cos(−250πt 2 ) (d) x(t) = cos(100πt) cos(400πt)

Spectrogram (2) Spectrogram (1)
(b) x(t) = cos(100πt − π/4) + cos(400πt) (e) x(t) = cos(200πt 2 )
(c) x(t) = cos(1000πt − 250πt 2 ) (f) x(t) = cos(30e2t )
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

Chapter 4
Sampling and Aliasing
4-1 Problems
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
60
CHAPTER 4. SAMPLING AND ALIASING
P-4.1 DSP First 2e
(a) A = 11, ϕ = −π/3, and ω/ f s = 7π/9 = ω̂1 .
(b) A = 11, ϕ = +π/3, and ω/ f s = 7π/6 → ω̂1 = 5π/6.
(c) A = 11, ϕ = −π/3, and ω/ f s = 7π/3 → ω̂1 = π/3.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.2 DSP First 2e
x 1 [n] = x 3 [n] and x 2 [n] = x 4 [n] = x 5 [n]
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.3 DSP First 2e
(a) y(t) = 8 cos(2160πt − π/3)
(b) y(t) = 4 cos(2880πt + π/4)
(c) y(t) = 2 cos(1440πt − π/5)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.4 DSP First 2e
(a) Sketch will have four spectral lines at ω̂ = ±4000π, ±12000π. The complex amplitude of each spectral line will be
0.25.
6
0.25 0.25 0.25 0.25
ω
-
−12000π −4000π 0 4000π 12000π
(b) The Nyquist rate for x(t) is 12,000 samples/s.
(c) The Nyquist rate for r (t) is 15 × 106 samples/s.
(d) The Nyquist rate for v(t) is 7 × 106 samples/s.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.5 DSP First 2e
(a) The Nyquist rate for x(t) is 50 samples/s.
(b) The Nyquist rate for r (t) is 72 samples/s.
(c) The Nyquist rate for v(t) is 45 samples/s.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.6 DSP First 2e
(a) Sketch will have four spectral lines: at ω̂ = ±0.2π with complex amplitudes equal to 13e±jπ/4 , and at ±0.5π with
complex amplitudes of 7e±j3π/4 .
There is no aliasing when f s = 2000 samples/s.
13e−jπ/4 6 13e jπ/4
7e−j3π/4 7e j3π/4
-
−π −0.5π −0.2π 0 0.2π 0.5π π ω̂
(b) The reconstructed output when the D-to-C rate is f s = 1600 samples/s will be y(t) = 26 cos(320πt+π/4)+14 cos(800πt+
3π/4).
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.7 DSP First 2e
(a) No, because there is aliasing.
(b) Sketch will have four spectral lines: at ω̂ = ±4π/9 with complex amplitudes equal to 13e±jπ/4 , and at ±2π/9 with
complex amplitudes of 7e±j3π/4 .
When f s = 450 samples/s, there will be an alias for the lines at ±500 Hz, i.e., the spectral line first maps to ω̂ = ±20π/9
and then aliases to ±2π/9.
13e−jπ/4 6 13e jπ/4
7e−j3π/4 7e j3π/4
-
−π −4π/9 −2π/9 0 2π/9 4π/9 π ω̂
(c) The reconstructed output when f s = 450 samples/s is y(t) = 26 cos(400πt + π/4) + 14 cos(100πt + 3π/4).
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.8 DSP First 2e
(a) Yes. f Nyquist = 1000 Hz.
(b) Sketch will have two spectral lines at ω̂ = ±4π/7 with complex amplitudes equal to 6e±jπ/4 .
When f s = 700 samples/s, there will be a folded alias for the lines at ±500 Hz, i.e., the spectral line first maps to
ω̂ = ±10π/7 and then aliases (via folding) to ∓4π/7. The complex amplitudes of the coincident spectral lines must be
summed.
6e−jπ/4 6 6e jπ/4 = 13e jπ/4 + 7e−j3π/4
-
−π −4π/7 0 4π/7 π ω̂
(c) The reconstructed output when f s = 700 samples/s is y(t) = 12 cos(400πt + π/4).
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.9 DSP First 2e
(a) ω = 2π(15200) rad/s.
(b) f s = 120 samples/s.
(c) f s = 75 samples/s.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.10 DSP First 2e
(a) ω1 = 18π rad/s.
(b) ω̂ = 0.6π rad.
(c) ω̂1 = π/7 and ω̂2 = −π/7, but b[n] involves a folded alias so ω̂2 = π/7.
(d) Unequal because the phases are different: a[n] = cos(πn/7 + 0.1π) and b[n] = cos(πn/7 − 0.1π).
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.11 DSP First 2e
(a) Average is 9 samples per period.
(b) ω = 8000π rad/s.
(c) Average is 0.9 samples per period, or 9 samples taken every 10 periods.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.12 DSP First 2e
Frequencies will be: 1000 Hz, 4000 Hz (but x(t) will have a different phase), and 6000 Hz.
x 1 (t) = 9.1 cos(2000πt + 0.4π)

x 2 (t) = 9.1 cos(8000πt − 0.4π)
x 3 (t) = 9.1 cos(6000πt + 0.4π)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.13 DSP First 2e
(a) f = 1600 Hz. Phase will be −2.03 rad.
(b) fsamp = 9600 Hz.
(c) f = 2200 Hz. Duration = 12.8 s.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.14 DSP First 2e
(a) Sketch will have six spectral lines at ω = ±2000π, ±(2000π ± 6000) rad/s.
In hertz, these frequencies are f ≈ ±47.07, ±1000, ±1954.93 Hz.
The complex amplitudes of the spectral lines are 41 e±jπ/2, 1.5e±j0, 41 e∓jπ/2 .
1.5 1.5
6
j
4 − 4j j
4 − 4j
-
−1955 −1000 −47 0 47 1000 1955 f (Hz)
(b) Not periodic.
(c) f s > 3909.859 Hz.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.15 DSP First 2e
(a) x 1 (t) = cos(2π(2800)t + 0.2π) and x 2 (t) = cos(2π(5200)t − 0.2π)
(b) y(t) = cos(2π(1750)t + 0.2π)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.16 DSP First 2e
(b) x(t) = cos(2π(7300)t − 0.2π)

ω̂
(c) f = 8000 − f s Hz, and ϕ = −0.2π rad.
2π
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.17 DSP First 2e
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.18 DSP First 2e
(a) ω1 = 12π rad/s.
(b) ω̂ = 0.8π rad.
(c) ω1 = ω2 = 200π rad/s.

( )
(d) The three possible answers are 2π(14 27 ), 2π(28 47 ), 2π(33 13 ) rad/s.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.19 DSP First 2e
(a) Run the code to make the stem plot.
(b) Period of x[n] in the plot is 5 samples.
(c) ω̂0 = 0.4π rad, and x[n] = 2 cos(0.4πn − 0.6π).
(d) 2π(380/100) = 7.6π > 2π and 7.6π = 4(2π) − 0.4π.

s
Over the total duration of 0.3 s, a 380 Hz continuous-time signal would have 114 periods, but x[n] only has 6 complete w
ng
la
hi
)
n
ht
eb
c
tio
periods. ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.20 DSP First 2e
(a) Expect 34 zero crossings in 17 periods. Number of positive peaks = number of negative peaks = 17.
(b) 11 zero crossings, 6 positive peaks and 6 negative peaks.
(c) T = 0.2 s, so f = 5 Hz. Sampling a 15-Hz sinusoid at 20 Hz (Ts = 0.05) s. involves aliasing, and a 5-Hz sinusoid is a
folded version of the 15-Hz sinusoid.
(d) x[n] = 1.005 cos(0.4πn − 0.1).
(e) To obey the Sampling Theorem, Ts < (1/30) s, but to make a smooth plot Ts < 0.05 s would be needed.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.21 DSP First 2e
x(t) = (3 − j0.5e jπt + j0.5e−jπt )(0.5e j13πt+jπ/2 + 0.5e−j13πt−jπ/2 )

= 1.5e j13πt+jπ/2 + 1.5e−j13πt−jπ/2 + 0.25e j14πt + 0.25e−j14πt + 0.25e j12πt+jπ + 0.25e−j12πt−jπ
= 3 cos(13πt + π/2) + 0.5 cos(14πt) + 0.5 cos(12πt + π)
(a) ω1 = 12π rad/s, A1 = 0.5, ϕ1 = π; ω2 = 13π rad/s, A2 = 3, ϕ2 = π/2; ω3 = 14π, A3 = 0.5, ϕ3 = 0;
(b) Use the second line above to plot six spectrum lines.
−3 j 3j
6
0.5 0.5 0.5 0.5

-
−14π −13π −12π 0 12π 13π 14π ω (rad/s)
(c) Nyquist rate is 28π rad/s, or 14 Hz.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.22 DSP First 2e
(a) Sampling rate is greater than 2(300π) rad/s, or 300 Hz.
(b)
x[n] = x(t) t=n/250 = 2 cos(2π(50)n/250 + π/2) + cos(2π(150)n/250)

= 2 cos(2πn/5 + π/2) + cos(6πn/5) = 2 cos(0.4πn + 0.5π) + cos(0.8πn)
(c) Plot four lines; two at ±0.4π with complex amps equal to 1e±j0.5π ; and two at ±0.8π with complex amps equal to 10.5.
Note: amplitudes are not to scale.
10.5 6 10.5
e−jπ/2 e jπ/2
-
−π −0.8π −0.4π 0 0.4π 0.8π π ω̂
s
w
ng
la
(d) f s = 300π rad/s, or 150 Hz.
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.23 DSP First 2e
(a) The discrete-time exponential when flashing at 15 flashes/s is p[n] = e+j2π (13)n/15+jπ/2 = e j2π (−2)n/15+jπ/2 . Thus, at
each flash the spot moves −4π/15 rad, which is 48◦ clockwise.
(b) p(t) = p[n] = e j2π (−2)(15t)/15+π/2

n=15t
appears to rotate through an angle of −4π every second. This is 2 rev/s clockwise.
(c) Thirteen flashes in one second; disk moving CCW which is negative. The disk moves 2π(−15/360) + 2π` rad per
flash. Thus the disk moves 2π(13)(−1/24 + `) in one second. When ` = 0, the rotation speed is 13/24 = 0.54 rev/s
CCW. When ` < 0, the rotation speed is |13` − 13/24| rev/s CCW. When ` > 0, the rotation speed is |13` − 13/24|
rev/s CW. For example, when ` = 1, the speed is 12.46 rev/s CW.
Complex Exponential Derivation of Rotation Speeds (r) in rev/s:
The discrete-time exponential after flashing at f s flashes/s is p[n] = e+j2π (r )n/fs . Thus, to get the movement of 15◦
CCW, which is −2π/24 rad/s CCW, we must determine whether or not there is is an integer ` for which the following
is true:
p[1] = e+j2π (r )(1)/fs = * 1e j2π` e−j2π (15/360)

p[0]

e+j2π(r )/fs = e−jπ/12 e j2π` ⇒ 2π(r)/ f s = 2π` − 2π/24 ⇒ r = f s (` − (1/24))

w
s
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.24 DSP First 2e
(a) p(t) = e+j2π (12)t rotates through an angle of 24π every second.
(b) The discrete-time exponential after flashing at n flashes/s is p[m] = e+j2π (12)m/n . Thus, given n, we must determine
whether or not there is is an integer ` for which the following is true:
p[1] = p[0]
+j2π(12)(1)/n
e = e j2π` ⇒ 24π/n = 2π` ⇒ 12/n = `
So, n must be a divisor of 12, which gives flashing rates of n = 1, 2, 3, 4, 6, 12 flashes/s.
(c) When n = 13, p[1] = e+j2π (12)/13 = e−j2π (1)/13 , so the disk rotates clockwise by 2π/13 rad, or 360/13 = 27.69◦ , for
every flash. The observed rotation is 1 rev/s clockwise.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.25 DSP First 2e
(a) Plot 8 unit-length vectors (with zvect). The angles will be {0.2π, 0.5π, 0.8π, 0, 0.3π, −0.5π, 0.5π, −0.5π}.
(b) Period is 20.
(c) Plot 5 unit-length vectors (with zvect). The angles will be {0, π/15, 4π/15, −14π/15, −14π/15}.
(d) Shortest period is 30.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.26 DSP First 2e
(a) θ[n] = 0.07π, 1.75π, 7π are the angles of the vectors. When reduced to the −π to +π range, these angles are
θ[n] = 0.07π, −0.25π, π.
(b) The instantaneous frequency of the signal below is 2π(44.8 × 103 )t Hz. The plot runs from t = 0 to t = 200/8000 =
1/40 s, so it starts at f = 0 and ends at (1/40)44.8 × 103 = 1120 Hz.
x(t) = x[n] = cos(π(0.7 × 10−3 × 64 × 106 )t 2 )

n=8000t
= cos(π(44.8 × 103 )t 2 )
(c) 2800 Hz
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-4.27 DSP First 2e
ALPHA = 437.5, BETA = 2000, fs = 8000
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

Chapter 5
FIR Filters
5-1 Problems
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
88
CHAPTER 5. FIR FILTERS
P-5.1 DSP First 2e
y[n] = 7x[n] + x[n − 3] − 5x[n − 4]
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.2 DSP First 2e

√
= 9 2 e+j3π/4

G = 9e−j0.5π (3) − 9e−j0.5π (4) e−j0.5πn
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.3 DSP First 2e
(a) First of all, y[n] = 0 for n = −2, −1. The system is causal. Then y[n] = n + 1 for n = 0, 1, 2, 3; and y[n] = 4 for n ≥ 4.
(b) Plot with Matlab.
(c)



 0 n<0
y[n] =  n + 1 0≤n<L



L n≥L

s
w
ng
 la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.4 DSP First 2e
(a) Use numerical convolution table:

n n<0 0 1 2 3 4 5 6 7 n>7
x[n] 0 1 2 3 2 1 1 1 1 1
h[n] 0 2 −3 2 0
h[0]x[n] 0 2 4 6 4 2 2 2 2 2
h[1]x[n−1] 0 0 −3 −6 −9 −6 −3 −3 −3 −3
h[2]x[n−2] 0 0 0 2 4 6 4 2 2 2
y[n] 0 2 1 2 −1 2 3 1 1 1
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
(b)
s
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
(c) h[n] = 2δ[n] − 3δ[n − 1] + 2δ[n − 2]

r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.5 DSP First 2e
Here are the figures:

2
2 xŒn
xŒn v2 Œn
Unit
Unit Delay
3
Delay 3
xŒn 1
v1 Œn
Unit Unit
Delay Delay
2 2
xŒn 2
yŒn yŒn
(a) (b)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.6 DSP First 2e
(a) P = N + M
(b) Input signal is time-shifted by N1 . Thus, the output starts at N3 = N1 , and the last nonzero value in the output is at
N4 = N2 + M. The length of y[n] is N4 − N3 + 1 = N2 − N1 + M + 1.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.7 DSP First 2e
(a) Make a plot in Matlab.
(b) First of all, y[n] = 0 for n = −2, −1. The system is causal. Then y[n] = 1 for n = 0; and y[n] = (−0.5) n + (−0.5) n−1
for n = 1, 2, 3, 4, 5, 6, 7, 8. The nonzero values starting at n = 0 are
1.0, 0.5, −0.25, 0.125, −0.0625, 0.03125, −0.015625, 0.0078125, −0.00390625
(c) Causality implies that y[n] = 0 for n < 0.

s
w
ng
la
Running onto the signal (partial overlap) region which is 0 ≤ n < L−1: sum (−0.5) k for k = 0 to k = n
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
n
s
(−0.5) 0 − (−0.5) n+1 1 − (−0.5) n+1

i
W
r
ss
to
es
X
(−0.5) k = = rld
c
i
at
.D
ru
o
St
.
st
1 − (−0.5) 1.5
ed
ng
W
in
d
itt
ni
k=0
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
Complete overlap region: sum (−0.5) k for k = n − L + 1 to k = n which is n ≥ L−1:

g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
n
le
si
pr
(−0.5) n−L+1 − (−0.5) n+1 (−0.5) −L − 1

k
k
es
so
or
or
is
X
(−0.5) k = = (−0.5) n+1
w
s
w
ed
k
as
or
e
id
hi
1 − (−0.5) 1.5
th
w
d
ov
t
an
of
of
is
k=n−L+1
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.8 DSP First 2e
Filter order is M = 4, and bk = {3, 7, 13, 9, 5}
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.9 DSP First 2e
Impulse response is h[n] = 13δ[n] − 13δ[n − 1] + 13δ[n − 2]. The output can be obtained via convolution: y[n] = h[n] ∗ x[n].
Assume that the input is nonzero for −∞ < n < ∞, which implies that x[n − 2] is equal to x[n]. Also, x[n − 1] will be equal
to 0 for n odd, and equal to 1 for n even.
 26
 n odd
y[n] = 13x[n] − 13x[n − 1] + 13x[n − 2] = 26x[n] − 13x[n − 1] = 
 −13
 n even
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.10 DSP First 2e
(a) y[n] = x[n] cos(0.2πn) (1) linear YES, (2) time-invariant NO, and (3) causal YES.
(b) y[n] = x[n] − x[n − 1] (1) linear YES, (2) time-invariant YES, and (3) causal YES.
(c) y[n] = |x[n]| (1) linear NO, (2) time-invariant YES, and (3) causal YES.
(d) y[n] = Ax[n] + B, where A and B are nonzero constants.

(1) linear NO, (2) time-invariant YES, and (3) causal YES.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.11 DSP First 2e
(a) Plot of y[n]
(b) Use LTI properties with x[n] = 7 (δ[n] − δ[n − 1]) +7 (δ[n − 1] − δ[n − 2]).
| {z }
first x[n] in table
Thus, y[n] = 7 (δ[n] − δ[n − 1] + 2δ[n − 3]) +7 (δ[n − 1] − δ[n − 2] + 2δ[n − 4])
| {z }
first x[n] in table
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.12 DSP First 2e
Note that the Delay by 2 block in (b) is the cascade of two unit delays.
1
xŒn
3
v3 Œn
xŒn
Delay
by 2
4
Unit
Delay 2
xŒn 1
v2 Œn
Unit
Unit Delay
2
Delay 4
xŒn 2
v1 Œn
Delay Unit
by 2 Delay
1 3
s
w
ng
la
xŒn 4
hi
)
n
yŒn
ht
yŒn
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
(a) rld (b)

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.13 DSP First 2e
(a) Use the LTI properties. The output will be y2 [n] = y1 [n] − 2y1 [n − 2].
y2 [n] = δ[n] − 3δ[n − 1] − 3δ[n − 2] − 2δ[n − 2] + 6δ[n − 3] + 6δ[n − 4]

= δ[n] − 3δ[n − 1] − 5δ[n − 2] + 6δ[n − 3] + 6δ[n − 4]
(b) A = 1 and B = −1.
(c) Using LTI properties,
h[n] = ( A/3)y1 [n] + (B/3)y1 [n − 1]

= (1/3) (δ[n] − 3δ[n − 1] − 3δ[n − 2]) − (1/3) (δ[n − 1] − 3δ[n − 2] − 3δ[n − 3])
= (1/3)δ[n] − (4/3)δ[n − 1] + δ[n − 3]
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.14 DSP First 2e
Do the convolutions with h[n] = δ[n] − aδ[n − 1]
(a) x[n] ∗ h[n] = a n u[n] ∗ (δ[n] − aδ[n − 1]) = a n u[n] − aa n−1 u[n − 1] = a n (u[n] − u[n − 1]) = δ[n]
| {z }
δ[n]
(b) Use the result from the previous part. h[n] ∗ a n (u[n] − u[n − 8]) = h[n] ∗ a n u[n]) − h[n] ∗ a8 a n−8 u[n − 8]
| {z } | {z }
δ[n] a8 δ[n−8]
(c) Sketch the plot which has only two nonzero stems, 1 at n = 0, and a8 at n = 8.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.15 DSP First 2e
Use trial and error, and just do the convolutions
(a) x[n] ∗ δ[n − 3] = x[n − 3], so we want x[n − 3] = 5u[n − 2] − u[n − 6]. Unshift to get x[n] = 5u[n + 1] − u[n − 3].
(b) y[n] = u[n − 2] − u[n − 3] which simplifies to y[n] = δ[n − 2]
(c) x[n] − x[n − 1] = y[n] = δ[n − 2] − δ[n − 7].

Rewrite y[n] as y[n] = u[n − 2] − u[n − 3] − u[n − 7] − u[n − 8].
| {z } | {z }
δ[n−2] δ[n−7]
Rewrite again as y[n] = u[n − 2] − u[n − 7] − (u[n − 3] − u[n − 8]) .
| {z } | {z }
x[n] x[n−1]
(d) x[n] + x[n − 1] = y[n] = −8δ[n] − 8δ[n − 3].

Mimic the previous part to obtain x[n] = −8u[n] + 8u[n − 3]
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.16 DSP First 2e
It is easy to verify these; just do the convolutions
(a) h[n] = δ[n − 1]
(b) h[n] = δ[n] − δ[n − 1]
(c) h[n] = δ[n − 1] − 0.5δ[n − 2] because ( 12 ) n−1 u[n − 1] − 12 ( 12 ) n−2 u[n − 2] is zero for n ≥ 2.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.17 DSP First 2e
If an FIR filter with impulse response h[n] is used to deconvolve, then
y[n] = h[n] ∗ (δ[n] + δ[n − 1]) = h[n] + h[n − 1]

*0
We want y[0] = δ[0] = 1 ⇒ 1 = y[0] = h[0] + ⇒ h[0] = 1

h[−1]

1
We want y[1] = δ[1] = 0 ⇒ 0 = y[1] = h[1] + ⇒ h[1] = 1
>

h[0]

At this point we know that the order of the filter is at least M = 1. For an M th order FIR filter the coefficient of z −M must
be nonzero, i.e., b M , 0. With that fact, the following shows that y[M + 1] will also be nonzero, so y[n] cannot be equal to
δ[n].
M
X
h[n] = bk δ[n − k] where b M , 0
k=0
*0
⇒ y[M + 1] = h[M
+
1] + h[M] = h[M] = b M , 0

s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.18 DSP First 2e
(a) Let x 1 [n] = δ[n], x 2 [n] = δ[n], and x 3 [n] = δ[n],
S1 : h1 [n] = δ[n] + δ[n − 1]

S2 : h2 [n] = δ[n] − δ[n − 2]
S3 : h3 [n] = δ[n − 1] − δ[n − 2]
(b) Carry out the convolutions:
h[n] = h1 [n] ∗ h2 [n] ∗ h3 [n]

h[n] = (δ[n] + δ[n − 1) ∗ (δ[n] − δ[n − 2]) ∗ (δ[n − 1] − δ[n − 2])
| {z }
h[n] = (δ[n] + δ[n − 1) ∗ (δ[n − 1] − 2δ[n − 2] + δ[n − 3])
h[n] = δ[n − 1] − δ[n − 2] − δ[n − 3] + δ[n − 4]
(c) Get the filter coefficients from h[n], and then
y[n] = x[n − 1] − x[n − 2] − x[n − 3] + x[n − 4]

s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-5.19 DSP First 2e
In this case, the three systems are as follows:
S1 : y1 [n] = x[n] + 2x[n − 1] + 3x[n − 2] + 4x[n − 3] h1 [n] = δ[n] + 2δ[n − 1] + 3δ[n − 2] + 4δ[n − 3]
S2 : y2 [n] = x[n] + x[n − 1] + x[n − 2] + x[n − 3] h2 [n] = δ[n] + δ[n − 1] + δ[n − 2] + δ[n − 3]
S3 : y[n] = w[n] − w[n − 1] h3 [n] = δ[n] − δ[n − 1]
where w[n] = y1 [n] + y2 [n]. The impulse responses are also given above
(a) We need three blocks and one adder. Systems S 1 and S 2 are in parallel; their outputs are added and the sum is
processed by S 3 .
(b) We can get h[n] from the combination of the individual systems, and the write the difference equation from the filter
coefficients.
h[n] = (h1 [n] + h2 [n]) ∗ h3 [n]

h[n] = (2δ[n] + 3δ[n − 1] + 4δ[n − 2] + 5δ[n − 3]) ∗ (δ[n] − δ[n − 1])
h[n] = 2δ[n] + δ[n − 1] + δ[n − 2] + δ[n − 3] − 5δ[n − 4]
y[n] = 2x[n] + x[n − 1] + x[n − 2] + x[n − 3] − 5x[n − 4]
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
If we had the z-transform at this point we would write em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.

H (z) = 1 + 2z −1 + 3z −2 + 4z −3 + 1 + z −1 + z −2 + z −3
st
1 − z −1
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U

H (z) = 2 + 3z −1 + 4z −2 + 5z −3 1 − z −1 = 2 + z −1 + z −2 + z −3 − 5z −4
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

Chapter 6
Frequency Response
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
108
CHAPTER 6. FREQUENCY RESPONSE
P-6.1 DSP First 2e
y[n] = x[n] − x[n − 1] = e j (0.4πn−0.5π) − e j (0.4π(n−1)−0.5π) = (1 − e−j0.4π )e j (0.4πn−0.5π) = H (e j0.4π )e j (0.4πn−0.5π)
H (e j0.4π ) = 1 − e−j0.4π = 1.1756e j0.3π

Therefore, A = 1.1756, ϕ = 0.3π − 0.5π = −0.2π, and ω̂0 = 0.4π.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.2 DSP First 2e
(a)
y[n] = x[n] + x[n − 1] = (e j0.1πn + e j0.4πn ) + (e j0.1π (n−1) + e j0.4π (n−1) )

= (1 + e−j0.1π )e j0.1πn + (1 + e−j0.4π )e j0.4πn
= H (e j0.1π )e j0.1πn + H (e j0.4π )e j0.4πn
where H (e j ω̂ ) = 1 + e−j ω̂ .
(b) The system is LTI, so superposition applies. Each complex exponential is modified by the frequency response evaluated
at its frequency.
(c)
H (e j0.1π ) = 1 + e−j0.1π = A1 e jϕ1 = 1.9754e−j0.05π
H (e j0.4π ) = 1 + e−j0.4π = A2 e jϕ2 = 1.6180e−j0.2π
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.3 DSP First 2e
2
(a) y[n] = e j0.1πn + e j0.4πn = e j0.2πn + 2e j0.5πn + e j0.8πn
(b) Clearly, the squaring operation produces three complex exponential signals each with a different frequency that is not
one of the original two frequencies.
(c) The system is clearly not linear since the output cannot be written as the sum of two terms involving the two input
frequencies. The principle of superposition is not satisfied, therefore the system is NOT linear.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.4 DSP First 2e
(a) H (e j ω̂ = 2 − 2e−j ω̂ + 2e−j ω̂2 = 2e−j ω̂ (e j ω̂ − 1 + e−j ω̂ ) = 2e−j ω̂ (2 cos(ω̂) − 1)

Thus, R(e j ω̂ ) = 4 cos(ω̂) − 2 and n0 = 1.
(b) The period = 2π

10
|H(ej ω̂ )|
0
−π −π/2 0 π/2 π 3π/2 2π 5π/2 3π
π
H(ej ω̂ )
0
6
−π
−π −π/2 0 π/2 π 3π/2 2π 5π/2 3π
(c) ω̂
s
w
(d) The output is zero whenever the magnitude of the frequency response is zero; i.e., when |2 cos(ω̂) − 1| = 0 or when
ng
la
hi
)
n
ht
eb
c
tio
ea
cos(ω̂) = 0.5. The solution is therefore ω̂ = π/3 rad.
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
(e) We know that for a cosine input with frequency ω̂0 = 0.5, the output is rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
y[n] = |H (e j0.5π )| cos(0.5πn + ∠H (e j0.5π ))

te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
where H (e j0.5π ) = 2e−j0.5π (2 cos(0.5π) − 1) = −2e−j0.5π = 2e j0.5π

he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
⇒ y[n] = 2 cos(0.5πn + 0.5π) = 2 cos(0.5π(n + 1)).

n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.5 DSP First 2e
(a) H (e j ω̂ ) = 3 + 2e−j ω̂ + e−j ω̂2
(b) y[n] = H (e j0 ) + 2|H (e j0.5π )| cos(0.5πn + 0.1π + ∠H (e j0.5π ))

√
where H (e j0 ) = 6 and H (e j0.5π ) = 3 − 2e−j0.5π + e−jπ = 2e−j0.25π .
√ √
Thus, y[n] = 6 + 2 2 cos(0.5πn + 0.1π − 0.25π) = 6 + 2 2 cos(0.5πn − 0.15π)
(c)

 0 n<0
n=0

3


y[n] = 3u[n] + 2u[n − 1] + u[n − 2] = 



 5 n=1


 6 n≥2
(d) The unit step input is identical to a suddenly applied complex exponential of zero frequency. Thus, for an FIR filter
like this, the steady state response is simply H (e j0 ) times 1. The steady state is reached at n = 2 because the filter
order is 2.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.6 DSP First 2e
(a) H (e j ω̂ ) = 1 − e−j ω̂4 = e−j ω̂2 (e j ω̂2 − e−j ω̂2 ) = 2 je−j ω̂2 sin(2ω̂)
(b) y[n] = 4H (e j0 ) + |H (e j0.25π )| cos(0.25πn − 0.25π + ∠H (e j0.25π ))

where H (e j0 ) = 0 and H (e j0.25π ) = 2 j (− j) sin(π/2) = 2. Thus,
y[n] = 2 cos(0.25πn − 0.25π)
(c) Since the impulse response has length 5 samples, the output reaches the steady state at n = 4 so the output due to the
suddenly applied signal is identical to the output in (b) for n ≥ 4.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.7 DSP First 2e
(a) h[n] = 3δ[n] − δ[n − 2]
(b) We need to expand H (e j ω̂ ) in terms of complex exponentials.

j2ω̂ + e−j2ω̂
!
j ω̂ −j3ω̂ e
H (e ) = 5e = 2.5e−j ω̂ + 2.5e−j5ω̂
2
Therefore h[n] = 2.5δ[n − 1] + 2.5δ[n − 5].
(c) We should recognize this as the form of an L sample running sum whose frequency response is
L−1
j ω̂ −j ω̂(L−1)/2 sin(ω̂L/2)
e−j ω̂n
X
H (e )=e =
sin(ω̂/2) n=0
Thus, it follows that L = 10 and the impulse response is

(
1 0≤n≤9
h[n] = u[n] − u[n − 10] =
0 otherwise
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.8 DSP First 2e
(a) To get the difference equation we need to expand the factors of H (e j ω̂ ) so that we get powers of e−j ω̂ .
√
H (e j ω̂ ) = (1 − e−j ω̂ )(1 − 2 cos(3π/4)e−j ω̂ + e−j2ω̂ ) = (1 − e−j ω̂ )(1 − 2e−j ω̂ + e−j2ω̂ )
√ √
= 1 + ( 2 − 1)e−j ω̂ − ( 2 − 1)e−j2ω̂ − e−j3ω̂
= 1 + 0.4142e−j ω̂ − 0.4142e−j2ω̂ − e−j3ω̂
Therefore, the difference equation is
y[n] = x[n] + 0.4142x[n − 1] − 0.4142x[n − 2] − x[n − 3]
(b)

 0 n<0
n=0




 1
y[n] = u[n] + 0.4142u[n − 1] − 0.4142u[n − 2] − u[n − 3] =  n=1

1.4142
n=2

1





0 n≥3

s

w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
(c) We need to find the values of ω̂ such that H (e j ω̂ ) = 0. These are easily found from the given factored form as ω̂ = 0
W
a
t
py
in
in
e
em
co
d
s
i
W
and ω̂ = ±3π/4.
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.9 DSP First 2e
(a) We need to multiply out the factors of H (e j ω̂ ) to get:
H (e j ω̂ ) = 1 + 0.5e−j ω̂ − 0.25e−j2ω̂ + 0.25e−j3ω̂
y[n] = x[n] + 0.5x[n − 1] − 0.25x[n − 2] + 0.25x[n − 3]
(b) By definition, the impulse response is:
h[n] = δ[n] + 0.5δ[n − 1] − 0.25δ[n − 2] + 0.25δ[n − 3]
(c) We need to find the values of ω̂ such that H (e j ω̂ ) = 0. The factored form shows that this can be true for values of ω̂
such that 1 + e−j2ω̂ = 0 or ω̂ = ±π/2 rad.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.10 DSP First 2e
(a) Plot {3, 0, −2, 1} for 0 ≤ n ≤ 3.
(b) The first entry in the table says that the impulse response is h[n] = δ[n] − δ[n − 3]. Therefore, y[n] = x[n] − x[n − 3].
For the given input the output is
y[n] = 3δ[n] − 2δ[n − 2] + δ[n − 3] − 3δ[n − 3] + 2δ[n − 5] − δ[n − 6]

= 3δ[n] − 2δ[n − 2] − 2δ[n − 3] + 2δ[n − 5] − δ[n − 6]
Plot {3, 0, −2, −2, 0, 2, −1} for 0 ≤ n ≤ 6.
(c) In this case we can use the third result in the table, which says that the frequency response at ω̂ = π/3 is H (e jπ/3 ) = 2;
i.e., the phase shift is zero. Therefore, the output for this input is y[n] = 2 cos(π(n − 3)/3)
(d) It is FALSE because the impulse response is δ[n] − δ[n − 3], so the frequency response is H (e j ω̂ ) = 1 − e−j3ω̂ which
is nonzero at ω̂ = π/2.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.11 DSP First 2e
(a) The zeros marked with o are at multiples of 2π/8 with gaps at 0, ±2π, ±4π, . . ..
D8 (ω̂) 1
−1
−3π −2π −π 0 π 2π 3π
Figure P-6.11
(b) The period is 4π.
(c) The maximum value occurs at ω̂ = 0. Using the small angle approximation for the sine functions we get
(L ω̂/2)
D L (0) ≈ =1
L(ω̂/2)
We get the same result using L’ Hospital’s Rule.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.12 DSP First 2e
(a) From the difference equation, we can write down H (e j ω̂ ) as
H (e j ω̂ ) = 1 − 3e−j ω̂ + 3e−j2ω̂ − e−j3ω̂ = (1 − e−j ω̂ ) 3 = e−j ω̂3/2 (e j ω̂/2 − e−j ω̂/2 ) 3

= e−j ω̂3/2 (2 j) 3 (sin(ω̂/2)) 3 = 8e−j ω̂3/2 (− j)(sin(ω̂/2)) 3
= 8 sin3 (ω̂/2)e j (−π/2−3ω̂/2)
10
|H(ej ω̂ )|
0
−2π −3π/2 −π −π/2 0 π/2 π 3π/2 2π
Figure P-6.12
(b)
s
w
ng
la
hi
)
n
(c) We need to determine the frequency response at 0 and 0.5π.
ht
eb
c
tio
ea
rig
W
a
√ √
t
py
in
in
e
These are H (e j0 ) = 0 and H (e j0.5π ) = 8(1/ 2) 3 e j (−π/2−3π/4) = 2 2e−j7π/4 . Therefore, em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
y[n] = 5H (e j0 ) + 6|H (e j0.5π )| cos(0.5π(n − 1) + ∠H (e j0.5π ))

in
d
itt
ni
e
te
of
th
m
ar
ni
√
er
e
le
on
U
us
tp
= 12 2 cos(0.5π(n − 1) − 7π/4)
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
(d) h[n] = δ[n] − 3δ[n − 1] + 3δ[n − 2] − δ[n − 3]

k
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
(e)
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
y[n] = 5H (e j0 ) + 6|H (e j0.5π )| cos(0.5πn + ∠H (e j0.5π )) + 9h[n − 3]

r
e
of
ei
th
th
√
y
l
sa
ro
= 12 2 cos(0.5πn − 7π/4) + 9h[n − 3]

st
or
de
ill
w
where h[n] is as determined in part (d).

P-6.13 DSP First 2e
(a)
S1 : H1 (e j ω̂ ) = 1 − e−j ω̂
S2 : H2 (e j ω̂ ) = 1 + e−j2ω̂
S3 : H3 (e j ω̂ ) = e−j ω̂ + e−j2ω̂
(b) H (e j ω̂ ) = (1 − e−j ω̂ )(1 + e−j2ω̂ )(e−j ω̂ + e−j2ω̂ )
(c) Multiply out all the factors in H (e j ω̂ ) and then pick off the coefficients to get the difference equation.
H (e j ω̂ ) = (1 − e−j ω̂ + e−j2ω̂ − e−j3ω̂ )(e−j ω̂ + e−j2ω̂ ) = e−j ω̂ − e−j5ω̂
Therefore, y[n] = x[n − 1] − x[n − 5]
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.14 DSP First 2e
1 − e−j5ω̂
4
sin(ω̂5/2)
(a) H (e j ω̂ ) = e−j ω̂n = = e−j ω̂2
X
n=0
1 − e−j ω̂ sin(ω̂/2)
(b) H (e j ω̂ ) = 0 for ω̂ = ±0.4π, ±0.8π
(c) We need the frequency response at ω̂ = 0, 0.25π, and 0.4π.
H (e j0 ) = 5
sin(5π/8)
H (e j0.25π ) = e−j0.5π = 2.4142e−j0.5π
sin(π/8)
H (e j0.4π ) = 0
Therefore, y[n] = 35 + 8(2.4142) cos(0.25πn − π/2)
(d) Since the length of the impulse response is 5, the steady-state output is attained for n ≥ 4.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.15 DSP First 2e
The sampled signal is
x[n] = x(n/ f s ) = 7 + 8 cos(1000πn/4000) + 9 cos(1600πn/4000)

= 7 + 8 cos(0.25πn) + 9 cos(0.4πn + 0.7π)
This is the same input signal as in Problem ?? and the discrete-time system is the same as in that problem. Therefore, from
the solution to Problem ??(c), the output sequence is
y[n] = 35 + 8(2.4142) cos(0.25πn − π/2)
The output of the D-to-C converter is therefore
y(t) = 35 + 8(2.4142) cos(1000πn − π/2)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.16 DSP First 2e
(a) f s > 1000 Hz.

Since h[n] = δ[n], y[n] = x[n]. In other words, the system is a C-to-D converter followed by a D-to-C converter, so
the issue is sampling above the Nyquist rate. The highest frequency in the input signal is 500 Hz, so the Nyquist rate
is 2(500) = 1000 Hz.
(b) A delay of 10 samples must correspond to 0.001 s. After sampling x(t) at a rate of f s
x[n] = 10 + 20 cos(ω0 (n/ f s ) + π/3)

y[n] = x[n − 10] = 10 + 20 cos(ω0 (n − 10)/ f s + π/3)
Reconstructing y(t) from y[n] with an ideal D-to-C converter gives
y(t) = 10 + 20 cos(ω0 ( f s t − 10)/ f s + π/3) = 10 + 20 cos(ω0 t − 10ω0 / f s + π/3)
We want y(t) = x(t − 0.001) = 10 + 20 cos(ω0 (t − 0.001) + π/3) which requires
−10ω0 / f s = ω0 (−0.001) ⇒ f s = 10, 000 Hz
s
w
ng
In addition, the output frequency must be the same as the input frequency, so there must be no aliasing, i.e., the
la
hi
)
n
ht
eb
relationship between ω0 and f s = 10, 000 Hz must obey the Nyquist condition. c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
2ω0 < 2π f s ⇒ ω0 < 2π(5000) rad/s rld

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
(c) To get y(t) = A, we need y[n] = constant. Since x[n] = 10 + 20 cos(ω0 (n/ f s ) + π/3), the filter must null out the cosine
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
term; or the cosine term must alias to DC.

is
st
c
fo
cl
d
te
in
an
y
n
o
sin(5ω̂/2) −j ω̂2
(
l
le
si
pr
Nulling: The zeros of H (e j ω̂ ) =

k
es
so
are at ω̂ = 2π/5, −2π/5, 4π/5, −4π/5.

or
or
is
e
w
s
w
ed
5 sin(ω̂/2)
k
as
or
e
id
hi
th
w
For ω0 and f s = 2000 Hz, these frequencies are 2π(±400) and 2π(±800) rad/s.
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
The negative frequencies must be used when we include aliases. The complete list of input frequencies that alias to
y
an
co
an
in
the zeros of H (e j ω̂ ) is
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
2π(400 + 2000`) = 2π(400), 2π(2400), 2π(4400), . . .

ill
w
2π(800 + 2000`) = 2π(800), 2π(2800), 2π(4800), . . .

2π(−400 + 2000`) = 2π(−400), 2π(1600), 2π(3600), . . .
2π(−800 + 2000`) = 2π(−800), 2π(1200), 2π(3200), . . .
For these input frequencies, the cosine term will be filtered out, and only the DC remains. Since H (e j0 ) = 1 and the
DC value of x(t) = 10, the output will be y(t) = A = 10.
Cosine aliases to DC: When ω0 is a frequency that aliases to DC, then ω̂0 = (ω0 + 2π` f s )/ f s = 0,
and x[n] = 10 + 20 cos(0n + π/3) = 10 + 20 cos(π/3) = 20.
The frequencies for which this happens are
2π(0 + 2000`) = 2π(0), 2π(2000), 2π(4000), . . .
In this case, the DC value of x(t) = 20 and H (e j0 ) = 1, so the output will be y(t) = A = 20.

P-6.17 DSP First 2e
(a) The constant component passes through the first system as a constant. Then the first difference system gives a zero
output for a constant input. By superposition, the output is thus equal to y1 [n], the output due to the other input
component x 1 [n].
(b) The overall frequency response function is the product of the frequency responses. Therefore,
1 − e−j5ω̂
4 !
j ω̂ −j ω̂n + −j ω̂
(1 − e−j ω̂ ) = 1 − e−j5ω̂
X
H (e )= * e (1 − e ) = −j ω̂
, n=0 - 1−e
(c) The frequency response is zero at the 5th roots of unity or ω̂ = 2πk/5 for k = 0, 1, 2, 3, 4.
(d) From the frequency response, the overall difference equation is
y[n] = x[n] − x[n − 5]
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.18 DSP First 2e
(a) Need a plot of h[n] = −δ[n − 2] + δ[n − 12]
e j5ω̂ − e−j5ω̂
(b) H (e j ω̂ ) = −e−j2ω̂ +e−j12ω̂ = e−j7ω̂ (−e j5ω̂ +e−j5ω̂ ) = e−j7ω̂ (−2 j) = j (−2 sin(ω̂5))e−j7ω̂ Therefore, R(e j ω̂ ) =
2j
−2 sin(5ω̂) and n0 = −7.
(c,d) Plots of |H (e j ω̂ )| and the principal value of ∠H (e j ω̂ ) for −π < ω̂ ≤ π.

2
|H(ej ω̂ )|
s
1 w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
0
s
i
W
r
ss
to
es
−π −π/2 0 π/2 rld π

c
i
at
.D
ru
o
St
.
st
5
ed
ng
W
in
H(ej ω̂ )
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
0
d
ud
rt
te
is
st
c
fo
cl
d
te
6
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
-5
or
e
id
hi
th
w
−π −π/2 0 π/2 π
ov
t
an
of
of
is
pr
Th
rt
ω̂
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.19 DSP First 2e
(a)
H (e j ω̂ ) = H1 (e j ω̂ )H2 (e j ω̂ )
= (1 + 2e−j ω̂ + e−j ω̂2 )(1 − e−j ω̂ + e−j ω̂2 − e−j ω̂3 )
= 1 + e−j ω̂ − e−j ω̂4 − e−j ω̂5
(b) h[n] = δ[n] + δ[n − 1] − δ[n − 4] − δ[n − 5]
s
w
ng
la
(c) y[n] = x[n] + x[n − 1] − x[n − 4] − x[n − 5]
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.20 DSP First 2e
(a) We need to have zeros of the frequency response at ω̂ = 0 and 0.2π. Thus,
H (e j ω̂ ) = (1 − e−j ω̂ )(1 − e−j0.2π e−j ω̂ )(1 − e j0.2π e−j ω̂ )

= (1 − e−j ω̂ )(1 − 2 cos(0.2π)e−j ω̂ + e−j ω̂2 )
= 1 − 2.6180e−j ω̂ + 2.6180e−j ω̂2 − e−j ω̂3
y[n] = x[n] − 2.6180x[n − 1 + 2.6180x[n − 2] − x[n − 3]
(b) We need to null the frequencies ω̂ = 0, ±0.2π, ±0.4π, ±0.6π, ±0.8π, π. A system whose frequency response is zero at
all these frequencies is H (e j ω̂ ) = 1 − e−j ω̂10 .
If we set H (e j ω̂ ) = 0 we get the equation 1 − e−j ω̂10 = 0 or e−j ω̂10 = 1 or, equivalently, e j ω̂10 = 1 = e j2πk where
k = 0, 1, . . . , 9. If we then equate the exponents of the last equation we get ω̂10 = 2πk, or ω̂ = 2πk/10 = 0.2πk.
The difference equation for this system is y[n] = x[n] − x[n − 10].
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.21 DSP First 2e
(a) The two normalized frequencies are 120π/ f s = 0.12π and 240π/ f s = 0.24π.
(b) We need to place zeros at ω̂ = ±0.12π and ±0.24π. Therefore, choose
H (e j ω̂ ) = (1 − e−j0.12π e−j ω̂ )(1 − e j0.12π e−j ω̂ )(1 − e−j0.24π e−j ω̂ )(1 − e j0.24π e−j ω̂ )
= 1 − 3.3175e−j ω̂ + 4.7111e−j ω̂2 − 3.3175e−j ω̂3 + e−j ω̂4
The corresponding difference equation is therefore:
y[n] = x[n] − 3.3175x[n − 1] + 4.7111x[n − 2] − 3.3175x[n − 3] + x[n − 4]
(c) Here is the Matlab plot of the frequency response (magnitude) versus ω̂, and also versus the analog frequency f in
hertz. The frequencies 60 Hz and 120 Hz are marked and are snown to be nulled out.
Note: the plot of |H (e j ω̂ )| versus ω̂ is zoomed vertically so as to see the detail of the zeros.
|H(ej ω̂ )|
s
0.5 w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
0
a
t
py
in
in
e
−π −π/2 0 em
π/2 π
co
d
s
i
W
r
ss
to
es
|H(ej2πf /fs )|
normalized frequency ω̂ rld

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
20 itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
10
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
0
w
s
w
ed
k
as
or
-500 -120-60 0 60 120 500

e
id
hi
th
w
d
ov
t
an
analog frequency f
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
Figure P-6.21
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.22 DSP First 2e
(a) To find y[n] we need the frequency response evaluated at frequencies 0 and 0.5π which are from the graph:
H (e j0 ) = 0 H (e j0.5π ) = 1.4e−jπ/4
It is easy to get these analytically as well since
H (e j ω̂ ) = 1 − e−j ω̂ = je−j ω̂/2 2 sin(ω̂/2) = 2 sin(ω̂/2)e j (π/2−ω̂/2)
Thus, the output is y[n] = 10 √1 cos(0.5πn − π/4) = 7.0711 cos(0.5πn − π/4).

2
(b) The discontinuity of size π radians is due to the fact that H (e j ω̂ ) is zero at ω̂ = 0 so there is a sign change from one
side of 0 to the other. This is evident in H (e j ω̂ ) equation above since the function sin(ω̂/2) is negative for −π < ω̂ < 0
and positive for 0 < ω̂ < π.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-6.23 DSP First 2e
(a) When the input is x[n] = 10 + 10 cos(0.2πn) + 10 cos(0.5πn), we need to determine the frequency response at fre-
quencies 0, 0.2π and 0.5π. From the curves we get
H (e j0 ) = 1 H (e j0.2π ) ≈ 0.9e−j0.4π H (e j0.5π ) = 0
Therefore, y[n] = 10 + 9 cos(0.2πn − 0.4π).
(b) There are two kinds of discontinuities in the phase. At ω̂ = 2π(0.17), the discontinuity of size 2π is due to taking the
principal value. At ω̂ = 0.5π the frequency response is zero and the discontinuity is a phase jump of ±π due to the
sign change as the frequency response crosses zero.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

Chapter 7
DTFT
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
132
CHAPTER 7. DTFT
P-7.1 DSP First 2e
This solution requires mostly table lookup and applying the delay property for the DTFT.
DTFT
(a) x 1 [n] = 2δ[n − 3] ←→ X1 (e j ω̂ ) = 2e−j ω̂3
DTFT
(b) x 2 [n] = 3δ[n − 2] − δ[n − 3] + 3δ[n − 4] ←→ X2 (e j ω̂ ) = 3e−j ω̂2 − e−j ω̂3 + 3e−j ω̂4
7
 1≤n≤8
(c) x 3 [n] = 7u[n − 1] − 7u[n − 9] = 
0 otherwise
Note how the step sequence is used to create a finite-length sequence. Plug into the DTFT definition and set n = m + 1

in the sum. Then apply the formula for the sum of terms of a geometric series.
1 − e−j ω̂8
8 7
sin(4ω̂) −j ω̂9/2
X3 (e j ω̂ ) = 7e−j ω̂n = 7e−j ω̂ e−j ω̂m = 7e−j ω̂
X X
=7 e
n=1 m=0
1−e −j ω̂ sin(ω̂/2)
1
|ω̂| < 0.25π
(
sin(0.25πn) DTFT
(d) x 4 [n] = ←→ X4 (e j ω̂ ) = 9
9πn 0 0.25π < |ω̂| ≤ π
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.2 DSP First 2e
This solution requires table lookup and use of DTFT properties

DTFT
(a) Y1 (e j ω̂ ) = 2π ←→ y1 [n] = 2πδ[n]
DTFT
(b) Y2 (e j ω̂ ) = 5e−j3ω̂ ←→ y2 [n] = 5δ[n − 3]
DTFT
(c) Y3 (e j ω̂ ) = 6 cos(3ω̂) = 3e j3ω̂ + 3e−j3ω̂ ←→ y3 [n] = 3δ[n + 3] + 3δ[n − 3]
DTFT
(d) Y4 (e j ω̂ ) = j sin(7ω̂) = 21 e j7ω̂ − 12 e−j7ω̂ ←→ y4 [n] = 21 δ[n + 7] − 21 δ[n − 7]
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.3 DSP First 2e
This solution involves either evaluating the definitions of the direct and inverse DTFT or using table lookup and the
properties of the DTFT.
0.3π
e j0.3πn − e−j0.3πn sin(0.3πn)
Z
1
(a) v1 [n] = e j ω̂n d ω̂ = =
2π j2πn πn
−0.3π
1 − e−j ω̂10 e−j ω̂5 (e j ω̂5 − e−j ω̂5 )

9
j ω̂ sin(ω̂5) −j ω̂4.5
e−j ω̂n =
X
(b) V2 (e )= = = e
n=0
1−e −j ω̂ e −j ω̂/2 (e j ω̂/2 −e −j ω̂/2 ) sin(ω̂/2)
sin(0.3πn)
(c) Note that V3 (e j ω̂ ) = 1 − V1 (e j ω̂ ). Thus, v3 [n] = δ[n] − v1 [n] = δ[n] −
πn
Alternatively, you can plug into the inverse DTFT integral as in part (a) obtaining
−0.3π
Z Zπ
1 j ω̂n 1 sin(πn) sin(0.3πn) sin(0.3πn)
v3 [n] = e d ω̂ + e j ω̂n d ω̂ = − = δ[n] −
2π 2π πn πn πn
−π 0.3π
s
w
ng
la
hi
9 9
sin((ω̂ − π)5) −j (ω̂−π)4.5
)
n
ht
eb
c
tio
j ω̂ jπn −j ω̂n
e−j (ω̂−π)n = V2 (e j (ω̂−π) ) =
X X ea
rig
)= =
W
a
(d) V4 (e e e e
t
py
in
sin((ω̂ − π)/2)
in
e
em
co
d
s
i
W
r
n=0 n=0
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.4 DSP First 2e
This solution requires using the linearity and delay properties of the DTFT and the following basic DTFT pair:
sin(ω̂0 n) DTFT |ω̂| < ω̂0

(
j ω̂ 1
h[n] = ←→ H (e ) =
πn 0 ω̂0 < |ω̂| ≤ π
It might be helpful to sketch the different bandlimited DTFTs that are involved.
|ω̂| < 0.25π |ω̂| < 0.25π

( (
sin(0.25πn) DTFT j ω̂ 1 0
(a) h1 [n] = 5δ[n] − ←→ H1 (e ) = 5 − 5 =
0.2πn 0 0.25π < | ω̂| ≤ π 5 0.25π < |ω̂| ≤ π
sin(0.4πn) sin(0.1πn)
(b) h2 [n] = −
0.1πn 0.1πn
 0 0 ≤ |ω̂| ≤ 0.1π
0 ≤ |ω̂| ≤ 0.4π 0 ≤ |ω̂| ≤ 0.1π

DTFT j ω̂ 1
 1
 

) = 10  = 10  0.1π < |ω̂| ≤ 0.4π

←→ H2 (e − 10  1
0 0.4π < |ω̂| ≤ π 0 0.1π < |ω̂| ≤ π 
0.4π < |ω̂| ≤ π

0

 

sin(0.4π(n − 8)) sin(0.1π(n − 8))

(c) h3 [n] = − = 0.1h2 [n − 8]
π(n − 8) π(n − 8)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
0 0 ≤ |ω̂| ≤ 0.1π
t
py
in
in
e
em

co
d

s
i
W

r
DTFT j ω̂ j ω̂ −j ω̂8 −j ω̂8

ss
to

es
) = 0.1H2 (e =e  1 0.1π < |ω̂| ≤ 0.4π


←→ H3 (e )e rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W

in
 0 0.4π < |ω̂| ≤ π

d

itt
ni

te
of
th
m
ar
ni
er
e
le
on
U

us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.5 DSP First 2e

6
Write down the system function directly from the difference equation obtaining H (e j ω̂ ) = 3 e−j ω̂k .
X
k=2
With a change of index in the sum we can apply the formula for the sum of terms in a geometric series.
sin(ω̂5/2) −j ω̂4
H (e j ω̂ ) = 3 e
sin(ω̂/2)
Thus, the answers as requested are: L = 5, α = 3, and β = 4.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.6 DSP First 2e
The exponential factors suggest a delay, but we cannot use the DTFT delay property directly because the delay would
be a fraction of a sample. This means that we should use the inverse DTFT if we can evaluate it, and, of course, we can.
(a) Inverse DTFT of G1 (e j ω̂ )

0.4π 0.4π
e j0.4π (n−0.3) − e−j0.4π (n−0.3)
Z Z
1 −j0.3ω̂ j ω̂n 1
5eω̂(n−0.3) d ω̂ = 5
DTFT
←→ g1 [n] = 5e e d ω̂ =
2π 2π 2π(n − 0.3)
−0.4π −0.4π
sin 0.4π(n − 0.3)

so g1 [n] = 5 .
π(n − 0.3)
This looks like it was obtained using the delay property, but remember that the delay property only holds in general
for integer shifts.
(b) Using the approach of (a), the solution is

sin π(n − 0.1) sin 0.7π(n − 0.1)
g2 [n] = 4 −4
π(n − 0.1) π(n − 0.1)
s
If the delay had been an integer, the first function would reduce to an impulse, but in this case it does not. w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
(c) Again using the direct evaluation of the inverse DTFT, the solution is
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
sin 0.7π(n − 0.2) sin 0.4π(n − 0.2)

o
St
.
st
g3 [n] = 9
ed
ng
−9
in
d
itt
ni
π(n − 0.2) π(n − 0.2)

te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.7 DSP First 2e
sin(0.25πn) sin(0.44πn) sin(0.25πn)

(a) ∗ =
9πn 4πn 36πn
The DTFT of this signal is the product of two ideal lowpass filter frequency responses. Therefore the result of
the convolution is a sinc function corresponding to the lowest cutoff frequency and scaled by the product of the gains.
sin(0.25πn) sin(0.25πn) sin(0.25πn) sin(0.25πn)
(b) ∗ ∗ =
9πn 8πn 7πn (9 × 8 × 7)πn
In this case we have the convolution of three sinc functions all with the same cutoff frequency so the result is
the same signal but with the product of the gains.
(c) Considerable simplification is possible.
(0.3) n u[n] ∗ (10δ[n − 2] − 3δ[n − 3])

= 10(0.3) n−2 u[n − 2] − 3(0.3) n−3 u[n − 3]
| {z }
break into 2 terms
= 10δ[n − 2] + 10(0.3)(0.3) n−3 u[n − 3] − 3(0.3) n−3 u[n − 3]
= 10δ[n − 2]
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.8 DSP First 2e
We will need the impulse response of the system to do parts (a) and (c). We can get it by expanding H (e j ω̂ ) using the
inverse Euler relation; i.e., H (e j ω̂ ) = 21 e j ω̂3 + 12 e−j ω̂3 , so h[n] = 12 δ[n + 3] + 21 δ[n − 3].
(a) By definition, y1 [n] = h[n − 3] = 21 δ[n] + 12 δ[n − 6].
(b) Here we can use the frequency response directly if we express x 2 [n] as x 2 [n] = 1.5e j0.25πn + 1.5e−j0.25πn . Then
y2 [n] = 1.5H (e j0.25π )e j0.25πn + 1.5H (e−j0.25π )e−j0.25πn

√
= 1.5 cos(0.75π)e j0.25πn + 1.5 cos(−0.75π)e−j0.25πn = −1.5 2 cos(0.25πn)
1
 0≤n≤8
(c) This signal is a length-9 pulse: x 3 [n] = u[n] − u[n − 9] = 
0
 otherwise
This one should be done by convolution because the impulse response is so simple and a frequency-domain solution
would not be feasible.

y3 [n] = x 3 [n] ∗ h[n] = x 3 [n] ∗ 21 δ[n + 3] + 12 δ[n − 3]
s
w
ng
la
hi
)
n
ht
eb
1
c
tio
−3 ≤ n ≤ 2 ea
rig
W

a
 2
t
py
in

in
e
 em
co
d

3≤n≤5
s
1
i

W
r
ss
= 12 (u[n + 3] − u[n − 6] + u[n − 3] − u[n − 12]) = 

to
es

rld
c
i
at
.D
ru
1
o
St
6 ≤ n ≤ 11
.
st
ed
ng
W

in
 2
d
itt
ni

e
te

of
th
m
ar

ni

0 otherwise
er
e
le
on
U
us
tp
t
by
en

he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.9 DSP First 2e
(a) Expand cosine with the inverse Euler relation:
e j ω̂0 n + e−j ω̂0 n

!
x[n] = a n cos(ω̂0 n)u[n] = a n u[n] = 12 e j ω̂0 n a n u[n] + 12 e−j ω̂0 n a n u[n]
2
DTFT 1
(b) Using the DTFT pair a n u[n] ←→ and the frequency-shift property of the DTFT we can write the DTFT of
1 − ae−j ω̂
x[n] as
1 1
X (e j ω̂ ) = 2
+ 2
1− ae−j (ω̂−ω̂0 ) 1− ae−j (ω̂+ω̂0 )
1
− 12 ae−j ω̂0 e−j ω̂ + 1
− 12 ae j ω̂0 e−j ω̂
= 2 2
(1 − ae j ω̂0 e−j ω̂ )(1 − ae−j ω̂0 e−j ω̂ )
1 − 12 a(e j ω̂0 + e−j ω̂0 )e−j ω̂

=
1 − a(e j ω̂0 + e−j ω̂0 )e−j ω̂ + a2 e−j2ω̂
s
w
ng
la
1 − a cos(ω̂0 )e−j ω̂
hi
)
n
ht
eb
c
tio
=
ea
rig
W
a
t
py
in
1 − 2a cos(ω̂0 )e−j ω̂ + a2 e−j ω̂2
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.10 DSP First 2e
(a) The plot of |V (e j ω̂ )| 2 for −π < ω̂ ≤ π is below.

1 1
From the table, V (e j ω̂ ) = ω̂
so |V (e j ω̂ )| 2 = .
1 − ae −j 1 + a − 2a cos ω̂
2
1
A sketch of this for a = 0.95 would show a maximum at ω̂ = 0 of size = 400 and it would fall off monotonically
(1 − a) 2
1
with increasing ω̂ to a value of = 0.2630 at ω̂ = ±π.
(1 + a) 2
(b) If V (e j ω̂ ) were a frequency response, is the filter lowpass, highpass or bandpass? LOWPASS
(c) Using the frequency shift property of the DTFT we obtain W (e j ω̂ ) = 21 V (e j (ω̂−0.3π) ) + 12 V (e j (ω̂+0.3π) ). Therefore, a
plot of |W (e−j ω̂ )| 2 will look like two shifted versions of the plot in part (a) with everything scaled by 14 . This will
place peaks at approximately ω̂ = ±0.3π with heights of approximately 100.
Note: the plots use a semilog scale to show details.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
(d) If W (e j ω̂ ) were a frequency response, is the filter lowpass, highpass or bandpass? BANDPASS
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.11 DSP First 2e
(a) Using the definition of the ideal lowpass filter we can write
 1 |ω̂| ≤ ω̂co  1 |ω̂| ≤ ω̂co  0 |ω̂| ≤ ω̂co

Hhp (e j ω̂ ) = 
  
=
 
−
 1 ω̂co < |ω̂| ≤ π  0 ω̂co < |ω̂| ≤ π  1 ω̂co < |ω̂| ≤ π
  
which is the frequency response of an ideal highpass filter.
Note: A plot of the function 1 − Hlp (e j ω̂ ) might make this easier to visualize.
(b) The cutoff frequency of the ideal HPF is ω̂co .
(c) From part (a) it follows that hhp [n] = δ[n] − hlp [n] so
sin(ω̂co n)
hhp [n] = δ[n] −
πn
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.12 DSP First 2e
(a) To get a bandpass filter, we need to subtract the low frequency band from a wider lowpass filter. Thus, we can write
 0 |ω̂| < 0.3π

|ω̂| ≤ 0.6π |ω̂| ≤ 0.3π

1 1 
Hbp (e j ω̂ ) = 
  
= 1 0.3π ≤ |ω̂| ≤ 0.6π

− 
0 0.6π < |ω̂| ≤ π 
0 0.3π < |ω̂| ≤ π 
0.6π < |ω̂| ≤ π

0



Therefore, the cutoff frequencies of the two lowpass filters are
ω̂lp1 = ω̂co2 = 0.6π, and ω̂lp2 = ω̂co1 = 0.3π.
(b) hbp [n] = −
πn πn
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.13 DSP First 2e
(a) A sketch would show a negative exponential decaying to the left for n ≤ −1.
−1
X
(b) The DTFT exists for values of b such that |x[n]| is absolutely summable; i.e., | − bn | < ∞.
−∞
This will be true for |b| > 1.
(c) By definition, the DTFT of x[n] = −bn u[−n − 1] when |b| > 1 is
−1 ∞ ∞
X (e j ω̂ ) = −bn e−j ω̂n = − b−n e j ω̂n = 1 − b−n e j ω̂n
X X X
n=−∞ n=1 n=0
1 −b−1 e j ω̂
=1− =
1 − b−1 e j ω̂ 1 − b−1 e j ω̂
−b−1 e j ω̂ 1
= =
−b e (1 − be ) 1 − be−j ω̂
−1 j ω̂ −j ω̂
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.14 DSP First 2e
∞ 0.25π
2 Z
X sin(0.25πn) 1 1 1
(a)
= d ω̂ =
n=−∞
3πn 2π 3 12
−0.25π
 1/3 0 ≤ |ω̂| ≤ 0.25π

sin(0.25πn) DTFT 
since ←→ 
3πn 0
 0.25π < |ω̂| ≤ π
Zπ 2
(b) 4 sin(ω̂5) e−j ω̂9/2 d ω̂= 20
sin(ω̂/2) π
−π
9
X DTFT sin(ω̂5) −j ω̂9/2
since 2δ[n − k] ←→ 2 e
k=0
sin(ω̂/2)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.15 DSP First 2e
(a) This one is easy with the DTFT since X1 (e j ω̂ ) = e−j ω̂4 and Cx1 x1 (e j ω̂ ) = |X1 (e j ω̂ )| 2 = 1. Therefore, cx1 x1 [n] = δ[n].
In the index domain, it is also easy: cx1 x1 [n] = x 1 [n] ∗ x 1 [−n] = δ[n − 4] ∗ δ[−n − 4] = δ[n]
(b) This one can be done easily by discrete convolution, but let’s use the DTFT. Since X2 (e j ω̂ ) = e−j ω̂4 − e−j ω̂6 , it follows
that
Cx2 x2 (e j ω̂ ) = |X2 (e j ω̂ )| 2 = (e−j ω̂4 − e−j ω̂6 )(e j ω̂4 − e j ω̂6 ) = −e j ω̂2 + 2 − e−j ω̂2
so cx2 x2 [n] = −δ[n + 2] + δ[n] − δ[n − 2].
(c) Let’s try this with the DTFT. First note that x 3 [n] is identical to the impulse response of a 10-point running sum system
with additional delay of 6 samples. Therefore,
!2
j ω̂ sin(ω̂5) −j ω̂4.5 −j ω̂6 j ω̂ j ω̂ 2 sin(ω̂5)
X3 (e ) = e e and Cx3 x3 (e ) = |X3 (e )| = .
sin(ω̂/2) sin(ω̂/2)
(Note that the exponential factors corresponding to delay disappear when we take the magnitude.)
To determine cx3 x3 [n] we have to determine the inverse DTFT. We cannot evaluate the integral with what we know and
we have no entry in our table of transforms for this function. Therefore, we should try convolution, which is not too
s
difficult. The result is w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
 10 − |n| −9 ≤ n ≤ 9
in
e
em

co
d
cx3 x3 [n] = 
s
i
W
r
ss
to
es
rld
0 otherwise
c
i
at
.D
ru
o
St
.
st
ed
ng
W

in
d
itt
ni
e
te
of
th
m
ar
ni
er
This might be a good transform pair to know in a general case. In particular, you can show the following for any
e
le
on
U
us
tp
t
by
en
g
he
no
in
sequence x[n] (including an autocorrelation function):

ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
!2
sin(ω̂L/2)
es
so
or
or
 L − |n| |n| ≤ (L − 1)
is
DTFT j ω̂
w
s

w
x[n] =  )=
ed
k
as
←→ X (e
or
e
id
hi
sin(ω̂/2)
th
w
d
ov
0 otherwise
t
an
of
of
is
pr
Th
rt

se
rit
is
pa
g
ur
d
te
y
an
co
an
in
 1/9 |ω̂| < 0.25π

r
e
of
ei
sin(0.25πn) DTFT
th
←→ X4 (e j ω̂ ) = 
th

(d) Recall the DTFT pair: x 4 [n] =
e
y
l
sa
ro
9πn 0.25π < |ω̂| ≤ π

st
0
or
de

ill
w
This one is a clear candidate for using the DTFT since it would be impossible to evaluate the discrete-time autocorre-
lation in the time domain.
We should recognize that x 4 [n] is identical to the impulse response of an ideal lowpass filter with cutoff frequency
0.25π and gain 1/9. If we take the squared-magnitude of X4 (e j ω̂ ), we get the same cutoff but the gain is squared.
sin(0.25πn)
Therefore, it follows that cx4 x4 [n] =
81πn

CHAPTER 7. DTFT
P-7.16 DSP First 2e
(a) This is accomplished by evaluating the discrete-time convolution cxx [n] = x[n] ∗ x[−n] and the result is cxx [n] =
 5 − |n| |n| ≤ 4 .


0 otherwise

(b) To do this, we don’t try to find the DTFT of cxx [n] in part (a) directly, but instead use the fact that Cxx (e j ω̂ ) = |X (e j ω̂ )| 2 .
We can look up the DTFT of x[n] in our table of DTFT pairs with L = 5. Thus
sin(ω̂5/2) 2 sin(ω̂5/2)
!2
j ω̂ −j ω̂(L−1)/2
Cxx (e ) = e =
Cxx (e j ω̂ ) is clearly real and positive because it is a squared-magnitude of the complex DTFT X (e j ω̂ ).
(c) Since Cyy (e j ω̂ ) = |Y (e j ω̂ )| 2 = |X (e j ω̂ )e−j ω̂2 | 2 = |X (e j ω̂ )| 2 , it follows that cyy [n] = cxx [n]. In fact the auto correlation
function for a delayed signal is always the same as the autocorrelation function for the signal itself. Delay always
cancels out when you evaluate a correlation.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.17 DSP First 2e
(a) Plot shows details of the passband and stopband magnitudes.
Magnitude of Frequency Response of Lowpass Filter M = 32

1.5
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Detail of Stopband: d1 = 0.0021663 ! s = 0.4932:

#10 -3
6
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
0 0.1 0.2 0.3 0.4 0.5 0.6
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
Detail of Passband: d2 = 0.0029736 ! p = 0.7034: rld

c
i
at
.D
ru
o
St
.
st
1.004
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
1.002
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
1
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
0.998
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
0.996
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
0.994
s
rt
y
se
rit
is
pa
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1

g
ur
d
te
y
an
co
!/:
an
in
r
e
of
ei
th
th
y
l
sa
ro
Figure 7-1: Solution for Problem 7-17(a).

st
or
de
ill
w
(b) From the plots the measured parameters are ω̂ p = 0.7034π, ω̂s = 0.4932π, δ p = 0.003 and δ s = 0.0022.
(c) A good rule of thumb is that with a Hamming window, the transition region satisfies ∆ω̂ M = 8π/M, where M
is the order of the filter. For M = 32, this formula gives ∆ω̂32 = 0.25π, whereas the measured value is ∆ω̂32 =
(0.7034 − 0.4932)π = 0.2102π, so the formula over-estimates the transition width in this case.
(d) However, the formula gives us a rationale for estimating the effect of going from M = 32 to M = 80. We should have
∆ω̂80 ≈ ∆ω̂32 32/80 = 0.2102π(32/80) = .0841π. The formula would give a value for ∆ω̂80 of 0.1π. The measured
cutoff frequencies for M = 80 are ω̂ p = (.6 + .0841/2)π = 0.642π and ω̂s = (0.6 − .0841/2)π = 0.558π.

CHAPTER 7. DTFT
1
0.8
0.6
0.4
0.2
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
s
Detail of Stopband: d1 = 0.0020404 ! s = 0.5584: w
ng
#10 -3
la
hi
)
n
ht
eb
5
c
tio
ea
rig
W
a
t
py
in
in
e
4 em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
3
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
2
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
1
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
0
an
y
n
o
(
l
le
si
pr
0 0.1 0.2 0.3 0.4 0.5 0.6

k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
Detail of Passband: d2 = 0.0018343 ! p = 0.6418:

th
w
d
ov
t
an
of
of
is
pr
Th
1.005
s
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
1
in
r
e
of
ei
th
th
0.995
l
sa
ro
st
or
de
0.99
ill
w
0.985
0.98
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
!/:
Figure 7-2: Solution for Problem 7-17d. The estimates of the cutoff frequencies from part (c) are seen to be quite accurate.

CHAPTER 7. DTFT
P-7.18 DSP First 2e

To
solve this problem use Matlab to design the filter using filterdesign. Save the impulse response to the work space as
h24 for use. The solution to each of the following parts requires that we evaluate the frequency response at a given frequency.
The frequencies are 0, 0.5π and 0.75π. This can be done in many ways, but the most straightforward way is to use Matlab’s
freqz( ) function as follows:
>> H24 = freqz(h24,1,[0,0.5*pi,0.75*pi])

H24 =
Columns 1 through 2
0.000134504777373 + 0.000000000000000i 0.041265977941552 + 0.000000000000000i
Column 3
-0.996401749465568 - 0.000000000000001i
We will need the magnitude and angle at these three frequencies so we do the following:
>> zprint(H24)
Z = X + jY Magnitude Phase Ph/pi Ph(deg)
0.0001345 0 0.0001345 0.000 0.000 0.00
0.04127 7.164e-17 0.04127 0.000 0.000 0.00
-0.9964 -1.142e-15 0.9964 -3.142 -1.000 -180.00
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
(a) If the input is x 1 [n] = 10 for all n, the output is y1 [n] = 10H24 (e j0 ), so we use Matlab as above to find H24 (e j0 ).H32=sum(h32)=
a
t
py
in
in
e
em
co
d
s
i
so y1 [n] = 0.001345. This is a highpass filter so the DC is greatly attenuated.
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
(b) When the input is a zero-phase cosine, the output is y2 [n] = 10|H24 (e j0.5π )| cos(0.5πn + ∠H24 (e j0.5π )), so from the
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
Matlab analysis, y2 [n] = 0.4127 cos(0.5πn + 0). This frequency is in the transition zone near the stopband so its
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
amplitude is much reduced.

c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
(c) As in part (b) the output in this case is y3 [n] = 10|H24 (e j0.75π )| cos(0.75πn + ∠H24 (e j0.75π )), so from the Matlab
w
s
w
ed
k
as
or
e
id
hi
th
w
analysis, y3 [n] = 9.964 cos(0.75πn − π). This frequency is in the passband so the amplitude is barely reduced.
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 7. DTFT
P-7.19 DSP First 2e

A
Matlab program to make the requested plots is as follows:
% solution to problem 7-19

% assumes that h37 has been saved in workspace by filterdesign GUI
% h37 = fir1(37,0.3,ones(1,38);
n=0:99;
x1=10*ones(1,100);
x2=10*cos(0.5*pi*n);
y1=filter(h37,1,x1);
sn=x1+x2;
snOutput=filter(h37,1,sn);
%
subplot(411)
stem(n,y1,’filled’);hold on; plot([37,37],[-9,14.5],’r--’);hold off
title(’Output y_1[n] due to x_1[n] = 10’)
text(20,1,5,’Transient region’);hold off
subplot(412)
s
w
ng
la
stem(n,y2,’filled’);hold on; plot([37,37],[-10,15],’r--’);hold off
hi
)
n
ht
eb
c
tio
ea
rig
W
title(’Output y_2[n] due to x_2[n] = 10cos(0.5 \pi n)’)
a
t
py
in
in
e
em
co
d
s
i
subplot(413)
W
r
ss
to
es
rld
c
i
at
.D
ru
o
stem(n,y1+y2,’filled’);hold on; plot([37,37],[-10,15],’r--’);hold off
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
title(’Sum of Outputs’)
ni
er
e
le
on
U
us
tp
t
by
en
subplot(414)
he
no
in
ud
d
ud
rt
te
is
st
c
fo
stem(n,snOutput,’filled’)
cl
d
te
in
an
y
n
o
(
l
le
si
pr
title(’Output Due to Sum of Inputs’)

es
so
or
or
is
w
s
w
ed
k
as
or
xlabel(’time index n’)

id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
hold on
s
rt
y
se
rit
is
pa
g
ur
d
te
y
an
han=stem(n,snOutput-y1-y2);hold on; plot([37,37],[-10,15],’r--’);hold off

co
an
in
r
e
of
ei
th
th
y
l
sa
ro
The resulting plot is shown in the figure above.

st
or
de
ill
w
(a) When the input x 1 [n] is DC with a DC level of 10, the output level y1 [n] converges to 10H (e j0 ) = 10(1) = 10 after a
brief transient of less than 37 samples.
(b) When the input is the sinusoid x 2 [n] = 10 cos(0.5πn) we must evaluate H (e j ω̂ ) at ω̂ = 0.5π rad. We obtain H (e j0.5π ) =
0.02814e j0.75π , so the output would be y2 [n] = 0.2814 cos(0.5πn + 0.75π).
(c) Note that the bottom two figures are identical, thereby showing that superposition holds. This is also shown by the
open circles, which show the difference between y1+y2 and snoutput.

CHAPTER 7. DTFT
Output y [n] due to x [n] = 10

1 1
10
0
Transient region is n=0 to 37
−10
0 10 20 30 40 50 60 70 80 90 100
Output y [n] due to x [n] = 10cos(0.5 π n)
2 2
2
s
w
ng
1
la
hi
)
n
ht
eb
c
tio
ea
rig
W
0
a
t
py
in
in
e
em
co
d
s
i
W
−1
r
ss
to
es
rld
c
i
at
.D
0 10 20 30 40 50 60 70 80 90 100
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
Sum of Outputs
er
e
le
on
U
us
tp
t
by
en
15
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
10
d
te
in
an
y
n
o
(
l
le
si
pr
5
es
so
or
or
is
w
s
w
ed
k
as
or
0
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
0 10 20 30 40 50 60 70 80 90 100
s
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
Output Due to Sum of Inputs

r
e
of
ei
th
th
15
y
l
sa
ro
st
or
de
10
ill
w
5
0
0 10 20 30 40 50 60 70 80 90 100
Time Index (n)
Figure 7-3: Solution for Problem 7-19.

CHAPTER 7. DTFT
P-7.20 DSP First 2e
(a) Use the GUI to find this frequency to be approximately 0.712π rad. The unwrapped phase at this frequency is
ϕ ≈ −10.5 rad. The principal value of the phase is −10.5 + 2π ≈ 2.01 rad.
(b) Here again, you can use the GUI to find the frequencies where |H (e j ω̂ )| = 1.
These are approximately {0.71π, 0.788π, 0.854π, 0.914π, 0.972π}
(c) Use the GUI to find the frequencies where |H (e j ω̂ )| = 0.

These are approximately {0.03π, 0.091π, 0.151π, 0.212π, 0.272π, 0.334π, 0.398π, 0.486π}.
The following figure shows the places where the frequency response is one and zero more accurately than is possible
with the GUI. This plot can be constructed by saving the impulse response to the work space and then using freqz to
compute the frequency response.

1.5
1
s
w
ng
la
hi
)
n
ht
eb
c
tio
0.5 ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
0
ru
o
St
.
st
ed
ng
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U

us
tp
t
#10 -3
by
en
g
he
no
in
ud
6
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
4
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
2
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
0
y
l
sa
ro
0 0.1 0.2 0.3 0.4 0.5 0.6

st
or
de
ill

w
1.002
1
0.998
0.996
0.994
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
!/:

Chapter 8
DFT
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
155
CHAPTER 8. DFT
P-8.1 DSP First 2e
(a) The DTFT of x 0 [n] is X0 (e j ω̂ ) = 1 for all ω̂. Therefore, X0 [k] = X0 (e j (2πk/10) ) = 1 for k = 0, 1, . . . , 9. In Matlab
>> X0=fft([1,zeros(1,9)])
X0 = 1 1 1 1 1 1 1 1 1 1
1 − e−j ω̂10
9
(b) The DTFT of x 1 [n] is X1 (e j ω̂ ) = e−j ω̂n =
X
. Thus, for k = 0, 1, . . . , 9 we get
n=0
1 − e−j ω̂
1 − e−j2πk  10
 k=0
X1 [k] = X1 (e j (2π/10)k ) = =
1−e −j (2π/10)k 0
 k = 1, 2, . . . , 9
1 − e−j2πk
Note that the closed form X1 [k] = is indeterminant when we try to evaluate for k = 0. L’ Hôpital’s
1 − e−j (2π/10)k
rule doesn’t work here because k is an integer variable. Therefore we need to go back to the sum form where we see
9
X
that X1 [0] = X1 (e j (2π/10)0 ) = 1 = 10.
s
n=0 w
ng
la
hi
)
n
ht
eb
c
Matlab verification:
tio
ea
rig
W
a
t
py
in
in
e
>> X1 = fft(ones(1,10)) em
co
d
s
i
W
r
ss
to
es
rld
c
X1 = 10 0 0 0 0 0 0 0 0 0
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
(c) The DTFT of x 2 [n] is X2 (e j ω̂ ) = e−j ω̂4 so X2 [k] = e−j (2π/10)k4

er
e
k = 0, 1, . . . , 9.
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
>> X2 = fft([zeros(1,4),1,zeros(1,5)])
es
so
or
or
is
w
s
w
ed
k
as
or
X2 = 1.0000 + 0.0000i -0.8090 - 0.5878i 0.3090 + 0.9511i

id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
0.3090 - 0.9511i -0.8090 + 0.5878i 1.0000 + 0.0000i

s
rt
y
se
rit
is
pa
g
ur
d
te
y
an
-0.8090 - 0.5878i 0.3090 + 0.9511i 0.3090 - 0.9511i

co
an
in
r
e
of
ei
th
-0.8090 + 0.5878i
th
y
l
sa
ro
st
or
de
1 − e−j (ω̂−2π/5)10
9 9
ill
(d) The DTFT of x 3 [n] is X3 (e j ω̂ ) = e j2πn/5 e−j ω̂n = e−j (ω̂−2π/5)n =

w
X X
.
n=0 n=0
1 − e−j (ω̂−2π/5)
Therefore,
 0 k = 0, 1
1 − e−j ((2π/10)k−2π/5)10 1 − e−j2π(k−2)



X3 [k] = X3 (e j (2π/10)k
)= = = k=2

 10
1 − e−j ((2π/10)k−2π/5) 1 − e−j (2π/10)(k−2) 
k = 3, 4, . . . , 9

0


As in part (b), we have to be careful in evaluating this expression for k = 2.
>> X3 = fft(exp(j*2*pi*(0:9)/5))
X3 = -0.0000 + 0.0000i -0.0000 + 0.0000i 10.0000 - 0.0000i
0.0000 + 0.0000i -0.0000 + 0.0000i 0.0000 + 0.0000i
-0.0000 + 0.0000i 0.0000 + 0.0000i 0.0000 + 0.0000i
-0.0000 + 0.0000i

CHAPTER 8. DFT
P-8.2 DSP First 2e
(a) Substituting into the inverse DFT definition we get,

9
1 X
x a [n] = δ[k]e j (2π/10)kn = 0.1 n = 0, 1, . . . , 9
10 k=0
>> xa = ifft([1,zeros(1,9)])
xa = 0.1000 0.1000 0.1000 0.1000 0.1000 0.1000
0.1000 0.1000 0.1000 0.1000
(b) This one is the first entry in Table 8-1, so we can write the answer down by inspection: x b [n] = δ[n]. Alternatively,
you can plug into the inverse DFT definition and get
9
1 X j (2π/10)kn 1 − e j (2π/10)10n 1 n=0
x b [n] = e = = = δ[n]
10 k=0 1−e j (2π/10)n 0
 n = 1, 2, . . . , 9
>> xb = ifft(ones(1,10))
xb = 1 0 0 0 0 0 0 0 0 0
s
w
ng
la
hi
)
n
ht
eb
c
tio
(c) Since there are only two nonzero terms, we can write down the answer directly and manipulate it if possible. ea
rig
W
a
t
py
in
in
e
em
co
d
x c [n] = 0.1(e j (2π/10)3n + e j (2π/10)7n )
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
= 0.1(e j (2π/10)3n + e j (2π/10)(10−3)n )
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
= 0.1(e j (2π/10)3n + e j (2π/10)10n e−j (2π/10)3n )

us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
= 0.1(e j (2π/10)3n + e−j (2π/10)3n )

st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
= 0.2 cos((2π/10)3n)
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
>> xc = ifft([0,0,0,1,0,0,0,1,0,0])
g
ur
d
te
y
an
co
an
in
xc = 0.2000 -0.0618 -0.1618 0.1618 0.0618 -0.2000

r
e
of
ei
th
th
y
l
sa
0.0618 0.1618 -0.1618 -0.0618

ro
st
or
de
ill
w
(d) We can observe that

X d [k] = cos(2πk/5)X b [k]
= 12 e j (2π/10)2k X b [k] + 12 e−j (2π/10)2k X b [k]
= 12 e j (2π/10)2k X b [k] + 12 e j (2π/10)(10−2)k X b [k]
= 12 e j (2π/10)2k X b [k] + 12 e j (2π/10)8k X b [k]
so by the time-shift property of Table 8-2,
x d [n] = 12 x b [n − 2] + 12 x b [n − 8] = 12 δ[n − 2] + 21 δ[n − 8]
>> xd = ifft(cos(2*pi*(0:9)/5))
xd = -0.0000 + 0.0000i -0.0000 + 0.0000i 0.5000 + 0.0000i
0.0000 - 0.0000i -0.0000 - 0.0000i 0.0000 + 0.0000i
-0.0000 + 0.0000i 0.0000 + 0.0000i 0.5000 - 0.0000i
-0.0000 - 0.0000i

CHAPTER 8. DFT
P-8.3 DSP First 2e
(a) Express (−1) n as e jπn = e j (2π/12)6n and substitute into the DFT definition.
11 11
X X 1 − e−j2π (k−6)
Y0 [k] = 3 e j (2π/12)6n e−j (2π/12)kn = 3 e−j (2π/12)(k−6)n = 3
n=0 n=0
1 − e−j (2π/12)(k−6)



 0 k = 0, 1, . . . , 5
= k=6

 36
k = 7, 8, . . . , 11

0


Note that the numerator of the closed form is zero for all integer k, but when k = 6 we have an indeterminant form
that evaluates to 3 × 12 = 36.
>> Y0 = 3*fft((-1).^(0:11))
Y0 = 0 0 0 0 0 0 36 0 0 0 0 0
(b) This can be looked up in Table 8-1. It is the third entry with L = 4 and N = 12 so
s
w
ng
la
hi
)
n
ht
eb
c
sin(πk/3) −jπk/4
tio
ea
rig
Y1 [k] =
W
a
e
t
py
in
in
e
sin(πk/12) em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
>> Y1 = fft([ones(1,4),zeros(1,8)])
ud
rt
te
is
st
c
fo
cl
d
te
in
an
Y1 = 4.0000 + 0.0000i 2.3660 - 2.3660i 0.0000 - 1.7321i

y
n
o
(
l
le
si
pr
k
es
so
or
or
is
0.0000 + 0.0000i 1.0000 + 0.0000i 0.6340 - 0.6340i

w
s
w
ed
k
as
or
e
id
hi
th
w
0.0000 + 0.0000i 0.6340 + 0.6340i 1.0000 + 0.0000i

ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
0.0000 + 0.0000i 0.0000 + 1.7321i 2.3660 + 2.3660i

g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
(c) In this problem, the even samples are 1 and the odd indexed samples are zero. This solution is facilitated if we write
y
l
sa
ro
st
or
11
de
X
y2 [n] as y2 [n] = 2 1 + e δ[n−m]. Therefore, it follows from the linearity and frequency-shift properties
ill
1 j (2π/12)6n
w
m=0
in Table 8-2 and the third entry in Table 8-1 with L = N = 12 that
1 sin(πk) 1 sin(π(k − 6))

Y2 [k] = 2 sin(πk/12) e
−jπk11/12
+ 2 sin(π(k − 6)/12) e
−jπ (k−6)11/12
The first term is zero except at k = 0 where its value is Y2 [0] = 6. The second term is zero except when k = 6 where
its value is Y2 [6] = 6. Thus
6
 k = 0, 6
Y2 = 
0
 k = 1, 2, . . . , 5, 7, 8, . . . , 11
We can get another closed form expression by summing the DFT expression only over the even indices using a index
m = 2n, n = 0, 1, . . . , 5. This gives us
5
X 1 − e−j (2π/12)6k 1 − e−jπk
Y2 [k] = e−j (2π/12)2km = =
m=0
1 − e−j (2π/12)2k 1 − e−j (2π/6)k

CHAPTER 8. DFT
11
X
In this case, the expression is indeterminant for k = 0, 6. The value at these DFT indices is Y2 [0] = y2 [n] = 6, and
n=0
because the odd-indexed samples are zero,
11
X 11
X
Y2 [6] = y2 [n]e−j (2π/12)6n = y2 [n]e−jπn = 6
n=0 n=0
>> Y2 = fft([1,0,1,0,1,0,1,0,1,0,1,0])
Y2 = 6 0 0 0 0 0 6 0 0 0 0 0
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.4 DSP First 2e
(a) The DFT of y[n] is

5
X 1 − e−j2πk
Y [k] = 7e−j (2π/6)kn − 3W [k] = 7 − 3W [k] = 7(6)δ[k] − 3W [k]
n=0
1 − e−j (2π/6)k
Therefore, Y [k] = {39, 0, −3 j, 0, +3 j, 0}
(b) If we write v[n] = 4(−1) n w[n] as v[n] = 4e j (2π/6)3n w[n], we see that the new DFT is V [k] = 4W [k − 3] where the
shift is periodic with period 6.
Therefore, V [k] = {0, −4 j, 0, 4, 0, 4 j}
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.5 DSP First 2e
255
X 255
X
(a) X[0] = x[n]e−j2π (0)n/256 = x[n] is real since all x[n] are real-valued.
n=0 n=0
255
X 255
X
X[128] = x[n]e−j2π (128)n/256 = x[n](−1) n is real since all x[n] are real-valued.
n=0 n=0
(b) X[k] = 3 − e−j (2π/4)k + 3e−j (2π/4)2k − e−j (2π/4)3k = {4, 0, 8, 0}

3
1 X
(c) v[2] = V [k]e j (2π/4)(2)k = 14 (3 + (5 j)(−1) + 0 + (−5 j)(−1)) = 3
4
N k=0
9
X 9
X
(d) Y [0] = y[n] = 0 Y [5] = 7(−1) n e−j (2π/10)5n = 70
n=0 n=0
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.6 DSP First 2e
(a) Sketch should show x[n] = {0, 1, 2, 3, 4, 0, 0, 0} starting at n = 0.
(b) Sketch should show y[n] = {2, 3, 4, 0, 0, 0, 0, 1} because y[n] = x[n − 6] evaluated with a period of 8.
Problem 8.6(a), x[n]
4
3
2
1
0
0 1 2 3 4 5 6 7
Problem 8.6(b), y[n]

4
3
s
2 w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
1
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
0 rld
c
i
at
.D
ru
o
St
.
st
ed
ng
0 1 2 3 4 5 6 7
in
d
itt
ni
e
te
of
th
m
ar
ni
Time Index (n)

er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.7 DSP First 2e
(a) Plot y[n] = {2, 1, 0, 0, −2, −1} starting at n = 0.
Problem 8.7, y[n]

2
1
0
−1
−2
0 1 2 3 4 5
Using the method of synthetic polynomial multiplication to evaluate the convolution, we get
1 1 1 1
2 -1 -1 so y[n] = 2δ[n] + δ[n − 1] − 2δ[n − 4] − δ[n − 5]
2 1 0 0 -2 -1
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
(b) Applying the definition of the DFT and writing out the individual terms gives
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
X[k] = 1 + e−j (2π/N )k + e−j (2π/N )k (2) + e−j (2π/N )k3 + e−j (2π/N )k (4)
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
H[k] = 2 − e−j (2π/N )k − e−j (2π/N )k (2)

th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
(c) Forming the product Y [k] = H[k]X[k] is the same as doing the polynomial multiplication as in part (a) so we get
d
te
in
an
y
n
o
(
l
le
si
pr
Y [k] = 2 + e−j (2π/N )k − 2e−j (2π/N )k (4) − e−j (2π/N )k (5) .

es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
(d) For N = 6, the DFT is Y [k] = 2 + e−j (2π/6)k − 2e−j (2π/6)k (4) − e−j (2π/6)k (5) . All we need to do is pick off the coeficients
pr
Th
rt
y
se
rit
is
pa
of Y [k] to obtain y[n] = {2, 1, 0, 0, −2, −1}

ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
(e) We need to adjust the exponents of W [k] so that any power of e−j (2π/4)k (n+4) is replaced by e−j (2π/4)kn .
sa
ro
st
or
de
Thus, Y [k] = 2 + e−j (2π/4)k − 2e−j (2π/4)k (4) − e−j (2π/4)k (5) is equivalent to
ill
w
Y [k] = 2 + e−j (2π/4)k − 2e−j (2π/4)k (0) − e−j (2π/4)k = 0.
(f) Since the length of the convolution is L + M − 1, we need to be able to represent the sequence of this length by the
DFT. Therefore, N ≥ L + M − 1 is required.

CHAPTER 8. DFT
P-8.8 DSP First 2e
In general, the 100-point DFT would be

sin(12.5(2π/100)k)
X[k] = X (e j (2π/100)k ) = 7e−j (2π/100)18k
sin(πk/100)
If this is evaluated for k = 13 we get X[13] = X (e j (2π/100)13 ) = 12.46 e j0.32π = 12.46 e j1.005
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.9 DSP First 2e
(a) From Q(e j ω̂ ) it is clear that the peaks will occur around DFT indices corresponding to the frequencies ω̂0 and 2π − ω̂0 ;
i.e. when (2π/32)k 1 = ω̂0 and (2π/32)(32 − k 1 ) = 2π − ω̂0 . Therefore, ω̂0 = (2π/32)10 = 0.625π.
(b) The second peak occurs when (2π/32)(32 − k1 ) = (2π/32)k 2 or when k2 = 22. The third nonzero value occurs at
k 3 = 0, corresponding to the DC component of 0.1 in s[n]. The DFT of that DC component is 0.1N δ[k], which is
zero for all k except k = 0. The Dirichlet components are also zero except at k 1 = 10 for the first one and k 2 = 22 for
the second one.
(c) The peak magnitude at k = 10 is 21 ( A)(32) = 50 so A = 100/32 = 3.125.

>> x=0.1+(100/32)*cos(2*pi*10*(0:31)/32);
>> X=fft(x);
>> abs(X)
ans = 3.2000 0.0000 0.0000 0.0000 0.0000 0.0000
0.0000 0.0000 0.0000 0.0000 50.0000 0.0000
0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
0.0000 0.0000 0.0000 0.0000 50.0000 0.0000 s
w
ng
la
hi
)
n
ht
eb
c
tio
0.0000 0.0000 0.0000 0.0000 0.0000 ea 0.0000
rig
W
a
t
py
in
in
e
em
co
0.0000 0.0000
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.10 DSP First 2e
(a) N = 32 + 48 = 80.
(b) k peaks = 8, and −8 → (80 − 8 = 72) because (2π/80)k peaks = ±0.2π

The peaks will occur around DFT indices corresponding to ω̂ = ±0.2π, or when (2π/80)k = 0.2π or k = 8 and when
(2π/80)(80 − 8) = 2π − 0.2π or k = 72.
(c) max{|R[k]|} = 118.4553. Evaluate in Matlab via

k
r=1+7*cos(0.2*pi*(0:31)); R=fft(r,80); [mx,ix]=max(abs(R))
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.11 DSP First 2e

sin(0.15πn)
The impulse response h[n] = corresponds to a frequency response
5πn
 1/5 |ω̂| ≤ 0.15π

H (e j ω̂ ) = 

0
 0.15π < |ω̂| ≤ π
The input signal has fundamental frequency ω̂0 = 0.08π. Thus, only DC and the first harmonic are below the cutoff of
the filter. Since the gain is 1/5, we get three terms in the output

y[n] = (1/5) 1 + 2e j0.08πn + 2e−j0.08πn = 0.2 + 0.4 cos(0.08πn)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.12 DSP First 2e
(a) In order to show the spectrum only for positive frequencies, we need to convert the frequencies by adding 2πn to the
exponent of each of the components having negative frequency. Therefore,
x[n] = 3 + 2e j0.2πn + 2e j (2π−0.2π)n − 7 je j0.7πn + 7 je j (2π−0.7π)n

= 3 + 2e j0.2πn + 2e j1.8πn − 7 je j0.7πn + 7 je j1.3πn
Therefore, the spectrum plot is as follows:
7j −7 j 7j
3 3
2 2 2
π
-
−π −.7π −.2π 0 .2π .7π 1.3π 1.8π 2π ω̂
The dashed lines show negative frequencies that differ by 2π from frequencies in the range 0 ≤ ω̂ < 2π. w
s
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
(b) If we carry out the multiplication and adjust any resulting negative frequencies as we did in part (a), we get
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
x 1 [n] = e j0.4πn x[n] = 3e j0.4πn + 2e j0.6πn + 2e0.2πn − 7 je j1.1πn + 7 je j (−0.3π)n

o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
= 3e j0.4πn + 2e j0.6πn + 2e0.2πn − 7 je j1.1πn + 7 je j1.7πn

er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
This is an example of the frequency shift property since by adding 0.4π to each frequency we shift the spectrum to the
g
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
right by that amount. The resulting spectrum is

w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
−7 j 7j −7 j 7j
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
3
ill
w
2 2
π 1.1π
-
−.9π −.3π 0 .2π .4π .6π 1.7π 2π ω̂
(c) Now we can write x 2 [n] = (−1) n x[n] = e jπn x[n] and we see that by the same process as in part (b), each of the original
frequencies will be increased by π radians. If we choose the alias frequencies that are in the range 0 ≤ ω̂ < 2π, we can
write
x[n] = 3e jπn + 2e j1.2πn + 2e j0.8πn − 7 je j1.7πn + 7 je j0.3πn
Therefore the spectrum plot is that of part (a) shifted to the right by π radians.

CHAPTER 8. DFT
−7 j 7j −7 j
3 3
2 2 2
π 1.2π
-
−π −.8π −.3π 0 .3π .8π 1.7π 2π ω̂
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.13 DSP First 2e
(a) The sampled signal is
x[n] = x(n/60) = 6 cos(42πn/60) + 4 cos(18πn/60 − 0.5π)

= 6 cos(2π(7/20)n) + 4 cos(2π(3/20)n − 0.5π)
= 6 cos(0.7πn) + 4 cos(0.3πn − 0.5π)
Since 7 and 3 have no common factors, the period is N = 20 and the fundamental frequency is ω̂0 = 2π/20.
(b) x[n] = 2e−j0.5π e j (2π/20)3n + 2e j0.5π e−j (2π/20)3n + 3e j (2π/20)7n + 3e−j (2π/20)7n
(c) From the answer to part (b), the DFS coefficients are c3 = 2e−j0.5π , c−3 = 2e j0.5π , c7 = 3, c−7 = 3.
(d) Here is the spectrum plot of the DFS:
3 3 3 3
2j −2 j 2j −2 j
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
.3π ω̂
-
a
−π
t
−.7π −.3π
py
0 1.3π 1.7π 2π 2.3π 2.7π 3π

in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
Dashed lines are aliases of the frequencies in the range −π < ω̂ ≤ π.
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.14 DSP First 2e
sin(4ω̂) −j ω̂7/2
(a) The frequency response of this filter is H (e j ω̂ ) = e . The zeros of this function are at frequencies where
sin(ω̂/2)
4ω̂ = πr where r is an integer; i.e., where ω̂ = 0.25π, 0.5π, 0.75π, π. The equivalent values of DFT index k are when
(2π/N )k = 0.25πr or k = Nr/8. For N = 512, these values are multiples of 64.
Problem 8.14
8
6
Magnitude
0
0 32 64 96 128 160 192 224 256
pi/2
s
w
ng
Phase (rad)
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
−pi/2
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
−pi
in
ud
d
ud
rt
te
is
st
c
fo
0 32 64 96 128 160 192 224 256

cl
d
te
in
an
y
n
o
(
l
le
si
pr
DFT Frequency Index (k)

k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
(b) Note that the given frequencies are all multiples of 0.25π so all the frequencies except DC fall at the zeros of H (e j ω̂ ).
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
Therefore, these frequencies do not appear in the output. The gain of the filter at ω̂ = 0 is H (e j0 ) = 8, so the output
co
an
in
r
e
of
ei
th
is y[n] = 3(8) = 24 for all n. You can pick all these values off the plot, or evaluate H (e j ω̂ ) at individual frequencies
th
y
l
sa
ro
st
or
with freqz.
de
ill
w

CHAPTER 8. DFT
P-8.15 DSP First 2e
(a) We have two constraints. To avoid aliasing we need 2(1000) < f s . To meet the frequency spacing condition, we need
f s /N ≤ 5, where N = 2ν .
The combined constraints are given by 2000 < f s ≤ 5 · 2ν , or equivalently 400 < f s /5 ≤ ·2ν .
Thus we require N = 512 = 29 . If we fix N at 512, the condition on f s is 2000 < f s ≤ 2560 Hz.
(b) As discussed in Section 8-7.4.1, the width of the main lobe of the Hann and Hamming windows is approximately
∆ω̂ = 8π/L where L is the window length. Therefore, for L = N/2 = 256, we should expect a main lobe width of
about 8π/256 = π/32 rad. This is equivalent to analog radian frequency of ∆ω = (π/32) f s or using analog frequency
in hertz, ∆ f = f s /64.
Using the constraints on f s obtained in part (a), we get 2000/64 < ∆ f ≤ 2560/64, or 31.25 < ∆ f ≤ 40.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.16 DSP First 2e
(a) Here’s a Matlab plot of the ideal spectrogram. A sketch should look like this.
Problem 8-16a: Ideal Spectrogram
2000
1800
1600
1400
frequency
1200
1000
800
600
400
200
0
0 0.2 0.4 0.6 0.8 1 1.2
time axis
(b) The width of the main lobe of the DTFT of the Hann window is approximately ∆ω̂ = 8π/L, where L is the window
length. Therefore the normalized main lobe width for the L = 200 window is ∆ω̂ = 0.04π. In terms of analog
s
w
ng
cyclic frequency, this would be equivalent to ∆ f = ∆ω̂ f s /(2π) = 0.04π8000/(2π) = 160 Hz. Thus, the spectrogram
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
sketch should show bars about 160 Hz wide and there should be blurred regions of approximately 200 samples (or
t
py
in
in
e
em
co
d
s
i
W
r
equivalently, 200/8000 = 0.025 s) duration at the beginning and end of each sinusoidal component. The following is a
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
Matlab spectrogram of the signal.

ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
Problem 8-16b: Spectrogram with 200-point Window

tp
t
by
en
g
he
2000
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
1800
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
1600
is
w
s
w
ed
k
as
or
e
id
hi
th
1400
w
d
ov
t
an
of
of
is
pr
frequency
Th
rt
1200
se
rit
is
pa
g
ur
d
te
y
an
co
an
1000
in
r
e
of
ei
th
th
800
y
l
sa
ro
st
or
600
de
ill
w
400
200
0
0 0.2 0.4 0.6 0.8 1 1.2
time axis

CHAPTER 8. DFT
P-8.17 DSP First 2e
Note that there are 8 equal duration notes in (approximately) 2 seconds of time. Therefore, each note has duration
2/8 s or 250 ms.
Spectrogram of C Major scale, L D 500; fs D 4 kHz
600
550
500
450
Frequency (Hz)
400
350
300
250
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
200
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

es
rld
c
i
at
.D
ru
Time (s)
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
Figure P-8.17
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.18 DSP First 2e
The ratio of the window lengths is 256/100 ≈ 2.5.

The sketch will look similar to the spectrogram in Fig. P-8.17, but it will have bars that are 2.5 times as wide in the
frequency dimension. The frequency width of the horizontal bars in Fig. P-8.17 is approximately 16 Hz, so it will change to
40 Hz in the new figure.
In the time dimension, the fuzzy regions at the time of frequency change will be shorter—their duration will be 40% of
the duration of those in Fig. P-8.17 which are approximately 0.03 s. Thus, in the new figure the duration of the fuzzy regions
will be approximately 0.01 s.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.19 DSP First 2e
(a) The Hamming window has a main lobe width of approximately ∆ω̂ = 8π/L, where L is the window length. For the
given parameters, this would give a main lobe width equivalent to ∆ f = 4 f s /L = 4(8000)/256 = 125 Hz. Thus, two
sinusoids whose frequencies are farther apart than 125 Hz will show up as two distinct peaks, but if the two frequencies
are closer that 125 Hz they will tend to blend together in the spectrum.
(b) The window length is equivalent to 256/8000 =32 ms so if there are abrupt changes in the waveform, the spectral
characteristics will look blurred over approximately 32 ms around the abrupt change.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.20 DSP First 2e
(a) The Hamming window has a main lobe width of approximately ∆ω̂ = 8π/L, where L is the window length. In terms
of analog cyclic frequency this corresponds to ∆ f = 4 f s /L, where L is the window length. For the given parameters,
we want 40000/L = 250 or L = 160.
(b) Here is the plot of the DTFT magnitude:
DTFT of Windowed Sinusoids at Frequencies 3000 and 3250 Hz

45
40
35
30
25
20
15
10
5
0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
analog frequency in Hz
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 8. DFT
P-8.21 DSP First 2e
(a) The notes are in the octave above A-440, so the frequencies are given by the equation f = 440(2n/12 ) with n =
0, 1, . . . , 12 covering one octave from A-440 to A-880. Here are the notes and their theoretical frequencies. (Subscript
denotes octave number.)
note A4 B4 [ B4 C5 C5 ] D5 D5 ] E5 F5 F5 ] G5 G5 ] A5
f 440 466 494 523 554 587 622 659 698 740 784 831 880
Now the first 9 notes of Für Elise are { E5 , D5 ], E5 , D5 ], E5 , B4 , D5 , C5 , A4 }. Therefore, the theoretical frequencies
for the first nine notes are { 659, 622, 659, 622, 659, 494, 587, 523, 440 }.
Spectrogram of Fur Elise (9 notes), L D 1000; fs D 8 kHz
800
750
700
650 E
600
D
Frequency (Hz)
550
C
s
500 B w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
450
W
A
a
t
py
in
in
e
em
co
d
s
i
W
400
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
350
in
d
itt
ni
e
te
E
of
th
m
ar
ni
er
e
le
on
U
300
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
250
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
200
or
is
w
s
w
ed
6 6.25 6.5 6.75 7 7.25 7.5 7.75 8 8.25 8.5 8.75 9

k
as
or
e
id
hi
th
w
Time (s)
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
(b) The spectrogram in the figure above is marked with the true frequencies. Observe that the passage of interest starts
an
in
r
e
of
ei
at t = 6.4 s (approximately). It is difficult to obtain very accurate estimates of the frequencies by measuring on the
th
th
y
l
sa
ro
st
or
spectrogram because the spectral peaks have a width of about 10 Hz. However, careful examination of the figure above
de
ill
confirms that the theoretical note frequencies are close to the center of the spectral peaks.
w
(c) Observe that the passage of interest starts at t = 6.4 s (approximately). From the score in Fig. 3-22, the first eight notes
are 16th notes while note nine is an eighth note, which should have twice the duration.
The first five notes are easy to distinguish, but overlap in time due to the sliding window of the spectrogram. Their total
duration is 7.7 − 6.4 = 1.3 s, so each 16th note has a duration of approximately 0.26 s. The sixth note (B4 ) appears
shorter (≈ 0.15 s), but might be softer so its second half is hard to see; likewise, for note eight (C5 ). The seventh note
(D5 ) is barely visible, but there is evidence of a spectral peak in the 0.2 s gap between B4 and C5 . The ninth note (A4 )
is easy to distinguish, and its duration seems a bit longer (≈ 0.3 s), but not twice as long.

CHAPTER 8. DFT
P-8.22 DSP First 2e
(a) From the spectrogram it appears that ω̂0 = 0.7π and ω̂1 = 0.75π. To estimate the chirp parameter α, recall that the
instantaneous frequency is ω̂i [n] = 2αn. From the spectrogram we see that ω̂i [n] starts at 0 and increases linearly to
ω̂i [2000] = 0.5π = 2α(2000). Therefore, α = π/8000.
(b) The Hann window has a main lobe width of approximately ∆ω̂ = 8π/L, where L is the window length. Thus, if we
multiply the window length by 4, we divide the main lobe width by 4. Thus, the spectrogram with L = 400 will like
Fig. ?? except that the bars will be much narrower—they will have 1/4 the width.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

Chapter 9
z -Transform
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
180
CHAPTER 9. z-TRANSFORM
P-9.1 DSP First 2e
X1 (z) = 7, X2 (z) = z −1, X3 (z) = 5z −4, X4 (z) = 7 − z −1 + 5z −4
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.2 DSP First 2e
Y (z) = X (z) − z −1 X (z) = (1 − z −1 )X (z) = H (z)X (z)
where H (z) = 1 − z −1 is the z-transform of h[n] = δ[n] − δ[n − 1], the impulse response of the system.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.3 DSP First 2e
(a) y[n] = x[n] + 5x[n − 2] − 3x[n − 3] + 2x[n − 5] + 4x[n − 7],
(b) h[n] = δ[n] + 5δ[n − 2] − 3δ[n − 3] + 2δ[n − 5] + 4δ[n − 7],

Plot {1, 0, 5, −3, 0, 2, 0, 4} for 0 ≤ n ≤ 8.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.4 DSP First 2e
(a) H (z) = 0.1(1 − z −1 + z −2 )
(b) The roots of H (z) are e±π/3 so we can write
(z − e jπ/3 )(z − e−jπ/3 )

H (z) = 0.1(1 − e jπ/3 z −1 )(1 − e−jπ/3 z −1 ) = 0.1
z2
0.5
Imaginary Part
2
0
-0.5
-1
-1 -0.5 0 0.5 1
Real Part
(c) H (e j ω̂ ) = 0.1(1 − e−j ω̂ + e−j ω̂2 )

0.3
s
w
ng
la
0.2π
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
0.1
in
e
em
co
d
s
i
W
r
ss
to
es
0
rld
c
i
at
.D
−π π
ru
−0.8π −0.6π −0.4π −0.2π 0 0.2π 0.4π 0.6π 0.8π o

St
.
st
ed
ng
W
5
in
d
H(ej ω̂ )
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
0
g
he
no
in
ud
d
6
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
-5
(
l
le
si
pr
−π −0.8π −0.6π −0.4π −0.2π 0 0.2π 0.4π 0.6π 0.8π π

es
so
or
or
is
w
s
(d)
w
ed
ω̂
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
(e) For this we need the frequency response evaluated at ω̂ = 0, π/4, 2π/3.
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
H (e j0 = 1 H (e jπ/4 ) = 0.4142e−jπ/4 H (e j2π/3 ) = 2e jπ/3

ei
th
th
y
l
sa
ro
st
or
de
Therefore, the output is

ill
w
y[n] = 9 − 8(0.4142) cos[0.25π(n − 1) − π/4] + 7(2) cos[(2π/3)n + π/3]

= 9 − 3.3136 cos[0.25π(n − 2)] + 14 cos[(2π/3)n + π/3]

P-9.5 DSP First 2e
(a) We have to multiply out the factors to get a polynomial in z −1 .
H (z) = 1 + 2z −1 + 2.64z −2 + 2.28z −3 + 1.28z −4 + 0.64z −5
Then the difference equation is:
y[n] = x[n] + 2x[n − 1] + 2.64x[n − 2] + 2.28x[n − 3] + 1.28x[n − 4] + 0.64x[n − 5]
(b) There are five poles at z = 0 and the five zeros are at z = −1, ± j0.9, and e±j2π/3 .
0.5
Imaginary Part
5
0
-0.5
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
-1
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
-1 -0.5 0 0.5 1 rld

c
i
at
.D
ru
o
St
.
st
ed
ng
Real Part
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
(c) We need to find the frequencies where H (e j ω̂ ) = 0.

g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
These frequencies are the angles of the zeros that are on the unit circle; i.e., (1 + z −1 ) = 0, (1 − e j2π/3 z −1 ) = 0 and
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
(1 − e−j2π/3 z −1 ) = 0, or ω̂ = π, ±2π/3.
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.6 DSP First 2e
(a)
H (z) = H2 (z)H1 (z)

= 14 (1 + z −1 + z −2 + z −3 ) 31 (1 + z −1 + z −2 )
= 1
12 (1 + 2z −1 + 3z −2 + 3z −3 + 2z −4 + z −5 )
(b)
Y2 (z) = X (z)H2 (z)

= (1 − z −1 ) 31 (1 + z −1 + z −2 )
= 13 (1 − z −3 )
W (z) = Y2 (z)H1 (z)

= 13 (1 − z −3 ) 14 (1 + z −1 + z −2 + z −3 )
= 1
12 (1 + z −1 + z −2 − z −4 − z −5 − 3z −3 )
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
(c) y[n] = 1
12 (x[n] + 2x[n − 1] + 3x[n − 2] + 3x[n − 3] + 2x[n − 4] + x[n − 5]) rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
Clearly, the samples are not weighted equally although the sum of the weights (coefficients) is equal to one.
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
(d) There are five poles at z = 0, and the five zeros are at z = −1, ± j, and e±j2π/3
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
1
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
0.5
Imaginary Part
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
5
y
l
sa
0
ro
st
or
de
ill
w
-0.5
-1
-1 -0.5 0 0.5 1
Real Part

P-9.7 DSP First 2e
>> z=roots([1,-.5,.5,-1])
z =
1.0000
-0.2500 + 0.9682i
-0.2500 - 0.9682i
>> magz=abs(z)
magz =
1.0000
1.0000
1.0000
>> angz=angle(z)/pi
angz =
0
0.5804
-0.5804
Thus, the zeros are at z = 1, e±j0.5804π and there is one pole at z = 0 for each of the zeros.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
0.5
st
ed
ng
W
Imaginary Part
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
3
en
g
he
no
0
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
-0.5
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
-1
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
0
th
-1 -0.5 0.5 1
e
y
l
sa
ro
Real Part
st
or
de
ill
w

P-9.8 DSP First 2e
(a) Impulse response:

4
X
h[n] = δ[n] − δ[n − 1] + δ[n − 2] − δ[n − 3] + δ[n − 4] = (−1) k δ[n − k]
k=0
4 4
X X 1 − (−z −1 ) 5 1 + z −5 z5 + 1
(b) H (z) = (−1) z k −k
= (−1z −1 ) k = = =
k=0 k=0
1 − (−z −1 ) 1 + z −1 z 4 (z + 1)
(c) Plot the poles and zeros of H (z) in the complex z-plane.
0.5
Imaginary Part
4
0
s
-0.5 w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
-1 em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
-1 -0.5 0 0.5 1
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
Real Part
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
1 + e−j ω̂5
ud
d
cos(ω̂2.5)
ud
rt
te
is
st
(d) H (e j ω̂ ) = = e−j ω̂2

c
fo
cl
d
te
in
an
y
n
o
1+e ω̂ cos(ω̂/2)
(
l
−j
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
(e) Sketch of the frequency response (magnitude and phase) , or Matlab plot.
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
6
th
th
e
|H(ej ω̂ )|
y
l
sa
ro
st
or
de
4
ill
w
0
−π −0.6π −0.2π 0 0.2π 0.6π π
5
H(ej ω̂ )
0
6
-5
−π −0.6π −0.2π 0 0.2π 0.6π π
ω̂
Figure P-9.8
(f) We need the frequency response evaluated at ω̂ = 0, 0.5π, 0.6π. These values are H (e j ω̂ ) = 1, 1, 0.
Therefore, y[n] = 5 + 4 cos(0.5πn).

P-9.9 DSP First 2e
(a) H (z) = (1 − z −1 ) 5
(b) Plot of the poles and zeros of H (z) in the z-plane.
0.5
Imaginary Part
5 5
0
-0.5
-1
-1 -0.5 0 0.5 1
Real Part
s
(c) w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
5
H (e j ω̂ ) = (1 − e−j ω̂ ) 5 = e−j ω̂5/2 = 2e jπ/2 e−j ω̂/2 sin(ω̂/2)
t
py

in
in
e
em
co
d
s
i
W
r
ss
to
es
= 32e (π−ω̂)5/2 sin5 (ω̂/2) rld

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
(d)
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
H (z) = (1 − z −1 ) 5 = 1 − 5z −1 + 10z −2 − 10z −3 + 5z −4 − z −5

(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
h[n] = δ[n] − 5δ[n − 1] + 10δ[n − 2] − 10δ[n − 3] + 5δ[n − 4] − δ[n − 5]

or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.10 DSP First 2e
(a) We can find N1 as the lowest power of z −1 of the product X (z)H (z). That will be N1 = 0 + 6 = 6. Similarly, N2 is the
highest power of z −1 if the product X (z)H (z). That is N2 = 9 + 9 = 18.
(b) Again, we can use the polynomials of X (z) and H (z) to get the answer. y[6] is the product of the coefficients of the
two lowest order terms or y[6] = 2 × 1 = 2. Similarly, y[18] is the product of the coefficients of the highest order
terms, or y[18] = −1 × 3 = −3.
(c) y[n] = 2δ[n − 6] − 4δ[n − 7] + 8δ[n − 8] + δ[n − 9] − 6δ[n − 10] + 12δ[n − 11] − 11δ[n − 12]
+16δ[n − 13] − 32δ[n − 14] + 11δ[n − 15] − 6δ[n − 16] + 12δ[n − 17] − 3δ[n − 18]
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.11 DSP First 2e
(a) H (z) = 1 + z −1 + z −2 = (1 − e j2π/3 z −1 )(1 − e−j2π/3 z −1 ), so there are zeros at z = e±j2π/3 .
(b) To solve this part we need the frequency response evaluated at ω̂ = π/6. That is,
sin(ω̂3/2)
H (e j ω̂ ) = 1 + e−j ω̂ + e−j ω̂2 = e−j ω̂
sin(ω̂/2)
so H (e jπ/6 ) = 2.7321e−jπ/6 and, therefore, the formula for the steady-state output is
y[n] = 2.7321 cos((π/6)(n − 1) − π/4)
(c) omega = 2*pi/3 gives zero for the steady-state output after n ≥ 2. (Until the last two samples.)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.12 DSP First 2e
(a) y[n] = x[n] − x[n − 4]
(b) H (e j ω̂ ) = 1 − e−j ω̂4 = 2e j (π/2−ω̂2) sin(ω̂2)

π/2 − 2ω̂ 0 < ω̂ < π/2
(
(c) |H (e j ω̂ )| = 2| sin(ω̂2)| and ∠H (e j ω̂ ) =
π/2 − 2ω̂ + 2π π/2 < ω̂ < π
(d) The frequencies nulled are the angles of the zeros of H (z); ω̂0 = 0, ±π/2, π.
(e) y[n] = |H (e jπ/3 )| cos(πn/3 + ∠H (e jπ/3 )) = 1.7321 cos(πn/3 − π/6)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.13 DSP First 2e
We should find the output as the sum of the response to DC plus the response to an impulse of size -70 plus the re-
sponse to the sinusoid of frequency 0.5π. We need the impulse response and the frequency response function to do
this.
From H (z) in expanded form, h[n] = δ[n] − 5δ[n − 2] + 4δ[n − 4] and
H (e j ω̂ ) = (1 − e−j ω̂2 )(1 − 4e−j ω̂2 ) = 1 − 5e−j ω̂2 + 4e−j ω̂4
so H (e j0 ) = 0 and H (e j0.5π ) = 1 − 5e−j2π/2 + 4e−j4π/2 = 1 − 5e−π + 4e−j4π/2 = 10. Therefore,
y[n] = −70δ[n] + 350δ[n − 2] − 280δ[n − 4] + 300 cos(0.5πn + π/4)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.14 DSP First 2e
After sampling, we get the sequence x[n] = 4+cos(1000πn/2000− π/4) −3 cos(500πn/2000). Thus, we need to evaluate the
frequency response at ω̂ = 0, π/2, π/4 to find the outputs due to each component of the input. These values are H (e j0 ) = 0,
H (e jπ/2 ) = 0, and H (e jπ/4 ) = 1.0824e jπ/8 . Therefore,
y[n] = −3(1.0824 cos(πn/4 + π/8) = 3.2472 cos(πn/4 − π + π/8) = 3.2472 cos(πn/4 − 2.7489)
and therefore the output of the D-to-C converter with f s = 2000 is
y(t) = 3.2472 cos(π2000t/4 − 2.7489) = 3.2472 cos(500πt − 2.7489)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.15 DSP First 2e
(a) α = 4, β = 10
From H (z) we get H (e j ω̂ ) = 5(1 + .8e−j ω̂ + e−j ω̂2 . By factoring e−j ω̂ out and using the inverse Euler relation we get
H (e j ω̂ ) = e−j ω̂ (4 + 10 cos(ω̂))
(b) In the plot the DC value is +14, H (e jπ ) = −6, and H (e j ω̂ ) = 0 at ω̂ = 0.631π = 1.9823.
15
|H(ej ω̂ )|
10
0
−π −0.8π −0.6π −0.4π −0.2π 0 0.2π 0.4π 0.6π 0.8π π
2
H(ej ω̂ )
0
s
w
ng
6
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
-2
in
e
em
co
d
s
i
−π −0.8π −0.6π −0.4π −0.2π π
W
0 0.2π 0.4π 0.6π 0.8π
r
ss
to
es
rld
c
i
at
.D
ru
ω̂
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
Figure P-9.15
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(c) We want the system function to have the form H (z) = (1−e j ω̂0 z −1 )(1−e−j ω̂0 z −1 ) = 1−2 cos ω̂0 z −1 +z −2 . Therefore, we
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
must set b1 = −2 cos ω̂0 . The normalized frequency that we want to null is ω̂0 = 2π f 0 / f s = 2π(1600/8000) = 0.4π.
k
as
or
e
id
hi
th
w
d
ov
t
an
Thus, we need to set b1 to b1 = −0.618.

of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.16 DSP First 2e
(a) Factoring the two second-order factors gives
H (z)(1 − z −1 )(1 + z −1 )(1 − 0.8e jπ/4 z −1 )(1 − 0.8e−jπ/4 z −1 )(1 + e jπ/2 z −1 )(1 + e−jπ/2 z −1 )
0.5
Imaginary Part
6
0
-0.5
-1
-1 -0.5 0 0.5 1
s
w
ng
la
Real Part
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
Figure P-9.16
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
(b) To satisfy the condition, we must group the zeros at ±1 and e±jπ/2 together to form H1 (z) as in
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
H1 (z) = (1 − z −2 )(1 + z −2 ) = (1 − z −4 )
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
. That leaves the remainder for H2 (z) as

or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
H2 (z) = (1 − 0.8e jπ/4 z −1 )(1 − 0.8e−jπ/4 z −1 ) = 1 − 1.6 cos(π/4)z −1 + 0.64z −2

g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
.
sa
ro
st
or
de
ill
(c) y[n] is the output of H2 (z) when the input is y1 [n] so from H2 (z) we get
w
y[n] = y1 [n] − 1.6 cos(π/4)y1 [n − 1] + 0.64y1 [n − 2]

P-9.17 DSP First 2e
(a) h[n] = 7δ[n] − 14.35δ[n − 1] + 22.4175δ[n − 2] − 14.35δ[n − 3] + 7δ[n − 4]

We need to determine the expanded version of H (z) with an additional gain factor.
H (z) = A(1 − 0.8e jπ/3 z −1 )(1 − 0.8e−jπ/3 z −1 )(1 − 1.25e jπ/3 z −1 )(1 − 1.25e−jπ/3 z −1 )
= (1 − 0.8z −1 + 0.64z −2 )(1 − 1.25z −1 + 1.5625z −2 )
Clearly, when expanded out, the H (z) polynomial will be of the form
H (z) = A + . . . + A(0.8) 2 (1.25) 2 z −4 , so if h[0] = 7 then A = 7.
(b) H (z) = 1.81406 − 3.71882z −1 + 5.80952z −2 − 3.71882z −3 + 1.81406z −4

If we expand H (z) into a polynomial with unknown gain we get
H (z) = A(1 − 2.05z −1 + 3.2025z −2 − 2.05z −3 + z −4 )
The DC value of the frequency response is
H (e j0 ) = A(1 − 2.05 + 3.2025 − 2.05 + 1) = 1.1025A = 2

s
w
ng
la
hi
)
n
ht
eb
c
tio
so A = 1.8141 and
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
H (z) = 1.8141(1 − 2.05z −1 + 3.2025z −2 − 2.05z −3 + z −4 ) rld

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.18 DSP First 2e
(a) H (1) = b0 + b1 − b1 − b0 = 0
(b)
H (e j ω̂ ) = b0 + b1 e−j ω̂ − b1 e−j ω̂3 − b0 e−j ω̂4 = e−j ω̂2 (b0 e j ω̂2 + b1 e j ω̂ − b1 e−j ω̂ − b0 e−j ω̂2 )
= [2b0 sin(2ω̂) + 2b1 sin(ω̂)]e j (π/2−j ω̂ (from the inverse Euler formula)
(c)
H (z −1 ) = b0 + b1 z − b1 z 3 − b0 z 4 = z 4 (b0 z −4 + b1 z −3 − b1 z −1 − b0 )
= −z 4 H (z)
(d) In the general case with bk = −b M−k , the condition H (1/z) = −z M H (z) holds for M even or odd.
The frequency response has the form
M/2
H (e j ω̂ ) = 2e j (π/2−ω̂ M/2)
X
b M/2−k sin(ω̂k) M an even integer
s
w
ng
la
k=1
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
and
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
(M+1)/2
o
St
.
st
ed
ng
H (e j ω̂ ) = 2e j (π/2−ω̂ M/2)
X
in
d
b(M+1)/2−k sin(ω̂(2k − 1)/2) itt

ni
e
te
M an odd integer
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
k=1
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-9.19 DSP First 2e
(a) H1 (z) = H2 (z) = 1 + z −1 + z −2 + z −3
(b) H (z) = H1 (z)H2 (z) = (1 + z −1 + z −2 + z −3 ) 2
(c) H (z) = 1 + 2z −1 + 3z −2 + 4z −3 + 3z −4 + 2z −5 + z −6
so h[n] = δ[n] + 2δ[n − 1] + 3δ[n − 2] + 4δ[n − 3] + 3δ[n − 4] + 2δ[n − 5] + δ[n − 6]
(d) y[n] = x[n] + 2x[n − 1] + 3x[n − 2] + 4x[n − 3] + 3x[n − 4] + 2x[n − 5] + x[n − 6]
(e) We can use the formula for the sum of N terms of a geometric series to write H1 (z), or H2 (z) as
1 − z −4
H1 (z) = H2 (z) = 1 + z −1 + z −2 + z −3 =
1 − z −1
(f) From part (e) we see that (1 + z −1 + z −2 + z −3 ) 2 so all the zeros at {−1, ± j} are double zeros, and there are six poles at
z = 0.
2
1
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
0.5 em
co
d
Imaginary Part
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
2 6
in
d
0 itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
-0.5
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
2
ed
k
as
or
-1
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
-1 -0.5 0 0.5 1
g
ur
d
te
y
an
co
an
in
Real Part
r
e
of
ei
th
th
y
l
sa
ro
Figure P-9.19
st
or
de
ill
w
(g) From the result in (e), we can write
1 − e−j ω̂4 e−j ω̂2 (e j ω̂2 − e−j ω̂2 ) sin(2ω̂) −j3ω̂/2

H1 (e j ω̂ ) = H2 (e j ω̂ ) = = = e
1−e −j ω̂ e −j ω̂/2 (e j ω̂/2 −e −j ω̂/2 ) sin(ω̂/2)
(h) Note that H (e j ω̂ ) = H2 (e j ω̂ )H1 (e j ω̂ ) = [H1 (e j ω̂ )]2 = [H2 (e j ω̂ )]2 .
!2
j ω̂ j ω̂ j ω̂ sin(2ω̂)
H (e ) = H1 (e )H2 (e )= e−j3ω̂
sin(ω̂/2)
Since the squared-term is nonnegative, it is the magnitude, and the system has linear phase with time delay equal to 3
samples. The sketch should approximate the Matlab plots of magnitude and phase shown below.

20
|H(ej ω̂ )|
s
w
ng
la
hi
10
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
0 rld
c
i
at
.D
ru
o
St
.
−π −0.8π −0.6π −0.4π −0.2π π
st
0 0.2π 0.4π 0.6π 0.8π

ed
ng
W
in
d
itt
ni
e
te
of
5
th
m
ar
ni
H(ej ω̂ )
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
0
g
in
an
y
n
o
(
l
le
si
pr
k
6
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
-5
of
of
is
pr
Th
rt
−π −0.8π −0.6π −0.4π −0.2π 0 0.2π 0.4π 0.6π 0.8π π

se
rit
is
pa
g
ur
d
te
y
an
ω̂
co
an
in
r
e
of
ei
th
th
Figure P-9.19
e
y
l
sa
ro
st
or
de
ill
w

Chapter 10
IIR Filters
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
201
CHAPTER 10. IIR FILTERS
P-10.1 DSP First 2e
(a) The system function is

1 1
H (z) = =
1− z −1 − z −2 − p2 z −1 )
(1 − p1 z −1 )(1
√
where, by the quadratic formula, p1,2 = (1 ± 5)/2 = 1.618, −0.618. The partial fraction expansion of H (z) is
√ √
p1 p2 5+1 5−1
√ √
p1 − p2 p2 − p2 2 5 2 5
H (z) = + = +
1 − p1 z −1 1 − p2 z −1 1 − p1 z −1 1 − p2 z −1
Therefore, the impulse response is
√ √ n √ √ n
5 + 1+ *1 + 5+ 5 − 1+ *1 − 5+
h[n] = * √ u[n] + * √ u[n]
, 2 5 -, 2 - , 2 5 -, 2 -
Try a few terms of this closed form expression to see if it generates the Fibonacci sequence, 1, 1, 2, 3, 5, 8, 13, ...
√
(b) The poles determined in part (a) are p1,2 = (1 ± 5)/2 = 1.618, −0.618. Since one of the poles is outside the unit s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
circle, this is an unstable system.
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-10.2 DSP First 2e
First, find the system function by inspection as
z −1
!
1
H (z) = = z −1 = G(z)z −1
1 − 0.9z −1 + 0.9z −2 1 − 0.9z −1 + 0.9z −2
The z −1 factored out suggests that the impulse response will be zero prior to n = 1. Then we can find the partial fraction
expansion of the rational function part and write G(z) as
0.568e−j0.4942 0.568e j0.4942

G(z) = +
1 − 0.9487e j1.0766 z −1 1 − 0.9487e−j1.0766 z −1
and
n n
g[n] = 0.568e−j0.4942 0.9487e j1.0766 u[n] + 0.568e j0.4942 0.9487e−j1.0766 u[n]
= 1.136(0.9487) n cos(1.0766n − 0.4942)u[n]
h[n] = g[n − 1] = 1.136(0.9487) n−1 cos(1.0766(n − 1) − 0.4942)u[n − 1]

s
w
ng
la
hi
)
n
ht
eb
which, by virtue of u[n − 1], is the desired formula showing h[n] being zero for n < 1 and being a decaying sinusoid for c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
n ≥ 1.
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
It should be noted that we could have made a partial fraction expansion of H (z) directly. The resulting inverse transform
o
St
.
st
ed
ng
W
in
d
would be a formula that is valid from n = 0 since the complex exponentials would be multiplied by u[n]. However, when the
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
formula is evaluated at n = 0, the result would be zero.

us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-10.3 DSP First 2e
The system function is
5z −7
H (z) =
1 − 0.5z −1
Us the z-transform table and then the delay property,
h[n] = 5(0.5) n−7 u[n − 7]
Note that h[n] = 0 for n < 7. It decays exponentially as (0.5) n for n ≥ 7.
4
h[n]
0
0 5 10 15 20 25 30
time index n
Figure P-10.3
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-10.4 DSP First 2e
The impulse response is h[n] = 5(0.5) n u[n], so the output for the given input is
y[n] = 3h[n − 1] − 2h[n − 2] + 3h[n − 4] − h[n − 5]

= 15(0.5) n−1 u[n − 1] − 10(0.5) n−2 u[n − 2] + 15(0.5) n−4 u[n − 4] − 5(0.5) n−5 u[n − 5]
Evaluate with Matlab in order to make a plot.
20
10
y[n]
-10
0 5 10 15 20 25 30
time index n
Figure P-10.4
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-10.5 DSP First 2e
The easiest way to work this is to write the output in terms of shifted impulse responses as
y[n] = h[n] + βh[n − 2] + h[n − 4]

√
and then examine y[n] for n ≥ 4. Define a = 1/ 2 so h[n] = 10a n u[n]. Then for n ≥ 4, the output is
y[n] = 10a n + 10 βa n−2 + 10a n−4 = 10a n (1 + βa−2 + a−4 )
To make y[n] = 0 for n ≥ 4, we need 1 + βa−2 + a−4 = 0. Solving this equation for β gives β = −(1 + a4 )/a2 , or for the
given value of a, β = −2.5. If you use this value of β, the output looks like the following
0
y[n]
-5
-10
0 5 10 15 20 25 30
time index n
s
Figure P-10.5 w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-10.6 DSP First 2e
−1 + 3z −1 − 2z −2
(a) The system function is H (z) =
1 − 13 z −1 − 21 z −2
The zeros are at z = 1, 2, and the poles at z = 0.8931, −0.5598
2
1
Imaginary Part
-1
-2
-2 -1 0 1 2
Real Part
−1 + 3z −1 + 2z −2 s
w
ng
la
(b) The system function is H (z) =
hi
)
n
ht
eb
c
tio
ea
rig
1 − 13 z −1 + 21 z −2
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
The zeros are now at z = 3.5616, −0.5616, and the poles at z = 0.7071e±j1.3329 .
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
4 itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
2
g
in
an
y
n
o
(
l
le
si
pr
Imaginary Part
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
0
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
-2
y
l
sa
ro
st
or
de
ill
w
-4
-4 -2 0 2 4
Real Part

P-10.7 DSP First 2e
(a) y[n] = 13 y[n − 1] − 21 y[n − 2] − x[n]
−1 −z 2
H (z) = =
1 − 31 z −1 + 12 z −2 z 2 − 31 z + 1
2
=⇒Poles: z = 0.7071e±j1.3329 Zeros: z = 0, 0
(b) y[n] = 31 y[n − 1] − 21 y[n − 2] − x[n − 2]
−z −2 −1
H (z) = =
1− 1 −1
3z + 1 −2
2z z 2 − 31 z + 1
2
=⇒Poles: z = 0.7071e±j1.3329 Zeros: z = ∞, ∞
(c) y[n] = 31 y[n − 1] − 21 y[n − 2] − x[n − 4]
−z −4 −1
H (z) = =
s
w
ng
1− 1 −1
3z + 1 −2
2z z 2 (z 2 − 13 z + 21 )
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
=⇒Poles: z = 0.7071e±j1.3329, 0, 0 Zeros: z = ∞, ∞, ∞, ∞
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
In (c) there are four zeros at infinity balanced by two additional poles at the origin.
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
(b) Two zeros at z = ∞ (c) Four zeros at z = ∞

ud
(a) Two zeros at origin

d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
1 1 1
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
s
Imaginary Part
Imaginary Part
Part
0.5 0.5 0.5

pr
hi
rt
y
T
se
rit
is
pa
g
ur
d
te
y
an
Imaginary
co
an
(2) (2)
in
r
0 0 0
e
of
ei
th
th
y
l
sa
ro
st
or
de
-0.5 -0.5 -0.5

ill
w
-1 -1 -1
-1 0 1 -1 0 1 -1 0 1
Real Part Real Part Real Part

P-10.8 DSP First 2e
(a) By inspection,
1 z
H (z) = =
1+ 1 −1
2z z− 1
2
H (z) has a pole at z = − 12 and a zero at z = 0. The corresponding impulse response is h[n] = (−0.5) n u[n].
(b) Write the input as x[n] = δ[n] + δ[n − 2] + δ[n − 5]. Therefore, it follows that for this input the output is y[n] =
h[n] + h[n − 2] + h[n − 5]. If we substitute for h[n] we get
y[n] = (−0.5) n u[n] + (−0.5) n−2 u[n − 2] + (−0.5) n−5 u[n − 5]
Taking the step sequence into account we get

 0 n<0
n = 0, 1

(−0.5) n


y[n] = 



 5(−0.5) n n = 2, 3, 4
−27(−0.5) n

 n≥5
s

w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-10.9 DSP First 2e
−15 + 20z −1 −15(z − 4/3)

(a) H (z) = =
1 − 0.75z −1 z − 0.75
(b) Need plot. Pole at z = 0.75, zero at z = 4/3.
−15 + 20e−j ω̂
(c) H (e j ω̂ ) =
1 − 0.75e−j ω̂
(d) |H (e j ω̂ )| 2 = 400
conjugate
−j ω̂
!z }|
j ω̂
{!
−15 + 20e −15 + 20e
|H (e j ω̂ )| 2 = H (e j ω̂ )H ∗ (e j ω̂ ) =
1 − 0.75e−j ω̂ 1 − 0.75e j ω̂
625 − 600 cos(ω̂)
=
1.5625 − 1.5 cos(ω̂)
400(1.5625 − 1.5 cos(ω̂))
= = 400
1.5625 − 1.5 cos(ω̂)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-10.10 DSP First 2e
(a) By inspection,
1 1
H (z) = = √ √
1 + 0.8z −2
(1 − 0.8 e jπ/2 z −1 )(1 − 0.8 e−jπ/2 z −1 )
z2
= √ √
(z − 0.8 e jπ/2 )(z − 0.8 e−jπ/2 )
√
Zeros at z = 0, 0 and poles at z = 0.8 e±jπ/2 .
(b) To find the output for the given input, determine the impulse response.
1 0.5 0.5
H (z) = = √ + √
1 + 0.8z −2 1 − 0.8 e jπ/2 z −1 1 − 0.8 e−jπ/2 z −1
so
n n √ n
0.8 e−jπ/2 u[n] = 0.8 cos( π2 n)u[n]
√ √
h[n] = 1
2 0.8 e jπ/2 u[n] + 1
2
Note: h[n] = 0 for n even, which could have been predicted from the fact that the initial H (z) has only z −2 in the w
s
ng
la
hi
)
n
ht
eb
c
tio
denominator—an even power. ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
Therefore,
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
y[n] = 5h[n] + 2.5h[n − 1] + 4h[n − 2] + 2h[n − 3] itt

ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
√ n √ n−1
en
= 5 0.8 cos( π2 n)u[n] + 2.5 0.8 cos( π2 (n − 1))u[n − 1]

he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
√ n−2 √ n−3
(
l
cos( π2 (n − 2))u[n − 2] + 2 0.8 cos( π2 (n − 3))u[n − 3]

le
si
pr
+ 4 0.8
k
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(a) By inspection,
1 z5
H (z) = =
1 − 0.97z −5 z 5 − 0.97
(b) There are 5 zeros at z = 0, and the poles are at the 5th roots of 0.97, zk = (0.97) 0.2 e j2πk/5 = (0.9939)e j2πk/5 .
0.5
Imaginary Part
5
0
-0.5
-1
s
w
ng
la
hi
)
n
ht
eb
c
tio
-1 -0.5 0 0.5 1 ea
rig
W
a
t
py
in
in
e
Real Part em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
Figure P-10.11 o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
(c) The z-transform of the output is

ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
1
or
or
is
Y (z) = H (z)X (z) = (1 + βz −p )

w
s
w
ed
k
as
or
e
id
hi
1 − 0.97z −5
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
If we choose β = −0.97 and p = 5, we get Y (z) = 1 or y[n] = δ[n], which satisfies the condition y[n] = 0 for n ≥ 5.
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(a) By using the delay and linearity properties,
h a [n] = (−0.77) n u[n] − (−0.77) n−1 u[n − 1]
or, by making the partial fraction expansion,

1.77/0.77
Ha (z) = −1/.77 + ⇐⇒ x a [n] = −1/0.77 + (1.77/0.77) (−0.77) n u[n]
1 + 0.77z −1 | {z } | {z }
−1.2987 2.2987
(b) By using the delay and linearity properties,
hb [n] = (0.9) n u[n] + 0.8(0.9) n−1 u[n − 1]

1.7/0.9
Ha (z) = −0.8/0.9 + ⇐⇒ x a [n] = −0.8/0.9 + (1.7/0.9) (0.9) n u[n]
1 − 0.9z −1 | {z } | {z }
−0.8889 1.8889
s
w
ng
la
hi
)
n
ht
eb
c
tio
(c) Using the delay property, ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
hc [n] = (0.9) n−2 u[n − 2]

to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
(d) Using the transform of an impulse and the delay property,

us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
hd [n] = δ[n] − δ[n − 1] + 2δ[n − 3] − 3δ[n − 4]

d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(a) Using residuez or hand calculation, determine the partial fraction expansion of H (z) as
1 − z −1 −0.6 1.6
X a (z) = = +
(1 − 1 −1
2 z )(1 + 1 −1
3z ) 1− 1 −1
2z 1 + 13 z −1
Therefore,
x a [n] = −0.6( 21 ) n u[n] + 1.6(− 13 ) n u[n]
(b) Determine the partial fraction expansion as
1 + z −2
X b (z) =
(1 − 0.9e j2π/3 z −1 )(1 − 0.9e−j2π/3 z −1 )
0.6556e j1.7506 0.6556e−j1.7506
= 1.2346 + +
1 − 0.9e j2π/3 z −1 1 − 0.9e−j2π/3 z −1
Now the inverse z-transform is by table look up,
s
w
ng
x b [n] = 1.2346 δ[n] + 0.6556e j1.7506 (0.9e j2π/3 ) n u[n] + 0.6556e−j1.7506 (0.9e−j2π/3 ) n u[n]
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
= 1.2346 δ[n] + 1.3112(0.9) n cos((2π/3)n + 1.7506)u[n]
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
(c) The partial fraction expansion is

in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
1 + z −1
g
−0.1176
he
1.1176
no
in
ud
d
X c (z) = = +
ud
rt
te
is
st
c
fo
cl
(1 − 0.9z −1 )(1 + 0.8z −1 ) 1 − 0.9z −1 1 + 0.8z −1

d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
so the inverse transform is

or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
x c [n] = 1.1176(0.9) n u[n] − 0.1176(−0.8) n u[n]

g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

S1 : y[n] = 0.9y[n − 1] + 12 x[n] + 12 x[n − 1] #2

S2 : y[n] = −0.9y[n − 1] + 9x[n] + 10x[n − 1] #3
1
2 (1 − z −1 )
S3 : H (z) = #6
1 + 0.9z −1
S4 : y[n] = 14 x[n] + x[n − 1] + 32 x[n − 2] + x[n − 3] + 14 x[n − 4] not used
S5 : H (z) = 1 − z −1 + z −2 − z −3 + z −4 #5
2
X
S6 : y[n] = x[n − k] #4
k=0
S7 : y[n] = x[n] + x[n − 1] + x[n − 2] + x[n − 3] + x[n − 4] + x[n − 5] #1
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

1
(a) By inspection, H (z) = . Therefore,
1 − 12 z −1
!
1 1 −1 2
Y (z) = H (z)X (z) = * +
−1
= +
1 −1
,1 − 2 z - 1 − z
1
1 − 2z −1 1 − z −1
Therefore, y[n] = 2u[n] − ( 12 ) n u[n] = (2 − ( 21 ) n )u[n].
(b) Use a partial fraction expansion:
1 1
1 −jπ/4
1 − 2e
!
1 1 jπ/4
Y (z) = H (z)X (z) = * + = 1 − 2e +
1 −1
,1 − 2 z - 1 − e
jπ/4 z −1 1 −1
1 − 2z 1 − e jπ/4 z −1
!
1 1
Therefore, y[n] = jπ/4
( 12 ) n u[n] + * 1
+ e j (π/4)n u[n]
1 − 2e , 1 − 2{ze −jπ/4
| }-
H (e j π /4 )
s
w
ng
la
hi
)
n
ht
eb
c
tio
1 ea
rig
(c) Note that H (e j ω̂ ) =
W
a
t
py
so the second term in y[n] in part (b) is the steady-state component.

in
in
e
em
co
1 −j ω̂
d
s
1−
i
2e
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
!
itt
ni
1
te
of
th
m
ar
y[n] = ( 1 ) n u[n] + H (e jπ/4 )e j (π/4)n u[n]

ni
er
e
le
on
U
us
1 − 2e jπ/4 2
tp
t
by
en
g
he
no
| {z }
in
ud
d
steady state
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

1 − 0.1756z −1 − 0.1756z −2 + z −3
(a) Multiply out the factors in numerator and denominator: H (z) =
1 + 0.4944z −1 + 0.64z −2
y[n] = −0.4944y[n − 1] − 0.64y[n − 2] + x[n] − 0.1756x[n − 1] − 0.1756x[n − 2] + x[n − 3]
(b) Get the poles and zeros from the given factored form of H (z); then plot in the complex z-plane.
0.5
Imaginary Part
-0.5
s
w
ng
la
-1
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
-1 -0.5 0 0.5 em
1
co
d
s
i
W
r
ss
to
es
Real Part rld

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
Figure P-10.16
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
(c) H (z) has zeros on the unit circle at ω̂ = 0 and ω̂ = ±0.3π. Therefore, any complex exponential with one of these
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
frequencies will have zero output.

or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

S1 : y[n] = 0.9y[n − 1] + 12 x[n] + 12 x[n − 1] #1

S2 : y[n] = −0.9y[n − 1] + 9x[n] + 10x[n − 1] #4
1
2 (1 − z −1 )
S3 : H (z) = #2
1 + 0.9z −1
S4 : y[n] = 14 x[n] + x[n − 1] + 32 x[n − 2] + x[n − 3] + 14 x[n − 4] #6
S5 : H (z) = 1 − z −1 + z −2 − z −3 + z −4 #5
2
X
S6 : y[n] = x[n − k] #3
k=0
S7 : y[n] = x[n] + x[n − 1] + x[n − 2] + x[n − 3] + x[n − 4] + x[n − 5] not used
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(A) #2
(B) #4
(C) #1
(D) #6
(E) #3
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(a) Expand the numerator and denominator to get
1 − 2 cos(π/3)z −1 + z −2 1 − z −1 + z −2
H (z) = =
1 − 2r cos(π/3)z −1 + r 2 z −2 1 − r z −1 + r 2 z −2
Therefore, the difference equation is y[n] = r y[n − 1] − r 2 y[n − 2] + x[n] − x[n − 1] + x[n − 2].
(b) Plot the poles and zeros of H (z) for the case r = 0.9. Angles to poles and zeros are ±π/3.
1
0.5
Imaginary Part
-0.5
-1
-1 -0.5 0 0.5 1
Real Part
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
(c) Magnitude of the frequency response plotted for the case r = 0.9. Notches are at ±π/3.
a
t
py
in
in
e
em
co
d
s
i
W
1.5
r
ss
to
es
rld
|H(ej ω̂ )|
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
1 itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
0.5
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
0
(
l
le
si
pr
k
es
so
or
or
−π −0.8π −0.6π −0.4π −0.2π π

is
0 0.2π 0.4π 0.6π 0.8π

w
s
w
ed
k
as
or
e
id
hi
ω̂
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
(d) The notches get sharper and the magnitude at other frequencies approaches 1.
te
y
an
co
an
in
r
e
of
ei
th
th
(e) Frequency = Angle. Change the angle of the poles and zeros from π/3 to whatever frequency is desired.
y
l
sa
ro
st
or
de
ill
w

(J) #6
(K) #5
(L) #4
(M) #2
(N) #3
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

−0.8 + z −1 z − 1.25
(a) H (z) = −1
= (−0.8)
1 − 0.8z z − 0.8
(b) The pole is at 0.8 and the zero is at 1.25 = 1/0.8.
(−0.8 + e−j ω̂ )(−0.8 + e j ω̂ ) 1.64 − 1.6 cos(ω̂)
(c) |H (e j ω̂ )| 2 = H (e j ω̂ )H ∗ (e j ω̂ ) = = =1
(1 − 0.8e−j ω̂ )(1 − 0.8e j ω̂ ) 1.64 − 1.6 cos(ω̂)
(d) This results because the magnitude response is unity for all ω̂. Only the phases will change.
(e) By evaluating H (e j ω̂ ) at the two frequencies (with Matlab), we get ϕ1 = −2.4826 and ϕ2 = −3.0285.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

If we sample with f s = 4000, the input to the LTI system is
x[n] = 9 + 8 cos(0.3πn) + 7 cos(0.15πn − 0.7π)
so we need to evaluate the frequency response at ω̂ = 0, 0.3π, 0.15π. The frequency response is obtained by substituting
z = e j ω̂ into the expression for H (z) and then evaluating at the required frequencies.
Since H (z) has zeros on the unit circle at ω̂ = ±0.3π, the component at that frequency is nulled out. Therefore, the
output of the LTI system is
y[n] = 9H (e j0 ) + 7|H (e j0.15π )| cos(0.15πn − 0.7π + ∠H (e j0.15π ))

y[n] = 9(0.7725) + 7(0.6010) cos(0.15πn − 0.7π − 0.3190))
y[n] = 6.9525 + 4.2070 cos(0.15πn − 0.7π − 0.3190))
Now use the sampling frequency to write the output of the D-to-C converter as
y(t) = 6.9525 + 4.2070 cos(600πt − 0.7π − 0.3190))
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(a) Plot the poles and zeros of H (z) in the z-plane.
0.5
Imaginary Part
-0.5
-1
-1 -0.5 0 0.5 1
Real Part
(b) The frequency response is constant for all ω̂ at (1.25) 2 = 1.5625. The vector lengths should cancel out to a constant.
1.5625
|H(ej ω̂ )|
s
w
ng
la
1.5625
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
1.5625 rld
c
i
at
.D
ru
π
o
−π
St
0
.
-5 5
st
ed
ng
W
in
d
itt
ni
e
te
of
ω̂
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
(c) The zeros of an all-pass filter are at reciprocals of the poles, so the angle stays the same but the radius of the zeros
is
st
c
fo
cl
d
te
in
an
y
n
o
must be decreased to 1/0.9 =1.1111.

l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(a) Expand H (z) to
(1 + z −2 ) (1 + z −2 )
H (z) = =
(1 − 0.9z −1 )(1 + 0.5z −1 ) 1 − 0.4z −1 − 0.45z −2
Therefore, the difference equation is y[n] = 0.4y[n − 1] + 0.45y[n − 2] + x[n] + x[n − 2].
(b) Plot the poles and zeros of H (z) in the z-plane.
0.5
Imaginary Part
-0.5
s
-1 w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
-1 -0.5 0 0.5 1
t
py
in
in
e
em
co
d
s
i
W
Real Part
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
(c) The pole at z = 0.9 dominates around ω̂ = 0 and the zeros pull down the frequency response around ω̂ = ±π/2. The
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
pole at z = −0.5 is symmetrical with the other pole, but much farther away from the unit circle. Therefore the maximum
t
by
en
g
he
no
in
ud
d
ud
rt
te
value occurs at ω̂ = 0.
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
The estimate of the maximum value is obtained as the ratio of the product of the lengths of the vectors from the
w
s
w
ed
k
as
or
√ 2 to z = 1 divided by the product of the lengths of the vectors from the poles to z = 1. The maximum value is
id
hi
th
w
zeros
ov
t
an
of
of
is
pr
Th
( 2) /(0.1 × 1.5) = 13.33.

s
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
(d) At ω̂ = ±π/2, |H (e j ω̂ )| = 0.
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(a) This system has a pole at z = −0.8 and a zero at z = 0.9.

(1 − 0.9e−j ω̂ )(1 − 0.9e j ω̂ ) 1.81 − 1.8 cos(ω̂)
(b) |H (e j ω̂ )| 2 = H (e j ω̂ )H ∗ (e j ω̂ ) = =
(1 + 0.8e−j ω̂ )(1 − 0.8e j ω̂ ) 1.64 + 1.6 cos(ω̂)
(c) This is a highpass filter since it has a zero near z = 1 or ω̂ = 0 and a pole near z = −1 or ω̂ = π.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

10 + 5z −1
! !
3 15 15z
(a) H (z) = H1 (z)H2 (z) = = =
2+z −1 1 − 0.7z −1 1 − 0.7z −1 z − 0.7
Therefore, H (z) has a zero at z = 0 and a single pole at z = 0.7.
1 + 21 z −1
(b) Since H1 (z) = , it follows that
1 − 12 z −1
3z −1 (1 + 12 z −1 ) 3z −1 + 23 z −2
H (z) = =
(1 − 12 z −1 )(1 − 56 z −1 + 16 z −2 ) 1 − 34 z −1 + 7 −2
12 z − 1 −3
12 z
(c) y[n] = 43 y[n − 1] − 7

12 y[n − 2] + 1
12 y[n − 3] + 3x[n] + 23 x[n − 2 − 1]
(d) For the systems in (b), plot the poles and zeros of H (z) in the complex z-plane.
s
w
ng
0.5
Imaginary Part
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
2 em
co
d
s
i
0
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
-0.5
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
-1
g
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
-1 -0.5 0 0.5 1
or
e
id
hi
th
w
d
ov
t
an
Real Part
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
Figure P-10.26
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(a) Since we require H (z) = H1 (z)H2 (z) = 1 for the output to equal the input, then H2 (z) = 1/H1 (z).
(b) Since H1 (z) = 1 − 43 z −1 ,
1 1
H2 (z) = =
H1 (z) 1 − 34 z −1
(c) Since the poles of H2 (z) must be the zeros of H1 (z), both the poles and the zeros of H1 (z) must be inside the unit
circle of the z-plane.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(a) Here is a stem plot of y[n] versus n, over the range 0 ≤ n ≤ 10.
y[n] 1
-1
0 5 10 15 20 25 30
time index n
Figure P-10.28
(b) We need to find a system whose impulse response is h[n] = (0.95) n cos(0.2πn + 0.4π)u[n]. Do this by finding the
corresponding system function, writing it as a ratio of polynomials in z −1 , and picking off the coefficients of the
corresponding difference equation.
First write h[n] as
s
w
ng
h[n] = 21 e j0.4π (.95e j0.2π ) n u[n] + 21 e−j0.4π (.95e−j0.2π ) n u[n]
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
The corresponding z-transform is
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
1 j0.4π 1 −j0.4π
in
d
2e 2e itt
ni
e
te
of
th
m
ar
H (z) = +
ni
er
e
le
on
U
us
tp
0.95e j0.2π z −1 0.95e−j0.2π z −1

t
1− 1−
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
If we put this over a common denominator and collect terms using the inverse Euler formula, we get
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
cos(0.4π) − 0.95 cos(0.2π)z −1 0.3090 − 0.7686z −1

e
id
hi
th
w
H (z) = =
ov
t
an
of
of
is
pr
1 − 1.9 cos(0.2π)z −1 + (0.95) 2 z −2 1 − 1.5371z −1 + 0.9025z −2

Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
The difference equation for implementing this system is therefore,

of
ei
th
th
y
l
sa
ro
st
or
de
y[n] = 1.5371y[n = 1] − 0.9025y[n − 2] + 0.3090x[n] − 0.7686x[n − 1]

ill
w

b0
(a) H3 (z) =
1 − a1 z −1 − a2 z −2 − a3 z −3
(b) In general, the system function overall is
2b0 z −1 (1 − z −2 )(1 + 21 z −1 ) 2b0 z −1 (1 − z −2 )(1 + 21 z −1 )

H (z) = H3 (z)H2 (z)H1 (z) = =
1 − a1 z −1 − a2 z −2 − a3 z −3 1 − a1 z −1 − a2 z −2 − a3 z −3
2b0 z −1 (1 + 12 z −1 − z −2 − 12 z −3 )
=
1 − a1 z −1 − a2 z −2 − a3 z −3
For the given coefficients, we get
14z −1 (1 − z −2 )(1 + 12 z −1 )
H (z) =
1 − 21 z −1 − 12 z −2
(c) From the given information, we want H (z) = 8z −nd . Since we have a third-order polynomial in both the numerator
and denominator, we just need to adjust the coefficients of the denominator so that the two polynomials are identical
and therefore cancel. The required values are
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
b0 = 4 a1 = − 12 a2 = 1 a3 = 1
W
a
t
py
in
in
e
2 em
co
d
s
i
W
r
ss
to
es
rld
c
For these conditions, nd = 1.

at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

(a) yy=filter([0,0,1,0,1],1,xx); ANS =1
(b) y[n] = 0.9y[n − 1] + 0.8y[n − 2] + x[n] − x[n − 3] ANS =3
(c) h[n] = u[n − 2] − u[n − 5] ANS =4

3 √
X
(d) y[n] = k − 1 x[n − k] ANS =5
k=1
(e) h[n] = 0.8n−2 u[n − 2] − 0.8n−3 u[n − 3] ANS =2
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

Appendix A
Complex Numbers
A-1 Problems
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
232
APPENDIX A. COMPLEX NUMBERS
P-A.1 DSP First 2e
(a) Cartesian: z = 0 + j2. Polar: 2e jπ/2 = 2∠90◦
(b) Cartesian: z = −3 − j4. Polar: 5e−j2.21 = 5e−j0.705π = 5 |

∠ − 126.87
{z }
◦
√ j3π/4
(c) Cartesian: z = −1 + j; Polar: 2e = 2 |{z}
∠135◦
θ
(d) Cartesian: z = 0 − j; Polar: 1e−jπ/2 = 1 |

∠− 90}◦
s
w
ng
{z la
hi
θ
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.2 DSP First 2e
√ j (3π/4) √
(a) 2e = 2(cos(3π/4) + j sin(3π/4) ) = −1 + j
| {z } | {z }
√ √
=− 2/2 = 2/2
(b) z = 3e−j (π/2) = 3∠ − 90◦ ⇒ z = −3 j
(c) 1.6 ∠ (π/6) = 1.6e jπ/6 = 1.6(cos(π/6) + j sin(π/6) ) = 1.386 + j0.8

|{z} | {z } | {z }
√
30◦ =− 3/2 =0.5
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
(d) 7∠ (7π) = 7∠π = 7e jπ = −7 + j0

o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
|{z}
th
m
ar
ni
1260◦
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.3 DSP First 2e
(a) j 3 = j ( j) 2 = j (−1) = − j = 0 − j
n  +1
 n even
(b) j 2n = j 2 = (−1) n = 
 −1 (n an integer)
 n odd
(c) e j (π+2πm) = e jπ e j2πm = (−1)(1) = −1 + j0 (m an integer)

1/2 √ √
(d) j 1/2 = e jπ/2 = e jπ/4 = 2/2 + j 2/2 (this is one answer)
1/2 √ √
j 1/2 = e jπ/2 e j2π = e jπ/4 |{z}
e jπ = −e jπ/4 = − 2/2 − j 2/2 (this is the second answer)
=−1
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.4 DSP First 2e

√
Use this fact in (a,b,c): 2 − j2 = 2.449e−j0.304π (Angle is −0.955 rad, or −54.74◦ )
√ 8
(a) ( 2 − j2) 8 = 2.449e−j0.304π = (2.449) 8 e−j8(.304π) = 1296e j0.433π
(Angle is −1.36 rad, or −77.88◦ )
√
(b) ( 2 − j2) −1 = 1/2.449e+j0.304π = 0.4082e+j0.304π
(Angle is +1.36 rad, or +77.88◦ )
√ 1/2
(c) ( 2 − j2) 1/2 = 2.449e−j0.304π e j2π` = 1.565e−j0.152π e jπ` ` = 0, 1
There are two answers: 1.565e−j0.152π and 1.565e j0.848π
(d) Add the two complex numbers below and obtain

−4.964 + j4.598 = 6.766e j0.762π
(Angle is 2.394 rad, or 137.2◦ )
√
3e j2π/3 = −3/2 + j3 3/2 w
s
ng
la
hi
√
)
n
ht
eb
c
tio
ea
−4e−jπ/6 = −4 3/2 + j2
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
( ) ( ) ( ) o
St
(e) = je−jπ/3 = = e jπ/2 e−jπ/3 = = e jπ/6 = sin(π/6) = 0.5

.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.5 DSP First 2e
(a) z1∗ = (−4 + j3) ∗ = −4 − j3 = 5e−j0.795π = 5e j2.50
(b) z22 = (1 − j) 2 = 1 − 2 j + (− j) 2 = 1 − 2 j = 1 = −2 j
√ 2
or, in polar form: z22 = 2e−jπ/4 = 2e−jπ/2 = − j2
(c) z1 + z2∗ = (−4 + j3) + (1 − j) ∗ = (−4 + j3) + (1 + j) = 3 + j4 = 5e j0.705π

(Angle is 2.214 rad, or 126.872◦ )
(d) j z2 = j (1 − j) = j − j ( j) = j + 1 = 1 + j
√ √
or, in polar form: e jπ/2 2e−jπ/4 = 2e jπ/4
(Angle is 0.785 rad, or 45◦ )
s
w
ng
(e) Use polar form: z1−1 = 1/(5e j0.795π ) = 0.2e−j0.795π
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
√ √
t
py
in
in
e
em
(f) Use polar form: z1 /z2 = (5e j0.795π )/( 2e−jπ/4 ) = (5 2/2)e j1.045π = 3.54e−j0.955π
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
(Angle is −3.0 rad, or −171.8◦ )

.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
(g) ez2 = e1−j = e1 e−j = e(cos(1) + j sin(1)) = 1.469 − j2.287

tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
Polar form is e1 e−j = 2.7183e−j0.318π

fo
cl
(angle is −1 rad, or −57.3◦ )

d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
(h) z1 z1∗ = (−4 + j3)(−4 + j3) ∗ = (−4 + j3)(−4 − j3) = 16 + 9 = 25

k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
√ √
s
rt
y
se
rit
is
pa
(i) Use polar form: z1 z2 = (5e j0.795π )/( 2e−jπ/4 ) = 5 2e j0.454π = 7.07e j1.713 (98.13◦ )
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.6 DSP First 2e
z1 = e j9π/3 = e j3π = −1 + j0
z2 = e−j5π/8 = −0.383 − j0.924
z3 = e j13π/8 = e j13π/8 = 0.383 − j0.924z = −1 − j1.848 = 2.1e−j2.07
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.7 DSP First 2e
(a) z = −3 + j4 = 5 e j0.705π
1/z = (1/5)e−j0.705π The angle is also 2.214 rad, or 126.87◦ .

√
(b) z = −2 + j2 = 2 2 e j3π/4
√ 5 √
z 5 = 2 2 e j15π/4 = 128 2 e−jπ/4 because 15π/4 = 4π − π/4.
(c) z = −5 + j13
|z| 2 = zz ∗ = (−5 + j13) (−5 − j13) = 25 + 169 = 194 = 194 e j0
(d) < (−2 + j5) |{z}e−jπ/2 = < (−2 + j5)(− j) = < − 2 j + 5 = 5 = 5e j0

=−j
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.8 DSP First 2e
Recall that e j2π` = 1 when ` = integer.
z 4 = j = e jπ/2 = e jπ/2 e j2π`

1/4
⇒ z = e jπ/2 e j2π` = e jπ/8 e jπ`/2 ; ` = 0, 1, 2, 3

z ∈ e jπ/8, e j5π/8, |{z}
e j9π/8, |
e j13π/8
{z }
e j π /8 e− j3π /8
There are four different answers shown as the four vectors below:
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.9 DSP First 2e

N −1
z0 = e j2π/N ⇒ (z0 ) N −1 = e j2π/N
* 1−j2π/N
z0N −1 = e j2π (N −1)/N = e j2π (N )/N e j2π (−1)/N =
e j2π

e
−1
z0N −1 = e−j2π/N = e j2π/N = z0−1 = 1/z0
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.10 DSP First 2e
Multiply by 1 = e j2π` which is true when ` = integer;

this will generate all possible solutions.
− j = 1e−jπ/2 = e j2π` e−jπ/2

1/2
⇒ (− j) 1/2 = e j2π` e−jπ/2
 e−jπ/4
 when ` is even
= e jπ` e−jπ/4 = 
 e jπ e−jπ/4 = e j3π/4
 when ` is odd
1
 when ` is even
Because e jπ` = 
 e jπ
 when ` is odd
Thus, there are two solutions (shown below)
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
For each of the

as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.11 DSP First 2e
Use a calculator to check these

√
34 e−j1.03
z1 = √ = 0.707e−j1.275
68 e j0.245
z2 = 0.25 e jπ/2
z3 = e j5π/6 = e j2.618
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.12 DSP First 2e
When N = 56, we can write

NX−1
z=− e j2πk/N + e j2π (N −1)/N = e j2π(55)/56 = e−jπ/28 = e−j0.1122
k=0
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.13 DSP First 2e

√
p = 32 and q = 200 3 = 346.4102
Approach: Express the right-hand side (RHS) in Cartesian form using Euler’s formula. Then cross multiply by the LHS
denominator to get a single complex-valued equation.
√
p = 200 + jq (0.08) −0.5 + j 3/2

The real and imaginary parts yield two (linear) equations in the two unknowns, p and q. One of these equations has only p,
or only q, so solve it and then substitute into the other equation.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-A.14 DSP First 2e
e jπt 1.5
Definite integral is
jπ 1
√
2 −j3π/4
re jθ = e = 0.4502 e−j2.3562 .
π
Approach: The integral of an exponential is an exponential, but you end up with a j in the denominator because the exponent
contains a j. After evaluating at the limits of the definite integral, the numerator contains a complex number. Finally, convert
the complex numerator-denominator into polar form with a calculator.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

Appendix C
Fourier Series
C-1 Problem Solutions
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w
247
APPENDIX C. FOURIER SERIES
P-C.1 DSP First 2e
(a) Sketch the periodic function x(t) over the time interval −2T0 < t < 2T0 for the case t c = T0 /4.
x(t)
−2T0 −T0 − 14 T0 0 1 T0 2T0 t

4 T0
T0 /2
Z Ztc /2
1 1 2t c
(b) a0 = x(t)dt = dt =
T0 T0 T0
−T0 /2 −tc /2
(c) Determine a formula for the Fourier series coefficients ak in the finite Fourier representation (??). Your final result
should depend on t c and T0 .
T0 /2 Ztc /2
sin((2πt c /T0 )k)
Z
1 1
ak = x(t)e −j (2π/T0 )kt
dt = e−j (2π/T0 )kt dt = k = ±1, ±2, . . .
T0 T0 πk
s
w
ng
−T0 /2 −tc /2
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
(d) See plot below
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ak
ed
ng
W
(d)
in
d
tc = T0 /4 itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
1
tp
t
by
en
4
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
–8f0 –4f0 4f0 8f0

rt
0
y
f
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
ak
l
(e)
sa
ro
tc = T0 /10
st
or
de
ill
1
w
10
–8f0 –4f0 0 4f0 8f0 f

(e)
(f) We see by comparing the figures that the high-frequency components get bigger relative to the DC component as t c
decreases. That is, very narrow pulses require much more bandwidth for an accurate representation by a finite Fourier
series.

P-C.2 DSP First 2e
(a) Plot the periodic signal x(t) for −T0 ≤ t ≤ T0 .
x(t)
−T0 − 21 T0 1
2 T0 T0 t
−1
Figure P-C.2
Z0 T0 /2
Z T0 /2
Z T0 /2
Z
1 1 1 1
(b) a0 = x(t)dt + x(t)dt = x(τ + T0 /2)dτ + x(t)dt = 0
T0 T0 T0 T0
−T0 /2
s
0 0 0
w
ng
la
hi
where we have made the substitution t = T0 /2 + τ in the first integral.
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
(c) The same manipulation can be made on the general Fourier series coefficient integral.
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
T0 /2 itt
ni
e
te
Z0
of
th
m
ar
ni
er
Z
e
le
1 1
on
U
us
tp
ak = −j (2π/T0 )kt
dt + x(t)e−j (2π/T0 )kt dt
t
by
en
x(t)e
g
he
no
in
ud
d
T0 T0
ud
rt
te
is
st
c
fo
cl
−T0 /2
d
te
in
0
an
y
n
o
(
l
le
si
pr
k
es
so
or
T0 /2 T0 /2
or
is
w
s
w
ed
k
as
Z Z
1 1
or
x(τ + T0 /2)e−j (2π/T0 )k (τ+T0 /2) dτ +

id
hi
=
th
x(t)e−j (2π/T0 )kt dt

w
d
ov
t
an
of
of
is
pr
T0 T0
Th
rt
y
se
rit
is
pa
g
ur
d
0 0
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
0 k even
sa
ro
T0 /2 
st
or

T0 /2
de
Z 
1


= x(t)(1 − e−jπk )e−j (2π/T0 )kt dt = 
ill
 Z
2
w
T0 
 x(t)e−j (2π/T0 )kt dt k odd
0  T0


 0
Thus, when we have half-wave symmetry, the odd-indexed coefficients can be found by integrating over only half the
period.

P-C.3 DSP First 2e
4+t −4 ≤ t < 0
(
(a) x(t) =
0 0≤t<4
(b) Here is a plot of x(t) over the range −8 ≤ t ≤ 8 s.
x(t)
−8 −4 4 8 t
Figure P-C.3
area under curve 8

s
w
(c) The DC value is just the average value of the waveform so a0 = = = 1.
ng
la
hi
)
n
ht
eb
8 8
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-C.4 DSP First 2e
(a) A sketch of the periodic function x(t) over the time interval −10 ≤ t ≤ 20.
x(t)
−10 −5 5 10 t
Figure P-C.4
area under one period

(b) The DC coefficient: a0 = = (2 × 5)/10 = 1
10
Z10
s
1 1 w
ng
(c) a1 = 2e−j (2π/10)t dt = (e−j (2π/10)10 − e−j (2π/10)5 ) = j/π
la
hi
)
n
ht
eb
c
tio
10 −10 j (2π/10) ea
rig
W
a
t
py
in
in
e
5 em
co
d
s
i
W
r
ss
to
es
(d) Adding a constant only changes the DC coefficient. b0 = 1 + a0 and b1 = a1 . rld

c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-C.5 DSP First 2e
3
(a) Because each of the terms ak e jkt = e jkt is paired with its complex conjugate ak∗ e−jkt giving a real function of
2 + jk
t.
(b) Define ω0 = 1 and ak = 3/(2 + j k) then

3
X 3
X 10
X
x(t) = (− jak /2)e j (k+7)t
+ ( jak /2)e j (k−7)t
= bk e jkt
k=−3 k=−3 k=−10
− j1.5
 4 ≤ k ≤ 10
2 + j (k − 7)





where bk =  j1.5

−10 ≤ k ≤ −4
 2 + j (k + 7)





 0 otherwise
Note that b−k = bk as required for real x(t).
∗
b−7 b7
s
b−8 b−6 b6 w b8
ng
la
b−5
hi
b−9 b5 b9
)
n
ht
eb
c
tio
ea
b−10 b−4
rig
b4 b10
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
−10 −4
te
−7 0 4 7 10 ω
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
Figure P-C.5
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
(c) See the explanation in part (b).

w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-C.6 DSP First 2e
j5 j5
(a) Define ak = . Each of the terms ak e j8πkt = e j8πkt is paired with its complex conjugate
k + j1.5 k + j1.5
∗ − j5 −j8πkt j5
ak e j8πkt = e = e j8π (−k)t
k − j1.5 (−k) + j1.5
for k = 1, 2 giving a real function of t for the sum. Multiplying by cos(3πt) is still real.
(b) The frequencies in q(t) are ±5π, ±11π, ±13π, ±19π. Note that the frequencies ±3π have zero amplitude since a0 = 0.
The fundamental frequency is determined by the gcd of 5,11,13,19, which is 1, so ω0 = π.
(c) The DC value is zero. There is no component at ω = 0.
(d) Table listing all the nonzero Fourier coefficients of the signal q(t).
ωk −19π −13π −11π −5π 5π 11π 13π 19π
bk a−2 /2 a−2 /2 a−1 /2 a−1 /2 a1 /2 a1 /2 a2 /2 a2 /2
j5
(e) Spectrum plot showing only the positive frequency components. Note: ak = .
k + j1.5
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
a2 /2 em a2 /2
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
a1 /2 a1 /2
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
5π 11π 13π 19π ω

is
0
w
s
w
ed
k
as
or
e
id
hi
th
w
Figure P-C.6
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-C.7 DSP First 2e
 0.5

 k=0
(a) ak =  sin(πk/2)
 k,0
πk


(b) Sketch of the waveform y(t).
y(t) = 2x(t − T0 /2)
−2T0 −T0 − 14 T0 0 1 T0 2T0 t

4 T0
Figure P-C.7
 1
 k=0
bk = 2ak e−j (2π/T0 )kT0 /2 = 2ak e−jπk = 

sin(πk/2)
 2e−jπk
 k,0
πk
s
w
ng
 la
hi
)
n
ht
eb
c
tio
Note that both ak and bk are zero for k a nonzero even integer. This is because both x(t) − 12 and y(t) − 1 have half-wave
ea
rig
W
a
t
py
in
in
e
em
co
d
s
symmetry.
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-C.8 DSP First 2e
(a) Plot of x(t) over the interval −10 ≤ t ≤ 10 s with a period of 8 s.

x(t)
−8 −4 4 8 t
area under one period 7 × 5 + 3 × 3 44

(b) a0 = = = = 5.5
8 8 8
Z5 Z8
1 1
(c) ak = 7e −j (2π/8)kt
dt + 3e−j (2π/8)kt dt
8 8
0 5
s
w
ng
la
hi
(d)
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
Z5 Z8
i
W
r
ss
to
es
1 1 7(1 − e−j (π/4)k5 ) 3(e j (π/4)k5 − e−j (π/4)k8 )

rld
c
i
at
.D
ru
ak = −j (2π/8)kt
dt + 3e−j (2π/8)kt dt = +
o
7e
St
.
st
ed
ng
W
in
d
8 8 j (π/4)k8 itt j (π/4)k8

ni
e
te
of
th
m
ar
ni
er
e
0
le
5
on
U
us
tp
t
by
en
e−j (π/4)k5 ) 4(e j5πk/8 − e−j5πk/8 )

he
no
in
4(1 − 4 sin(5πk/8)
ud
d
ud
rt
= = e−j5πk/8 = e−j5πk/8
te
is
st
c
fo
cl
πk
te
in
an
j2πk j2πk
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
(e) Clearly, every eighth coefficient is zero since sin(5πk/8) = 0 for k a multiple of 8.
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-C.9 DSP First 2e
The Fourier series coefficients for the half-wave rectified sine wave are:
1

 k even
π(1 − k 2 )




ak =  ∓ j/4

k = ±1





 0
 k odd and ≥ 3
The fundamental period is T0 = 0.02 s and therefore the fundamental frequency is f 0 = 1/0.02 = 50 Hz. Therefore the
spectrum is
1
π −j
4
−1
3π
−1 −1
15π −1 −1
35π 63π 99π
0 50 100 200 300 400 500 f

s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
The Fourier series coefficients for the full-wave rectified sine wave are:
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
2
st
ed
ng
W
in
d
k even itt
ni
e
te

of
th
 m
ar
π(1 − 4k 2 )
ni

er
e
le

on
U
ak = 

us
tp
t
by
en
g
he
no
in
 2
ud
d

k=0
ud
rt
te

is
st

c
fo
cl
 π
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
The fundamental period is T0 = T1 /2 = 0.01 s and therefore the fundamental frequency is f 0 = 1/0.01 = 100 Hz. The
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
spectrum for the full-wave rectified sine wave is:

pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
2
ei
th
th
π
y
l
sa
ro
st
or
de
ill
w
−2
3π
−2 −2
15π −2 −2
35π 63π 99π
0 100 200 300 400 500 f

In both cases, the negative-frequency Fourier series coefficients are complex conjugates of the positive-frequency coefficients.

P-C.10 DSP First 2e
(a) Sketch of q(t) = p(t/5). The period is T0 = 5 × 0.1 = 0.5s and the pulse width is 5 × 0.04 = 0.2s.
p(t)
1
−.1 −.05 −.02 0 .02 .05 .1 t

q(t) = p(t/5)
1
−.5 −.25 −.1 0 .1 .25 .5 t

Figure P-C.10
Z.02 .02
1 e−j20πkt sin(0.4πk)
(b) ak = e−j20πkt dt = 10 =
.1 πk

− j20πk −.02
−.02
s
w
ng
la
hi
)
n
ht
eb
Z.1 c
tio
ea
rig
.1
W
a
e−j4πkt
t
py
in
1 sin(0.4πk)
in
e
em
bk = e−j4πkt dt = 2 =
co
d
s
i
W
r
.5 πk
ss
to
es
− j4πk −.1 rld

c
i
at
.D
ru
o
St
.
st
−.1
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
(c) Make the change of variable u = αt, and dt = du/α, in bk

er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
ZT0 ZT0
d
te
in
an
y
n
o
(
l
1 1
le
si
pr
bk = x(u)e−j (2π/T0 )ku du/α = x(u)e−j (2π/T0 )ku du = ak

es
so
or
or
is
w
s
T0 /α
w
ed
k
as
T0
or
e
id
hi
th
w
d
ov
t
an
0 0
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-C.11 DSP First 2e
(a) One-sided spectrum for x(t). The negative frequency components are complex conjugates.
j12
j6
j3
j2
0 3 6 12 18 f
Figure P-C.11
(b) Since 24 sin(6πt) = − j12e j6πt + j12e−j6πt it follows that the k = ±1 terms in x(t) are cancelled. Therefore, these
components of the spectrum are missing in the spectrum of y(t). All other components are the same as in the spectrum
of x(t).
s
w
ng
(c) The original 3 Hz fundamental is is not present. Now the fundamental is the gcd of 6, 12, 18 or f 0 = 6 Hz.
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
(d) For x(t) with f 0 = 3 Hz and the Fourier coefficients are em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
ak = {0, ± j12, ± j6, 0, ± j3, 0, ± j2}

in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
Thus y(t) has spectrum lines at {±6, ±12, ±18} hertz. The true fundamental frequency from the GCD is f 0 = 6 Hz.
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
With respect to 6 Hz the formula for x(t) consists of the first, second and third harmonics. The formula for the relabeled
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
Fourier coefficients is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
for k = ±1, ±2, ±3

pr
Th
 j6/k
s
rt
y
se
rit
is

pa
bk = a2k = 
g
ur
d
te
y
an
co
an
in
0 elsewhere
r
e
of
ei
th

th
y
l
sa
ro
st
or
de
ill
w

P-C.12 DSP First 2e
(a) T0 = 4 s
(b) f 0 = 1/4 = 0.25 Hz

area under one period (2 × 1/2) + (3 × 1) − (2 × 1/2)
(c) a0 = = = 0.75
4 4
(d) The simplest expression would use symmetric limits:
Z−1 Z1
1 1
ak = 3e−jπkt/2 dt + 2te−jπkt/2 dt
4 4
−2 −1
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-C.13 DSP First 2e
(a) T0 = 100 s
(b) a3 = 5e−j0.644 , a2 = 0, a1 = 2.236e−j0.464 , a0 = 1,

√
a−3 = 5e j0.644 , a−2 = 0, a−1 = 5e j0.464
(c) b3 = 3e−jπ/2 , b2 = 0, b1 = 2.236e−j0.464 , b0 = 2.2e jπ ,

√
b−3 = 3e j1.571 , b−2 = 0, b−1 = 5e j0.464
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-C.14 DSP First 2e
(a) N-th harmonic is N = 3.
(b) Since 3F0 = 2000, it follows that F0 = 2000/3 = 667 Hz
(c) The signal x 2 (t) is the square wave since its Fourier series coefficients are
sin(πk/2)
ak = k = ±1, ±2, . . .
πk
which are zero at the even harmonics.
s
w
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
W
in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

P-C.15 DSP First 2e
(a) First, the fundamental frequenty of s(t) = | sin(2πt)| is F0 = 2 Hz. The FWRC can be expressed as
x(t) = | cos(2πt)| = | sin(2π(t − .25))| = s(t − 0.25)
so if ak denotes the Fourier coefficients of s(t), then the Fourier coefficients of x(t) are
bk = ak e−j2π (2)(k)0.25 = ak e−jπk = (−1) k ak
(b) The half-wave rectified sine is defined in (??) as

0 ≤ t < T0 /2
(
sin((2π/T0 )t)
x(t) =
0 T0 /2 ≤ t < T0
where the fundamental period is T0 . Therefore, the FWRS can be expressed as
s(t) = x(t) + x(t − T0 /2)
i.e., we fill in the gaps in the HWRS by adding in a delayed version. Now if bk denotes the Fourier coefficients of the
HWRS, the additive and delay properties give us
ak = bk + bk e−j (2π/T0 )T0 /2 = (1 + e−jπk )bk w

s
ng
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
Note that the fundamental frequency of the HWRS x(t) is 1/T0 but when we delay and add a copy the resulting FWRS
in
e
em
co
d
s
i
W
r
has fundamental frequency 2/T0 . This is confirmed by observing that (1 + e−jπk )bk = 0 for k odd, which zeroes out
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
the odd harmonics.

in
d
itt
ni
e
te
of
th
m
ar
ni
er
e
le
on
U
us
Since the odd harmonics are missing, the fundamental frequency is twice the fundamental of the HWRS. Thus, if we
tp
t
by
en
g
he
no
in
ud
d
reindex the coefficients, we get ak = b2k (1 + e−j2πk ) = 2b2k . Recall the Fourier coefficients of the HWRS given in
ud
rt
te
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
(??), which are

le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
1
id
hi
th
w
d
ov
t
an
for k even
of
of
is

pr

 π(1 − k 2 )
Th
rt

se
rit

is
pa

bk = 
g
ur
d
te
for k = ±1
y
∓ j 14
an
co
an
in
r

e
of
ei

th

th
for k = ±3, ±5, . . .

e

0
y
l
sa
ro
st

or
de
ill
Finally, we can write

w
2
ak = b2k =
π(1 − 4k 2 )
(c) Since (−1) k = e−jπk = e−j (2π/T0 )kT0 /2 , multiplying the Fourier coefficients by (−1) k is equivalent to delaying the
waveform by T0 /2. For the pulse wave with period 0.1s and pulse width 0.04s the plots are:
p(t)
1
−.2 −.02 0 .02 .05 .1 t

q(t) = p(t − 0.05)
1
−.3 −.2 −.07 −.03 0 .03 .05 .07 .1 .3 t

P-C.16 DSP First 2e
(a) Sketch of the triangle wave x(t), and its derivative y(t) = d
dt x(t).
x(t)
1
−T0 − T20 0 T0 T0 T0 t
4 2
4
y(t) = dx(t)
dt
T0
−T0 − T20 0 T0 T0 t
2
Figure P-C.16
(b) The derivative property gives the result bk = ( j2πk/T0 )ak .
(c) The Fourier coefficients of the square wave of amplitude 4/T0 are
4 2 sin(πk/2) 8 sin(πk/2)
bk = = with b0 = 0
πk πkT0 s
w
ng
T0
la
hi
)
n
ht
eb
c
tio
ea
rig
W
a
t
py
in
in
e
em
co
d
(d) Using the derivative property, we can write
s
i
W
r
ss
to
es
rld
c
i
at
.D
ru
o
St
.
st
ed
ng
8 sin(πk/2) 4 sin(πk/2)
in
d
( j2πk/T0 )ak = =⇒ ak = − j itt

ni
e
te
of
th
m
ar
ni
πkT0 π2 k 2
er
e
le
on
U
us
tp
t
by
en
g
he
no
in
ud
d
ud
rt
te
for k , 0. Note that a0 = 0, and thatt all even harmonics are zero since the triangle wave has half-wave symmetry.
is
st
c
fo
cl
d
te
in
an
y
n
o
(
l
le
si
pr
k
es
so
or
or
is
w
s
w
ed
k
as
or
e
id
hi
th
w
d
ov
t
an
of
of
is
pr
Th
rt
y
se
rit
is
pa
g
ur
d
te
y
an
co
an
in
r
e
of
ei
th
th
y
l
sa
ro
st
or
de
ill
w

CHAPTER 3. SPECTRUM
P-3.1 DSP First 2e
(a) x(t) = 11 + 14 cos(100πt − π/3) + 8 cos(350πt − π/2)

(b) Since the gcd of 50 and 175 is 25, x(t) is periodic with period T0 = 1/25 = 0.04 s.
(c) Negative frequencies are implicit in the cosine terms. They are needed to give a real signal when combined with
their corresponding positive-frequency terms.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d
©J. H. McClellan, R. W. Schafer, & M. A. Yoder July 29, 2015

CHAPTER 3. SPECTRUM
P-3.2 DSP First 2e
(a) A single plot labeled with complex amplitudes is sufficient. The spectrum consists of the lines
{(400, 5e jπ/4 ), (−400, 5e− jπ/4 ), (600, 3.5e− jπ/3 ), (−600, 3.5e jπ/3 ), (800, 1.5), (−800, 1.5)}
where the frequencies are in Hz.
5e− jπ/4 5e jπ/4

3.5e jπ/3 3.5e− jπ/3
1.5 1.5
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
-
−800 −600 −400 0 400 600 800 f
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
(b) The signal x(t) is periodic with fundamental frequency 200 Hz or period 1/200 = 0.005 s since the gcd of
e rt ss fo U
gr hi in e
{400, 600, 800} is 200.

th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
(c) The spectrum has the added components {(500, 2.5e jπ/2 ), (500, 2.5e− jπ/2 )}. Now we seek the gcd of {400, 500, 600, 800}
de o rse de ot
ill le u vi pr
so the fundamental frequency changes to 100 Hz and the period is 0.01 s.

w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.3 DSP First 2e
(a) x(t) = 2 cos(2t) + 4 cos(4t − π/3) + 2 cos(6t + π/4)

(b) The spectrum is {(2, 1), (−2, 1), (4, 2e− jπ/3 ), (−4, 2e jπ/3 ), (6, e jπ/4 ), (−6, e− jπ/4 )}
The frequencies are all in rad/s.
2e jπ/3 2e− jπ/3

e− jπ/4 1 1 e jπ/4
)
ω
-
0 2 4 6
eb
−6 −4 −2
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.4 DSP First 2e
(a) Determine a formula for x(t) as the real part of a sum of complex exponentials.
Use Euler’s formula for the sine function obtaining

!3
3 e j27πt − e− j27πt
sin (27πt) =
2j
1 j27πt3
= e − 3e j27πt2 e− j27πt + 3e j27πt e− j27πt2 − e− j27πt3
)
eb
−8 j
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
3 1
e
d on g. in t la
m ld a
= sin(27πt) − sin(81πt)

4 4
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
(b) What is the fundamental period for x(t)?
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
The fundamental frequency is 27/2 so the fundamental period is 2/27.
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
(c) Plot the spectrum for x(t).

ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
The spectrum is {(27π, − j3/8), (−27π, j3/8), (81π, j/8), (−81π, − j/8)}, where the frequencies are in rad/s.
or eir is p rk
o
w
an his
T
th d
j3/8 − j3/8
− j/8 j/8
ω
-
−81π −27π 0 27π 81π

CHAPTER 3. SPECTRUM
P-3.5 DSP First 2e
There are seven spectral components:

{(−3, 1/(1 − j3)), (−2, 1/(1 − j2)), (−1, 1/(1 − j)), (0, 1), (1, 1/(1 + j)), (2, 1/(1 + j2)), (3, 1/(1 + j3))},
where the frequencies are all in rad/s.
Putting all the complex numbers in polar form gives the following plot:
1
0.707e j0.25π 0.707e− j0.25π
0.447e j0.3524π 0.447e− j0.3524π
)
eb
er or in ing
0.316e j0.398π 0.316e− j0.398π
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
ω
-
−3 −2 −1 0 1 2 3
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.6 DSP First 2e
(a) In this case we need to find the gcd of 36 and 84, which is 12. Thus, the fundamental frequency is ω0 = 1.2π rad/s.
(b) The fundamental period is T0 = 2π/ω0 = 1/0.6 = 5/3 s.
(c) The DC value is −7.
(d) The ak coefficients are nonzero for k = 0, ±3, ±7.
Here is the list of the nonzero Fourier series coefficients in a table.
)
k 0 3 7
eb
−7 −3
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
ak 3e− jπ/4 4e jπ/3 7e jπ 4e− jπ/3 3e jπ/4
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.7 DSP First 2e
(a) The phasor representation is z(t) = Ae j2π( f c − f∆ )t + Be j2π ( f c + f∆ )t

(b)
z(t) = e j2π f c t ( Ae− j2π f∆ t + Be j2π f∆ t )

= e j2π f c t ( A cos(2π f ∆ t) − j A sin(2π f ∆ t) + B cos(2π f ∆ t) − j B sin(2π f ∆ t))
= e j2π f c t [( A + B) cos(2π f ∆ t) − j ( A − B) sin(2π f ∆ t)]
)
eb
er or in ing
ed id n
W
Therefore, the real part is
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
x(t) = <{z(t)} = ( A + B) cos(2π f ∆ t) cos(2π f c t) + ( A − B) sin(2π f ∆ t) sin(2π f c t)
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
so C = A + B and D = A − B. If A = B = 1, C = 2 and D = 0, so using the trigonometric identity cos α cos β =
ity s w g us d S
1 1
2 cos(α − β) + 2 cos(α + β), it follows that
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
x(t) = 2 cos(2π f ∆ t) cos(2π f c t) = 2[ 12 cos(2π( f c − f ∆ )t) + 21 cos(2π( f c + f ∆ )t)]
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
(c) The values are A = 1 and B = −1. In this case,

w sa co ro is
or eir is p rk
o
w
an his
e j2π f∆ t − e− j2π f∆ t e − e− j2π f c t

! j2π f c t !
x(t) = 2
T
d
2j 2j
th
j2π( f c + f ∆ )t
1 j2π( f c − f ∆ )t
− e− j2π( f c − f∆ )t + e− j2π( f c + f∆ )t

= −2 e −e
The spectrum is {(− f c − f ∆, −0.5), (− f c + f ∆, 0.5), ( f c − f ∆, 0.5), ( f c + f ∆, −0.5)}, and the plot is
−0.5 0.5 0.5 −0.5
−( f c + f ∆ ) ( fc + f∆)
-
− fc −( f c − f ∆ ) 0 ( fc − f∆) fc f

CHAPTER 3. SPECTRUM
P-3.8 DSP First 2e
(a) Using Euler’s relation we get
x(t) = 10 + 10e jπ/4 e j2π(100)t + 10e− jπ/4 e− j2π(100)t + 5e j2π(250)t + 5e− j2π (250)t
The gcd of 100 and 250 is 50 so f 0 = 50 and therefore N = 5. The nonzero Fourier coefficients are, therefore,
a−5 = 5, a−2 = 10e− jπ/4 , a0 = 10, a2 = 10e jπ/4 , and a5 = 5.
(b) The signal is periodic because all the frequencies are multiples of 50 Hz. Therefore, the fundamental period is
T0 = 1/50 = 0.02s.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(c) Here is the spectrum plot of this signal versus f in Hz.
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
10e− jπ/4 10 10e jπ/4
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
5 5
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
-
−250 −100 0 100 250 f
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.9 DSP First 2e
(a) Use phasors to show that x(t) can be expressed in the form
x(t) = A1 cos(ω1 t + ϕ1 ) + A2 cos(ω2 t + ϕ2 ) + A3 cos(ω3 t + ϕ3 )
where ω1 < ω2 < ω3 ; i.e., find values of the parameters A1, A2, A3, ϕ1, ϕ2, ϕ3, ω1, ω2, ω3 .
Using Euler’s relation we get
)
eb
f g
x(t) = 12 − 3.5 je− jπ/3 e jπt + 3.5 je jπ/3 e− jπt 0.5e j13πt + .5e− j13πt
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
= 12 cos(13πt) − 1.75 je− jπ/3 (e j14πt + e− j12πt ) + 1.75 je jπ/3 (e j12πt + e− j14πt )
.
r
or ud a uc y
w cl le tr p
= 12 cos(13πt) + 1.75e− jπ/2 e− jπ/3 (e j14πt + e− j12πt ) + 1.75e jπ/2 e jπ/3 (e j12πt + e− j14πt )
e in nt ns co
D
th k ( de f i es
of or stu e o tat
= 12 cos(13πt) + 1.75e− j5π/6 e j14πt + 1.75e− j5π/6 e− j12πt + 1.75e j5π/6 e j12πt + 1.75e j5π/6 e− j14πt
ity s w g us d S
is
te f t ss th nite
= 12 cos(13πt) + 3.5 cos(14πt − 5π/6) + 3.5 cos(12πt + 5π/6)
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
The requested parameters are easily picked off from this equation.
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
(b) Sketch the two-sided spectrum of this signal on a frequency axis. Be sure to label important features of the plot.
w sa co ro is
or eir is p rk
Label your plot in terms of the numerical values of Ai , ϕi , and ωi .

o
w
an his
T
d
12 12
th
3.5e j5π/6 3.5e− j5π/6 3.5e j5π/6 3.5e− j5π/6
ω
-
−14π −13π −12π 0 12π 13π 14π

CHAPTER 3. SPECTRUM
P-3.10 DSP First 2e
(a) Assume without limitation that ω2 − ω1 > 0. For periodicity with period T0 we require that ω0 = 2π/T0 . This
means that k 1 ω0 = ω2 − ω1 and k 2 ω0 = ω2 + ω1 , where k1 and k 2 are integers and k 2 > k1 .
(b) Part (a) gives two equations for ω1 and ω2 . If we solve them in terms of ω0 we get ω1 = (k2 − k1 )ω0 /2 and
ω2 = (k2 + k 1 )ω0 /2, so the main condition is that both ω1 and ω2 are integer multiples of ω0 /2. This is the most
general condition.
Therefore, the relationship between ω2 and ω1 is
)
eb
k2 + k1
er or in ing
ed id n
W
no the iss tea s
ω2 = ω1
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
k2 − k1
.
r
or ud a uc y
if x(t + T0 ) = x(t). Thus, ω2 could be an integer multiple of ω1 if k2 − k1 divides into k 2 + k1 with no remainder,
w cl le tr p
e in nt ns co
D
th k ( de f i es
but that is not necessary for periodicity of x(t).
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.11 DSP First 2e
(a) The gcd of 40, 60, 120 is 20 so ω0 = 20π and the fundamental period is T0 = 2π/ω0 = 0.1s. The finite Fourier series
has components indexed by 0, ±2, ±3, ±6 so N = 6. The coefficients are a0 = 2, a±2 = 2e∓ jπ/5 , a±3 = 1.5e∓ jπ/2 ,
a±6 = 2e∓π/3
(b) The spectrum is
{(−120π, 2e jπ/3 ), (−60π, 1.5e jπ/2 ), (−40π, 2e jπ/5 ), . . .
(0, 2), (40π, 2e− jπ/5 ), (60π, 1.5e− jπ/2 ), (120π, 2e− jπ/3 )}
)
eb
er or in ing
ed id n
2
W
2e jπ/3 2e jπ/5 2e− jπ/5 2e− jπ/3
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
1.5e jπ/2 1.5e− jπ/2
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
ω
-
in o e r
0 40π 60π 120π

y y p d le d
−120π −60π −40π

ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
(c) Now the fundamental frequency is 10π rad/s because the gcd of 20, 40, 50, and 120 is 10. Therefore, the period is
an his
T
d
T0 = 2π/10π = 1/5 = 0.2s. The spectrum is the same as in part (b) except there are two additional components at
th
±50π rad/s: (−50π, 5e jπ/6 ) and (50π, 5e− jπ/6 ).

CHAPTER 3. SPECTRUM
P-3.12 DSP First 2e
(a) Make a table of the frequencies of the tones of the octave beginning with middle C, assuming that the A above
middle C is tuned to 440 Hz.
Note name C C# D E[ E F F#
Note number 40 41 42 43 44 45 46
Frequency 262 277 294 311 330 349 370
Note name F# G G# A B[ B C
Note number 46 47 48 49 50 51 52
)
eb
er or in ing
ed id n
370 392 415 440 466 494 523
W
no the iss tea s
itt W tio
Frequency
w
t p W em ch
e
d on g. in t la
m ld a
(b) The formula for the frequency f as a function of note number n is
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
f = 440 · 2(n−49)/12
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
(c) The spectrum would have the form:

e rt ss fo U
gr hi in e
th a a ly by
k
{(−440, a3∗ ), (−370, a2∗ ), (−294, a1∗ ), (294, a1 ), (370, a2 ), (440, a3 )}
in o e r
y y p d le d
ro n an o te
st f a s d s ec
To sound like a musical chord, the coefficients should have similar magnitudes, but the phases could be arbitrarily
de o rse de ot
ill le u vi pr
chosen. A chord from a real instrument would have overtones (higher harmonics) of each individual note.
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.13 DSP First 2e
(a) The frequency of the DC component is by definition 0. The waveform is periodic with period 25 ms so the frequency
is 1/0.025 = 40 Hz.
(b) The DC level is (20 − 10)/2 = 5, the amplitude of the cosine is (20 + 10)/2 = 15, and the cosine is delayed by
0.005 s, so
x(t) = 5 + 15 cos(2π(40)(t − .005)) = 5 + 15 cos(80πt − 0.4π)
)
(c) x(t) = 5 + 7.5e j (80πt−0.4π) + 7.5e− j (80πt−0.4π) = 7.5e j0.4π e− j80πt + 5 + 7.5e− j0.4π e j80πt
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(d) Plot of the two-sided spectrum of the signal x(t).
.
r
or ud a uc y
w cl le tr p
7.5e j0.4π 7.5e− j0.4π
e in nt ns co
D
th k ( de f i es
of or stu e o tat
5
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
ω
-
−80π 0 80π
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.14 DSP First 2e
(a) Using symmetry we obtain

√ √
X−1 = 2 − j 2 = 2e− jπ/4 X2 = 8e jπ/3 ω1 = 70π ω2 = −100π
(b) x(t) = 20 + 4 cos(70πt + π/4) + 16 cos(100πt + π/3)

(c) The gcd of 70 and 100 is 10, so the fundamental frequency of the signal is f 0 = 5 Hz and the fundamental period
is T0 = 1/5 = 0.2 s.
)
eb
er or in ing
ed id n
W
no the iss tea s
(d) Note that the −20 ≤ 4 cos(70πt+π/4)+16 cos(100πt+π/3) ≤ 20 since the individual terms satisfy −4 ≤ 4 cos(70πt+
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
π/4) ≤ 4 and −16 ≤ 16 cos(100πt + π/3) ≤ 16. The value ±20 would be attained only if the phases of the two
.
r
or ud a uc y
cosines are such that 4 cos(70π(t − t 0 )) + 16 cos(100π(t − t 0 )).
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.15 DSP First 2e
(a) x(t) = cos2 (7πt − 0.1π)

We need to express x(t) in terms of complex exponentials
2
x(t) = 0.5e j (7πt−0.1π) + 0.5e− j (7πt−0.1π) = 0.25e j0.2π e− j14πt + 0.5 + 0.25e− j0.2π e j14πt

0.5
)
eb
er or in ing
0.25e j0.2π 0.25e− j0.2π
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
ω
-
0 14π
of or stu e o tat
−14π
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
(b) y(t) = cos2 (7πt − 0.1π) cos(77πt + 0.1π)

ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr

x(t) = 0.25e j0.2π e− j14πt + 0.5 + 0.25e− j0.2π e j14πt 0.5e j0.1π e j77πt + 0.5e− j0.1π e− j77πt
w sa co ro is
or eir is p rk
o
w
= 0.125e j0.1π e− j91πt + 0.25e− j0.1π e− j77πt + 0.125e− j0.3π e− j63πt

an his
T
d
+ 0.125e j0.3π e j63πt + 0.25e j0.1π e j77πt + 0.125e− j0.1π e j91πt

th
0.25e− j0.1π 0.25e j0.1π
0.125e j0.1π 0.125e− j0.3π 0.125e j0.3π 0.25e− j0.1π
ω
-
−91π −77π −63π 0 63π 77π 91π
Note:
ω axis is not to scale.

CHAPTER 3. SPECTRUM
P-3.16 DSP First 2e
(a) Plotting spectrum of v(t) versus f in kHz,
0.5e− j0.3π 0.5e j0.3π
f (kHz)
-
0 3
)
−3
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
(b) The spectrum for x(t) versus f in kHz.
e in nt ns co
D
th k ( de f i es
of or stu e o tat
0.75 0.75
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
0.25e− j0.3π 0.25e j0.3pi 0.25e− j0.3π 0.25e j0.3π

de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
f (kHz)
-
0 677 680 683
an his
−683 −680 −677

T
th d
Note: f axis is not to scale.

CHAPTER 3. SPECTRUM
P-3.17 DSP First 2e
(a) The gcd of 105 and 180 is 15, so the given frequencies are the 7th and 12th harmonics of f 0 = 15 Hz.
(b) x(t) = 22 cos(2π(105)t − 0.4π) + 14 cos(2π(180)t − 0.6π)
(c) Simplify the numerical values for the complex amplitudes, i.e., phases should be in [−π, π].
x 2 (t) = 22 cos(2π(105)(t − 0.05) − 0.4π) + 14 cos(2π(180)(t − 0.05) − 0.6π)

= 22 cos(2π(105)t − 10.5π − 0.4π) + 14 cos(2π(180)t − 18π − 0.6π)
)
= 22 cos(2π(105)t − 10π − 0.9π) + 14 cos(2π(180)t − 18π − 0.6π)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
= 22 cos(2π(105)t − 0.9π) + 14 cos(2π(180)t − 0.6π)
m ld a
.
r
or ud a uc y
w cl le tr p
Note that even multiples of 2π rad can be dropped from the equation. Thus, the spectrum is:
e in nt ns co
D
th k ( de f i es
{(−180, 7e j0.6π ), (−105, 11e j0.9π ), (105, 11e− j0.9π ), (180, 7e− j0.6π )} where the frequencies are in hertz. Therefore
of or stu e o tat
ity s w g us d S
the plot of the spectrum looks just like Fig. P-3.17 except the phase is different at frequencies ±105.
is
te f t ss th nite
e rt ss fo U
gr hi in e
(d) The effect of this operation is simply to increase all the frequencies by 105 Hz, or, in other words, to shift the
th a a ly by
k
in o e r
y y p d le d
spectrum of x(t) to the right by 105 Hz.

ro n an o te
st f a s d s ec
de o rse de ot
Therefore the spectrum line at −105 Hz will move to f = 0. The DC component is therefore, the value of the
ill le u vi pr
w sa co ro is
spectrum at f = −105, or 11e j0.9π .

or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.18 DSP First 2e
(a) The spectrum of y(t) is the spectrum of x(t) with an added DC component of size 8.
11e j0.4π 11e− j0.4π

8
7e j0.6π 7e− j0.6π
)
eb
er or in ing
ed id n
-
0 105 180 f
W
no the iss tea s
itt W tio
−180 −105
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
(b) The spectrum of z(t) is the same as that of x(t) with the addition of components of size 9e± j0.8π at frequencies
of or stu e o tat
ity s w g us d S
±40 Hz.
is
te f t ss th nite
e rt ss fo U
gr hi in e
11e j0.4π 11e− j0.4π

th a a ly by
k
in o e r
y y p d le d
9e− j0.8π 9e j0.8π

ro n an o te
st f a s d s ec
7e j0.6π 7e− j0.6π

de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d
-
−180 −105 −40 0 40 105 180 f
(c) The fundamental frequency is the gcd of 40, 105, 180, which is 5 Hz.
(d) The derivative operation multiplies each spectrum component by j2π f , where f is the frequency of the complex
exponential component. So we get
− j2π(105)11e j0.4π j2π(105)11e− j0.4π
− j2π(180)7e j0.6π j2π(180)7e− j0.6π
-
−180 −105 0 105 180 f

CHAPTER 3. SPECTRUM
P-3.19 DSP First 2e
(a)
X1 = 8e− jπ/3 and ω1 = −10π
(b) Here is the plot of the spectrum of x(t).
8e jπ/3 8e− jπ/3
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
ω
-
0 10π
th k ( de f i es
−10π
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
(c) The symmetry implies that ωb = 20π and B = +4 j. Furthermore, symmetry requires that ωa = 0. To find A, ωc ,
e rt ss fo U
gr hi in e
th a a ly by
and ϕ we can write y(t) as y(t) = 0.5x(t)e jϕ e jω c t + 0.5x(t)e− jϕ e− jω c t , which shows that the spectrum of y(t)
k
in o e r
y y p d le d
ro n an o te
will consist of the sum of scaled copies of the spectrum of x(t) shifted right (up) by ωc and left (down) by ωc . In
st f a s d s ec
de o rse de ot
order to have only three components we must choose ωc = 10π so that two of the shifted spectrum lines over lap
ill le u vi pr
w sa co ro is
at ω = 0.
or eir is p rk
o
w
an his
4e jπ/3 e jϕ + 4e− jπ/3 e− jϕ

T
th d
4e jπ/3 e− jϕ 4e− jπ/3 e jϕ
ω
-
−20π 0 20π
Now, 4e− jπ/3 e jϕ

√ = 4e √ so ϕ = −π/6. Finally, note that the DC value can be written as A = 8 cos(π/3 + ϕ) =
− jπ/2
8 cos(π/6) = 8 3/2 = 4 3.

CHAPTER 3. SPECTRUM
P-3.20 DSP First 2e
(a) The gcd of 40 and 90 is 10, so f 0 = 10 Hz.

(b) The fundamental period is T0 = 1/ f 0 = 1/10 = 0.1 s.
(c) From the plot, the DC value is 0.5.
(d) With f 0 = 10, the harmonics are k = 0, ±4, ±9.
k −9 −4 0 4 9
)
eb
er or in ing
ak 0.4e− j2 0.6e j1.4 0.5 0.6e− j1.4 0.4e j2
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.21 DSP First 2e
dψ
(a) The instantaneous frequency is ωi (t) = dt = 2αt + β, so ω1 = ωi (0) = β and ω2 = ωi (T2 ) = 2αT2 + β.
(b) The instantaneous frequency versus time is ωi (t) = 80t + 27
(c) Here is the plot of the instantaneous frequency (in Hz) versus time over the range 0 ≤ t ≤ 1 sec.
ω i (t )
f i (t) = 2π
6
107
2π
)

eb

er or in ing
ed id n
W
no the iss tea s
itt W tio

w
t p W em ch

e
d on g. in t la
m ld a


.
r
or ud a uc y

w cl le tr p
e in nt ns co
D

th k ( de f i es

of or stu e o tat

ity s w g us d S

is
te f t ss th nite

27
e rt ss fo U

gr hi in e
2π
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
-
w sa co ro is
0 1 t
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.22 DSP First 2e
(a) The general form for the chirp signal is x(t) = cos(αt 2 + βt + ϕ). The instantaneous frequency of this signal
is ωi (t) = 2αt + β. From this we observe that ω1 = 2π f 1 = 2π(4800) = ωi (0) = β. To obtain α, we note that
ω2 = 2π(800) = ωi (2) = 2α(2) + β = 4α + 9600π so α = −2000π. Therefore, the signal is
x(t) = cos(−2000πt 2 + 9600πt + ϕ)
where ϕ is an arbitrary phase constant.

(b) The instantaneous frequency is ωi = 800πt + 500π, so ω1 = ωi (0) = 500π and ω2 = ωi (3) = 800π(3) + 500π =
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
2900π rad/s.
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.23 DSP First 2e
(a) The instantaneous frequency is ωi (t) = 2αt + β. Substituting the given parameters gives α = 4π and β = 2π, so
the signal with the given parameters is x(t) = cos(4πt 2 + 2πt + ϕ).
(b–f) The solution to this problem is given in the following figure. Note that the times at which f i (t) is equal to 4 Hz and
8 Hz are indicated with dashed lines. Careful scrutiny of the plots confirms that the waveform of the chirp signal
does match the waveforms of the 4 Hz and 8 Hz constant-frequency sinusoids at the corresponding two times.
Instantaneous Frequency of Chirp Signal
frequency in Hz
10
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
5
m ld a
.
r
or ud a uc y
w cl le tr p
0
e in nt ns co
D
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
th k ( de f i es
of or stu e o tat
4 Hz Signal
1
ity s w g us d S
is
te f t ss th nite
0.5
e rt ss fo U
gr hi in e
0
th a a ly by
k
in o e r
y y p d le d
-0.5
ro n an o te
st f a s d s ec
de o rse de ot
-1
ill le u vi pr
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

w sa co ro is
Chirp Signal
or eir is p rk
1
o
w
an his
T
d
0
th
-1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
8 Hz Signal
1
0.5
0
-0.5
-1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
time in seconds

CHAPTER 3. SPECTRUM
P-3.24 DSP First 2e
1
(a) f 1 (t) = √ (c) f 3 (t) = e2t /π
2π (t)
(b) f 2 (t) = t/π (d) f 4 (t) = − sin(2πt)
Solution to Problem 3.24

0.8
frequency in Hz
0.6
)
eb
0.4
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
0.2

0
.
r
or ud a uc y
w cl le tr p
0 0.5 1 1.5 2 2.5 3 3.5 4
e in nt ns co
D
th k ( de f i es
frequency in Hz
of or stu e o tat
ity s w g us d S
0.5
is
te f t ss th nite
e rt ss fo U
0
gr hi in e
th a a ly by
k
in o e r
y y p d le d
-0.5
ro n an o te
st f a s d s ec
de o rse de ot
-1
ill le u vi pr
-3 -2 -1 0 1 2 3
w sa co ro is
or eir is p rk
o
frequency in Hz
150
w
an his
T
d
100
th
50
0
0 0.5 1 1.5 2 2.5 3
frequency in Hz
1
0.5
0
-0.5
-1
0 0.5 1 1.5 2 2.5 3
time in seconds

CHAPTER 3. SPECTRUM
P-3.25 DSP First 2e
∞
X
(a) Let x(t) be given by the Fourier series x(t) = ak e j (2π/T0 )k t . Then it follows that
k=−∞
∞
X :1 X
∞
(2π/T )k (0)

x(0) = e j
ak 0 = ak .
k=−∞ k=−∞
(b) Let f 3 = − f 2 = f 0 and f 4 = − f 1 = 3 f 0 so that from the spectrum we can write
x(t) = 12 cos(2π f 0 t + π/4) + 4 cos(6π f 0 t + 3π/4)
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
√ √ √
e
d on g. in t la
m ld a
Therefore x(0) = 12 cos(π/4) + 4 cos(3π/4) = 6 2 − 2 2 = 4 2. Now if we add the coefficients of the Fourier
.
r
series we get
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
√
of or stu e o tat
a1 + a2 + a3 + a4 = 2e− j3π/4 + 6e− jπ/4 + 6e jπ/4 + 2e j3π/4 = 12 cos(π/4) + 4 cos(3π/4) = 4 2
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.26 DSP First 2e
The equations corresponding to the spectra are:
The matches are (a) (3) x 1 (t) = 4 cos(4πt + π) + 4 cos(6πt + π/2)

(b) (1) x 2 (t) = 2 cos(4πt + π/4) + 4 cos(6πt − 0.333π)
(c) (2)
x 3 (t) = −3 + 2 cos(4πt + π/4)
(d) (5)
(e) (4) x 4 (t) = −2 + 4 cos(4πt + π)
x 5 (t) = 4 cos(2πt + π) + 4 cos(4πt + π)
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 3. SPECTRUM
P-3.27 DSP First 2e
(a) x(t) = cos(−250πt 2 ) (d) x(t) = cos(100πt) cos(400πt)

(b) x(t) = cos(100πt − π/4) + cos(400πt) (e) x(t) = cos(200πt 2 )
(c) x(t) = cos(1000πt − 250πt 2 ) (f) x(t) = cos(30e2t )
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.1 DSP First 2e
(a) A = 11, ϕ = −π/3, and ω/ f s = 7π/9 = ω̂1 .

(b) A = 11, ϕ = +π/3, and ω/ f s = 7π/6 → ω̂1 = 5π/6.
(c) A = 11, ϕ = −π/3, and ω/ f s = 7π/3 → ω̂1 = π/3.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.2 DSP First 2e
x 1 [n] = x 3 [n] and x 2 [n] = x 4 [n] = x 5 [n]
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.3 DSP First 2e
(a) y(t) = 8 cos(2160πt − π/3)

(b) y(t) = 4 cos(2880πt + π/4)
(c) y(t) = 2 cos(1440πt − π/5)
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.4 DSP First 2e
(a) Sketch will have four spectral lines at ω̂ = ±4000π, ±12000π. The complex amplitude of each spectral line will be
0.25.
0.25 0.25 0.25 0.25

6
ω
-
−12000π −4000π 0 4000π 12000π
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(b) The Nyquist rate for x(t) is 12,000 samples/s.
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
(c) The Nyquist rate for r (t) is 15 × 106 samples/s.
th k ( de f i es
of or stu e o tat
ity s w g us d S
(d) The Nyquist rate for v(t) is 7 × 106 samples/s.
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.5 DSP First 2e
(a) The Nyquist rate for x(t) is 50 samples/s.

(b) The Nyquist rate for r (t) is 72 samples/s.
(c) The Nyquist rate for v(t) is 45 samples/s.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.6 DSP First 2e
(a) Sketch will have four spectral lines: at ω̂ = ±0.2π with complex amplitudes equal to 13e± jπ/4 , and at ±0.5π with
complex amplitudes of 7e± j3π/4 .
There is no aliasing when f s = 2000 samples/s.
13e− jπ/4 6 13e jπ/4
7e− j3π/4 7e j3π/4
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
-
e in nt ns co
D
−π −0.5π −0.2π 0 0.2π 0.5π π ω̂
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
(b) The reconstructed output when the D-to-C rate is f s = 1600 samples/s will be y(t) = 26 cos(320πt + π/4) +
gr hi in e
th a a ly by
k
in o e r
14 cos(800πt + 3π/4).
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.7 DSP First 2e
(a) No, because there is aliasing.

(b) Sketch will have four spectral lines: at ω̂ = ±4π/9 with complex amplitudes equal to 13e± jπ/4 , and at ±2π/9 with
complex amplitudes of 7e± j3π/4 .
When f s = 450 samples/s, there will be an alias for the lines at ±500 Hz, i.e., the spectral line first maps to
ω̂ = ±20π/9 and then aliases to ±2π/9.
)
eb
13e− jπ/4 13e jπ/4
er or in ing
ed id n
6
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
7e− j3π/4 7e j3π/4
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
-
2π/9 4π/9 π
in o e r
0 ω̂
y y p d le d
−π −4π/9 −2π/9
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
(c) The reconstructed output when f s = 450 samples/s is y(t) = 26 cos(400πt + π/4) + 14 cos(100πt + 3π/4).
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.8 DSP First 2e
(a) Yes. f Nyquist = 1000 Hz.

(b) Sketch will have two spectral lines at ω̂ = ±4π/7 with complex amplitudes equal to 6e± jπ/4 .
When f s = 700 samples/s, there will be a folded alias for the lines at ±500 Hz, i.e., the spectral line first maps to
ω̂ = ±10π/7 and then aliases (via folding) to ∓4π/7. The complex amplitudes of the coincident spectral lines must
be summed.
6e− jπ/4 6e jπ/4 = 13e jπ/4 + 7e− j3π/4
)
eb
6
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite -
−π −4π/7 0 4π/7 π ω̂
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
(c) The reconstructed output when f s = 700 samples/s is y(t) = 12 cos(400πt + π/4).
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.9 DSP First 2e
(a) ω = 2π(15200) rad/s.

(b) f s = 120 samples/s.
(c) f s = 75 samples/s.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.10 DSP First 2e
(a) ω1 = 18π rad/s.

(b) ω̂ = 0.6π rad.
(c) ω̂1 = π/7 and ω̂2 = −π/7, but b[n] involves a folded alias so ω̂2 = π/7.
(d) Unequal because the phases are different: a[n] = cos(πn/7 + 0.1π) and b[n] = cos(πn/7 − 0.1π).
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.11 DSP First 2e
(a) Average is 9 samples per period.

(b) ω = 8000π rad/s.
(c) Average is 0.9 samples per period, or 9 samples taken every 10 periods.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.12 DSP First 2e
Frequencies will be: 1000 Hz, 4000 Hz (but x(t) will have a different phase), and 6000 Hz.
x 1 (t) = 9.1 cos(2000πt + 0.4π)

x 2 (t) = 9.1 cos(8000πt − 0.4π)
x 3 (t) = 9.1 cos(6000πt + 0.4π)
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.13 DSP First 2e
(a) f = 1600 Hz. Phase will be −2.03 rad.

(b) fsamp = 9600 Hz.
(c) f = 2200 Hz. Duration = 12.8 s.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.14 DSP First 2e
(a) Sketch will have six spectral lines at ω = ±2000π, ±(2000π ± 6000) rad/s.
In hertz, these frequencies are f ≈ ±47.07, ±1000, ±1954.93 Hz.
The complex amplitudes of the spectral lines are 41 e± jπ/2, 1.5e± j0, 14 e∓ jπ/2 .
1.5 1.5
6
)
j
− 4j j
− 4j
eb
er or in ing
4 4
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
-
−47 0

−1955 −1000 47 1000 1955 f (Hz)
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
(b) Not periodic.
th k ( de f i es
of or stu e o tat
ity s w g us d S
(c) f s > 3909.859 Hz.
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.15 DSP First 2e

(b) y(t) = cos(2π(1750)t + 0.2π)
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.16 DSP First 2e

(b) x(t) = cos(2π(7300)t − 0.2π)
ω̂
(c) f = 8000 − f s Hz, and ϕ = −0.2π rad.
2π
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.17 DSP First 2e
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.18 DSP First 2e
(a) ω1 = 12π rad/s.

(b) ω̂ = 0.8π rad.
(c) ω1 = ω2 = 200π rad/s.
(d) The three possible answers are 2π(14 27 ), 2π(28 74 ), 2π(33 13 ) rad/s.
( )
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.19 DSP First 2e
(a) Run the code to make the stem plot.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
(b) Period of x[n] in the plot is 5 samples.
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
(c) ω̂0 = 0.4π rad, and x[n] = 2 cos(0.4πn − 0.6π).

y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
(d) 2π(380/100) = 7.6π > 2π and 7.6π = 4(2π) − 0.4π.

ill le u vi pr
w sa co ro is
or eir is p rk
Over the total duration of 0.3 s, a 380 Hz continuous-time signal would have 114 periods, but x[n] only has 6
o
w
an his
complete periods.
T
th d

P-4.20 DSP First 2e
(a) Expect 34 zero crossings in 17 periods. Number of positive peaks = number of negative peaks = 17.
(b) 11 zero crossings, 6 positive peaks and 6 negative peaks.
(c) T = 0.2 s, so f = 5 Hz. Sampling a 15-Hz sinusoid at 20 Hz (Ts = 0.05) s. involves aliasing, and a 5-Hz sinusoid
is a folded version of the 15-Hz sinusoid.
(d) x[n] = 1.005 cos(0.4πn − 0.1).
(e) To obey the Sampling Theorem, Ts < (1/30) s, but to make a smooth plot Ts < 0.05 s would be needed.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.21 DSP First 2e
x(t) = (3 − j0.5e jπt + j0.5e− jπt )(0.5e j13πt+ jπ/2 + 0.5e− j13πt− jπ/2 )
= 1.5e j13πt+ jπ/2 + 1.5e− j13πt− jπ/2 + 0.25e j14πt + 0.25e− j14πt + 0.25e j12πt+ jπ + 0.25e− j12πt− jπ
= 3 cos(13πt + π/2) + 0.5 cos(14πt) + 0.5 cos(12πt + π)
(a) ω1 = 12π rad/s, A1 = 0.5, ϕ1 = π; ω2 = 13π rad/s, A2 = 3, ϕ2 = π/2; ω3 = 14π, A3 = 0.5, ϕ3 = 0;

(b) Use the second line above to plot six spectrum lines.
)
−3 j 3j
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
6
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
0.5 0.5 0.5 0.5
th k ( de f i es
of or stu e o tat
ity s w g us d S
-
0 12π 13π 14π ω (rad/s)
is
te f t ss th nite
−14π −13π −12π
e rt ss fo U
gr hi in e
th a a ly by
k
(c) Nyquist rate is 28π rad/s, or 14 Hz.
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.22 DSP First 2e
(a) Sampling rate is greater than 2(300π) rad/s, or 300 Hz.

(b)
x[n] = x(t) t=n/250 = 2 cos(2π(50)n/250 + π/2) + cos(2π(150)n/250)

= 2 cos(2πn/5 + π/2) + cos(6πn/5) = 2 cos(0.4πn + 0.5π) + cos(0.8πn)
(c) Plot four lines; two at ±0.4π with complex amps equal to 1e± j0.5π ; and two at ±0.8π with complex amps equal to
)
eb
er or in ing
ed id n
10.5. Note: amplitudes are not to scale.
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
10.5 10.5
e in nt ns co
D
6
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
e− jπ/2 e jπ/2
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
-
0 0.4π 0.8π π ω̂
ill le u vi pr
−π −0.8π −0.4π
w sa co ro is
or eir is p rk
o
w
(d) f s = 300π rad/s, or 150 Hz.

an his
T
th d

P-4.23 DSP First 2e
(a) The discrete-time exponential when flashing at 15 flashes/s is p[n] = e+ j2π(13)n/15+ jπ/2 = e j2π(−2)n/15+ jπ/2 . Thus,
at each flash the spot moves −4π/15 rad, which is 48◦ clockwise.
(b) p(t) = p[n] = e j2π(−2)(15t )/15+π/2

n=15t
appears to rotate through an angle of −4π every second. This is 2 rev/s clockwise.
(c) Thirteen flashes in one second; disk moving CCW which is negative. The disk moves 2π(−15/360) + 2π` rad per
flash. Thus the disk moves 2π(13)(−1/24 + `) in one second. When ` = 0, the rotation speed is 13/24 = 0.54 rev/s
)
eb
CCW. When ` < 0, the rotation speed is |13` − 13/24| rev/s CCW. When ` > 0, the rotation speed is |13` − 13/24|
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
rev/s CW. For example, when ` = 1, the speed is 12.46 rev/s CW.
.
r
Complex Exponential Derivation of Rotation Speeds (r) in rev/s:
or ud a uc y
w cl le tr p
e in nt ns co
D
The discrete-time exponential after flashing at f s flashes/s is p[n] = e+ j2π (r )n/ f s . Thus, to get the movement of
th k ( de f i es
of or stu e o tat
15◦ CCW, which is −2π/24 rad/s CCW, we must determine whether or not there is is an integer ` for which the
ity s w g us d S
is
te f t ss th nite
following is true:
e rt ss fo U
gr hi in e
th a a ly by
k
* 1e j2π` e− j2π(15/360)
in o e r
y y p d le d
ro n an o te
p[1] = e+ j2π (r )(1)/ f s =

p[0]
st f a s d s ec

de o rse de ot
ill le u vi pr
e+ j2π (r )/ f s = e− jπ/12 e j2π` ⇒ 2π(r)/ f s = 2π` − 2π/24 ⇒ r = f s (` − (1/24))

w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.24 DSP First 2e
(a) p(t) = e+ j2π(12)t rotates through an angle of 24π every second.

(b) The discrete-time exponential after flashing at n flashes/s is p[m] = e+ j2π(12)m/n . Thus, given n, we must determine
whether or not there is is an integer ` for which the following is true:
p[1] = p[0]
+ j2π (12)(1)/n
e = e j2π` ⇒ 24π/n = 2π` ⇒ 12/n = `
)
eb
So, n must be a divisor of 12, which gives flashing rates of n = 1, 2, 3, 4, 6, 12 flashes/s.
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(c) When n = 13, p[1] = e+ j2π(12)/13 = e− j2π(1)/13 , so the disk rotates clockwise by 2π/13 rad, or 360/13 = 27.69◦ ,
.
r
or ud a uc y
for every flash. The observed rotation is 1 rev/s clockwise.
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.25 DSP First 2e
(a) Plot 8 unit-length vectors (with zvect). The angles will be {0.2π, 0.5π, 0.8π, 0, 0.3π, −0.5π, 0.5π, −0.5π}.
(b) Period is 20.
(c) Plot 5 unit-length vectors (with zvect). The angles will be {0, π/15, 4π/15, −14π/15, −14π/15}.
(d) Shortest period is 30.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.26 DSP First 2e
(a) θ[n] = 0.07π, 1.75π, 7π are the angles of the vectors. When reduced to the −π to +π range, these angles are
θ[n] = 0.07π, −0.25π, π.
(b) The instantaneous frequency of the signal below is 2π(44.8 × 103 )t Hz. The plot runs from t = 0 to t = 200/8000 =
1/40 s, so it starts at f = 0 and ends at (1/40)44.8 × 103 = 1120 Hz.
x(t) = x[n] = cos(π(0.7 × 10−3 × 64 × 106 )t 2 )

n=8000t
)
= cos(π(44.8 × 103 )t 2 )
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(c) 2800 Hz
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-4.27 DSP First 2e
ALPHA = 437.5, BETA = 2000, fs = 8000
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.1 DSP First 2e
y[n] = 7x[n] + x[n − 3] − 5x[n − 4]
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.2 DSP First 2e

√
G = 9e− j0.5π(3) − 9e− j0.5π(4) e− j0.5πn = 9 2 e+ j3π/4

)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.3 DSP First 2e
(a) First of all, y[n] = 0 for n = −2, −1. The system is causal. Then y[n] = n + 1 for n = 0, 1, 2, 3; and y[n] = 4 for
n ≥ 4.
(b) Plot with Matlab.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
(c)
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
0 n<0
de o rse de ot

ill le u vi pr


w sa co ro is

y[n] =  n + 1 0≤n<L

or eir is p rk

o
w
L

n≥L
an his

T
d

th

P-5.4 DSP First 2e
(a) Use numerical convolution table:

n n<0 0 1 2 3 4 5 6 7 n>7
x[n] 0 1 2 3 2 1 1 1 1 1
h[n] 0 2 −3 2 0
h[0]x[n] 0 2 4 6 4 2 2 2 2 2
h[1]x[n−1] 0 0 −3 −6 −9 −6 −3 −3 −3 −3
)
eb
h[2]x[n−2] 0 0 0 2 4 6 4 2 2 2
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
y[n] 0 2 1 2 −1 2 3 1 1 1
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d
(b)
(c) h[n] = 2δ[n] − 3δ[n − 1] + 2δ[n − 2]

P-5.5 DSP First 2e
Here are the figures:

2
2 xŒn
xŒn v2 Œn
Unit
Unit Delay
3
)
eb
Delay
er or in ing
ed id n
3
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
xŒn 1
.
r
or ud a uc y
v1 Œn
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
Unit
ity s w g us d S
Unit
Delay
is
te f t ss th nite
Delay 2 2
e rt ss fo U
gr hi in e
th a a ly by
xŒn 2
k
yŒn yŒn
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
(a) (b)
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.6 DSP First 2e
(a) P = N + M
(b) Input signal is time-shifted by N1 . Thus, the output starts at N3 = N1 , and the last nonzero value in the output is at
N4 = N2 + M. The length of y[n] is N4 − N3 + 1 = N2 − N1 + M + 1.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.7 DSP First 2e
(a) Make a plot in Matlab.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
(b) First of all, y[n] = 0 for n = −2, −1. The system is causal. Then y[n] = 1 for n = 0; and y[n] = (−0.5) n + (−0.5) n−1
gr hi in e
th a a ly by
k
in o e r
y y p d le d
for n = 1, 2, 3, 4, 5, 6, 7, 8. The nonzero values starting at n = 0 are

ro n an o te
st f a s d s ec
1.0, 0.5, −0.25, 0.125, −0.0625, 0.03125, −0.015625, 0.0078125, −0.00390625

de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
(c) Causality implies that y[n] = 0 for n < 0.

o
w
Running onto the signal (partial overlap) region which is 0 ≤ n < L−1: sum (−0.5) k for k = 0 to k = n
an his
T
th d
(−0.5) 0 − (−0.5) n+1 1 − (−0.5) n+1

n
X
(−0.5) k = =
k=0
1 − (−0.5) 1.5
Complete overlap region: sum (−0.5) k for k = n − L + 1 to k = n which is n ≥ L−1:

n
X (−0.5) n−L+1 − (−0.5) n+1 (−0.5) −L − 1
(−0.5) k = = (−0.5) n+1
k=n−L+1
1 − (−0.5) 1.5

P-5.8 DSP First 2e
Filter order is M = 4, and bk = {3, 7, 13, 9, 5}
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.9 DSP First 2e
Impulse response is h[n] = 13δ[n]−13δ[n−1]+13δ[n−2]. The output can be obtained via convolution: y[n] = h[n]∗ x[n].
Assume that the input is nonzero for −∞ < n < ∞, which implies that x[n − 2] is equal to x[n]. Also, x[n − 1] will be
equal to 0 for n odd, and equal to 1 for n even.
 26
 n odd
y[n] = 13x[n] − 13x[n − 1] + 13x[n − 2] = 26x[n] − 13x[n − 1] = 
 −13
 n even
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.10 DSP First 2e
(a) y[n] = x[n] cos(0.2πn) (1) linear YES, (2) time-invariant NO, and (3) causal YES.
(b) y[n] = x[n] − x[n − 1] (1) linear YES, (2) time-invariant YES, and (3) causal YES.
(c) y[n] = |x[n]| (1) linear NO, (2) time-invariant YES, and (3) causal YES.
(d) y[n] = Ax[n] + B, where A and B are nonzero constants.
(1) linear NO, (2) time-invariant YES, and (3) causal YES.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.11 DSP First 2e
(a) Plot of y[n]
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
(b) Use LTI properties with x[n] = 7 (δ[n] − δ[n − 1]) +7 (δ[n − 1] − δ[n − 2]).
of or stu e o tat
ity s w g us d S
| {z }
first x[n] in table
is
te f t ss th nite
e rt ss fo U
gr hi in e
Thus, y[n] = 7 (δ[n] − δ[n − 1] + 2δ[n − 3]) +7 (δ[n − 1] − δ[n − 2] + 2δ[n − 4])
th a a ly by
k
in o e r
y y p d le d
ro n an o te
| {z }
st f a s d s ec
first x[n] in table

de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.12 DSP First 2e
Note that the Delay by 2 block in (b) is the cascade of two unit delays.
1
xŒn
3
v3 Œn
xŒn
Delay
by 2
4
Unit
)
eb
er or in ing
ed id n
Delay
W
no the iss tea s
2
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
xŒn 1

v2 Œn
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
Unit
of or stu e o tat
ity s w g us d S
Unit Delay
2
is
te f t ss th nite
Delay 4
e rt ss fo U
gr hi in e
th a a ly by
k
xŒn 2
in o e r
y y p d le d
ro n an o te
st f a s d s ec
v1 Œn
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
Unit
o
Delay
w
Delay
an his
by 2 1 3
T
d
xŒn 4
th
yŒn yŒn
(a) (b)

P-5.13 DSP First 2e
(a) Use the LTI properties. The output will be y2 [n] = y1 [n] − 2y1 [n − 2].
y2 [n] = δ[n] − 3δ[n − 1] − 3δ[n − 2] − 2δ[n − 2] + 6δ[n − 3] + 6δ[n − 4]

= δ[n] − 3δ[n − 1] − 5δ[n − 2] + 6δ[n − 3] + 6δ[n − 4]
(b) A = 1 and B = −1.

(c) Using LTI properties,
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
h[n] = ( A/3)y1 [n] + (B/3)y1 [n − 1]
e
d on g. in t la
m ld a
= (1/3) (δ[n] − 3δ[n − 1] − 3δ[n − 2]) − (1/3) (δ[n − 1] − 3δ[n − 2] − 3δ[n − 3])
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
= (1/3)δ[n] − (4/3)δ[n − 1] + δ[n − 3]
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.14 DSP First 2e
Do the convolutions with h[n] = δ[n] − aδ[n − 1]

(a) x[n] ∗ h[n] = a n u[n] ∗ (δ[n] − aδ[n − 1]) = a n u[n] − aa n−1 u[n − 1] = a n (u[n] − u[n − 1]) = δ[n]
| {z }
δ[n]
(b) Use the result from the previous part. h[n] ∗ a n (u[n] − u[n − 8]) = h[n] ∗ a n u[n]) − h[n] ∗ a8 a n−8 u[n − 8]
| {z } | {z }
δ[n] a 8 δ[n−8]
(c) Sketch the plot which has only two nonzero stems, 1 at n = 0, and a8 at n = 8.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.15 DSP First 2e
Use trial and error, and just do the convolutions

(a) x[n] ∗ δ[n − 3] = x[n − 3], so we want x[n − 3] = 5u[n − 2] − u[n − 6]. Unshift to get x[n] = 5u[n + 1] − u[n − 3].
(b) y[n] = u[n − 2] − u[n − 3] which simplifies to y[n] = δ[n − 2]
(c) x[n] − x[n − 1] = y[n] = δ[n − 2] − δ[n − 7].
Rewrite y[n] as y[n] = u[n − 2] − u[n − 3] − u[n − 7] − u[n − 8].
| {z } | {z }
δ[n−2] δ[n−7]
Rewrite again as y[n] = u[n − 2] − u[n − 7] − (u[n − 3] − u[n − 8]) .
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
| {z } | {z }
t p W em ch
e
d on g. in t la
m ld a
x[n] x[n−1]

(d) x[n] + x[n − 1] = y[n] = −8δ[n] − 8δ[n − 3].
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
Mimic the previous part to obtain x[n] = −8u[n] + 8u[n − 3]
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.16 DSP First 2e
It is easy to verify these; just do the convolutions

(a) h[n] = δ[n − 1]
(b) h[n] = δ[n] − δ[n − 1]
(c) h[n] = δ[n − 1] − 0.5δ[n − 2] because ( 12 ) n−1 u[n − 1] − 12 ( 12 ) n−2 u[n − 2] is zero for n ≥ 2.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.17 DSP First 2e
If an FIR filter with impulse response h[n] is used to deconvolve, then
y[n] = h[n] ∗ (δ[n] + δ[n − 1]) = h[n] + h[n − 1]

0
We want y[0] = δ[0] = 1 ⇒ 1 = y[0] = h[0] + h[−1] ⇒ h[0] = 1
>

1
We want y[1] = δ[1] = 0 ⇒ 0 = y[1] = h[1] + h[0] ⇒ h[1] = 1

)
eb
er or in ing
ed id n
W
no the iss tea s
At this point we know that the order of the filter is at least M = 1. For an M th order FIR filter the coefficient of z −M must
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
be nonzero, i.e., b M , 0. With that fact, the following shows that y[M + 1] will also be nonzero, so y[n] cannot be equal
.
r
or ud a uc y
to δ[n].
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
M
ity s w g us d S
X
h[n] = bk δ[n − k] where b M , 0
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k=0
k
in o e r
*0
y y p d le d
ro n an o te
st f a s d s ec

⇒ y[M + 1] = h[M
+ 1] + h[M] = h[M] = b M , 0
de o rse de ot

ill le u vi pr
w sa co ro is

or eir is p rk
o
w
an his
T
th d

P-5.18 DSP First 2e
(a) Let x 1 [n] = δ[n], x 2 [n] = δ[n], and x 3 [n] = δ[n],
S1 : h1 [n] = δ[n] + δ[n − 1]

S2 : h2 [n] = δ[n] − δ[n − 2]
S3 : h3 [n] = δ[n − 1] − δ[n − 2]
(b) Carry out the convolutions:
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
h[n] = h1 [n] ∗ h2 [n] ∗ h3 [n]
t p W em ch
e
d on g. in t la
m ld a
h[n] = (δ[n] + δ[n − 1) ∗ (δ[n] − δ[n − 2]) ∗ (δ[n − 1] − δ[n − 2])
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
| {z }
th k ( de f i es
h[n] = (δ[n] + δ[n − 1) ∗ (δ[n − 1] − 2δ[n − 2] + δ[n − 3])
of or stu e o tat
ity s w g us d S
h[n] = δ[n − 1] − δ[n − 2] − δ[n − 3] + δ[n − 4]
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
(c) Get the filter coefficients from h[n], and then
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
y[n] = x[n − 1] − x[n − 2] − x[n − 3] + x[n − 4]

w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-5.19 DSP First 2e
In this case, the three systems are as follows:
S1 : y1 [n] = x[n] + 2x[n − 1] + 3x[n − 2] + 4x[n − 3] h1 [n] = δ[n] + 2δ[n − 1] + 3δ[n − 2] + 4δ[n − 3]
S2 : y2 [n] = x[n] + x[n − 1] + x[n − 2] + x[n − 3] h2 [n] = δ[n] + δ[n − 1] + δ[n − 2] + δ[n − 3]
S3 : y[n] = w[n] − w[n − 1] h3 [n] = δ[n] − δ[n − 1]
where w[n] = y1 [n] + y2 [n]. The impulse responses are also given above
)
(a) We need three blocks and one adder. Systems S 1 and S 2 are in parallel; their outputs are added and the sum is
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
processed by S 3 .
e
d on g. in t la
m ld a
.
r
(b) We can get h[n] from the combination of the individual systems, and the write the difference equation from the
or ud a uc y
w cl le tr p
e in nt ns co
D
filter coefficients.
th k ( de f i es
of or stu e o tat
h[n] = (h1 [n] + h2 [n]) ∗ h3 [n] ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
h[n] = (2δ[n] + 3δ[n − 1] + 4δ[n − 2] + 5δ[n − 3]) ∗ (δ[n] − δ[n − 1])

th a a ly by
k
in o e r
y y p d le d
ro n an o te
h[n] = 2δ[n] + δ[n − 1] + δ[n − 2] + δ[n − 3] − 5δ[n − 4]

st f a s d s ec
de o rse de ot
ill le u vi pr
y[n] = 2x[n] + x[n − 1] + x[n − 2] + x[n − 3] − 5x[n − 4]

w sa co ro is
or eir is p rk
o
w
an his
If we had the z-transform at this point we would write

T
th d

H (z) = 1 + 2z −1 + 3z −2 + 4z −3 + 1 + z −1 + z −2 + z −3 1 − z −1

H (z) = 2 + 3z −1 + 4z −2 + 5z −3 1 − z −1 = 2 + z −1 + z −2 + z −3 − 5z −4

P-6.1 DSP First 2e
y[n] = x[n] − x[n − 1] = e j (0.4πn−0.5π) − e j (0.4π (n−1)−0.5π) = (1 − e− j0.4π )e j (0.4πn−0.5π) = H (e j0.4π )e j (0.4πn−0.5π)
H (e j0.4π ) = 1 − e− j0.4π = 1.1756e j0.3π

Therefore, A = 1.1756, ϕ = 0.3π − 0.5π = −0.2π, and ω̂0 = 0.4π.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.2 DSP First 2e
(a)
y[n] = x[n] + x[n − 1] = (e j0.1πn + e j0.4πn ) + (e j0.1π (n−1) + e j0.4π (n−1) )

= (1 + e− j0.1π )e j0.1πn + (1 + e− j0.4π )e j0.4πn
= H (e j0.1π )e j0.1πn + H (e j0.4π )e j0.4πn
where H (e j ω̂ ) = 1 + e− j ω̂ .
)
eb
er or in ing
ed id n
W
no the iss tea s
(b) The system is LTI, so superposition applies. Each complex exponential is modified by the frequency response
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
evaluated at its frequency.
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
(c)
th k ( de f i es
of or stu e o tat
ity s w g us d S
H (e j0.1π ) = 1 + e− j0.1π = A1 e jϕ1 = 1.9754e− j0.05π
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
H (e j0.1π ) = 1 + e− j0.4π = A2 e jϕ2 = 1.6180e− j0.2π

k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.3 DSP First 2e

2
(a) y[n] = e j0.1πn + e j0.4πn = e j0.2πn + 2e j0.5πn + e j0.8πn
(b) Clearly, the squaring operation produces three complex exponential signals each with a different frequency that is
not one of the original two frequencies.
(c) The system is clearly not linear since the output cannot be written as the sum of two terms involving the two input
frequencies. The principle of superposition is not satisfied, therefore the system is NOT linear.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.4 DSP First 2e
(a) H (e j ω̂ = 2 − 2e− j ω̂ + 2e− j ω̂2 = 2e− j ω̂ (e j ω̂ − 1 + e− j ω̂ ) = 2e− j ω̂ (2 cos(ω̂) − 1)

Thus, R(e j ω̂ ) = 4 cos(ω̂) − 2 and n0 = 1.
(b) The period = 2π
10
|H(ej ω̂ )|
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
0

−π −π/2 0 π/2 π 3π/2 2π 5π/2 3π
.
r
or ud a uc y
w cl le tr p
π
e in nt ns co
D
H(ej ω̂ )
th k ( de f i es
of or stu e o tat
ity s w g us d S
0
is
te f t ss th nite
6
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
−π
st f a s d s ec
−π −π/2 0 π/2 π 3π/2 2π 5π/2 3π

de o rse de ot
ill le u vi pr
(c) ω̂
w sa co ro is
or eir is p rk
o
(d) The output is zero whenever the magnitude of the frequency response is zero; i.e., when |2 cos(ω̂) − 1| = 0 or when
w
an his
cos(ω̂) = 0.5. The solution is therefore ω̂ = π/3 rad.

T
th d
(e) We know that for a cosine input with frequency ω̂0 = 0.5, the output is
y[n] = |H (e j0.5π )| cos(0.5πn + ∠H (e j0.5π ))
where H (e j0.5π ) = 2e− j0.5π (2 cos(0.5π) − 1) = −2e− j0.5π = 2e j0.5π
⇒ y[n] = 2 cos(0.5πn + 0.5π) = 2 cos(0.5π(n + 1)).

P-6.5 DSP First 2e
(a) H (e j ω̂ ) = 3 + 2e− j ω̂ + e− j ω̂2

(b) y[n] = H (e j0 ) + 2|H (e j0.5π )| cos(0.5πn + 0.1π + ∠H (e j0.5π ))
√
where H (e j0 ) = 6 and H (e j0.5π ) = 3 − 2e− j0.5π + e− jπ = 2e− j0.25π .
√ √
Thus, y[n] = 6 + 2 2 cos(0.5πn + 0.1π − 0.25π) = 6 + 2 2 cos(0.5πn − 0.15π)
(c)
)
0 n<0
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio

w
3 n=0

t p W em ch
e

d on g. in t la
m ld a

y[n] = 3u[n] + 2u[n − 1] + u[n − 2] = 


5 n=1
.
r

or ud a uc y

w cl le tr p
6 n≥2

e in nt ns co
D

th k ( de f i es

of or stu e o tat
ity s w g us d S
(d) The unit step input is identical to a suddenly applied complex exponential of zero frequency. Thus, for an FIR filter
is
te f t ss th nite
like this, the steady state response is simply H (e j0 ) times 1. The steady state is reached at n = 2 because the filter
e rt ss fo U
gr hi in e
th a a ly by
k
order is 2.
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.6 DSP First 2e
(a) H (e j ω̂ ) = 1 − e− j ω̂4 = e− j ω̂2 (e j ω̂2 − e− j ω̂2 ) = 2 je− j ω̂2 sin(2ω̂)

(b) y[n] = 4H (e j0 ) + |H (e j0.25π )| cos(0.25πn − 0.25π + ∠H (e j0.25π ))
where H (e j0 ) = 0 and H (e j0.25π ) = 2 j (− j) sin(π/2) = 2. Thus,
y[n] = 2 cos(0.25πn − 0.25π)
(c) Since the impulse response has length 5 samples, the output reaches the steady state at n = 4 so the output due to
)
eb
the suddenly applied signal is identical to the output in (b) for n ≥ 4.
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.7 DSP First 2e
(a) h[n] = 3δ[n] − δ[n − 2]

(b) We need to expand H (e j ω̂ ) in terms of complex exponentials.
j2ω̂ + e− j2ω̂
− j3ω̂ e
!
j ω̂
H (e ) = 5e = 2.5e− j ω̂ + 2.5e− j5ω̂
2
Therefore h[n] = 2.5δ[n − 1] + 2.5δ[n − 5].
)
eb
(c) We should recognize this as the form of an L sample running sum whose frequency response is
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
L−1
sin(ω̂L/2) X − j ω̂n
.
r
or ud a uc y
H (e j ω̂ ) = e− j ω̂(L−1)/2
w cl le tr p
= e
e in nt ns co
D
sin(ω̂/2)
th k ( de f i es
of or stu e o tat
n=0
ity s w g us d S
Thus, it follows that L = 10 and the impulse response is
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
1 0≤n≤9
in o e r
(
y y p d le d
ro n an o te
h[n] = u[n] − u[n − 10] =

st f a s d s ec
0 otherwise
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.8 DSP First 2e
(a) To get the difference equation we need to expand the factors of H (e j ω̂ ) so that we get powers of e− j ω̂ .
√
H (e j ω̂ ) = (1 − e− j ω̂ )(1 − 2 cos(3π/4)e− j ω̂ + e− j2ω̂ ) = (1 − e− j ω̂ )(1 − 2e− j ω̂ + e− j2ω̂ )
√ √
= 1 + ( 2 − 1)e− j ω̂ − ( 2 − 1)e− j2ω̂ − e− j3ω̂
= 1 + 0.4142e− j ω̂ − 0.4142e− j2ω̂ − e− j3ω̂
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
y[n] = x[n] + 0.4142x[n − 1] − 0.4142x[n − 2] − x[n − 3]
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
(b)
th k ( de f i es
of or stu e o tat
ity s w g us d S
0 n<0
is
te f t ss th nite 

1 n=0

e rt ss fo U
gr hi in e



th a a ly by
k

y[n] = u[n] + 0.4142u[n − 1] − 0.4142u[n − 2] − u[n − 3] =  1.4142 n=1

in o e r
y y p d le d
ro n an o te
1 n=2
st f a s d s ec



de o rse de ot


ill le u vi pr
0 n≥3


w sa co ro is
or eir is p rk

o
w
(c) We need to find the values of ω̂ such that H (e j ω̂ ) = 0. These are easily found from the given factored form as
an his
T
d
ω̂ = 0 and ω̂ = ±3π/4.
th

P-6.9 DSP First 2e
(a) We need to multiply out the factors of H (e j ω̂ ) to get:
H (e j ω̂ ) = 1 + 0.5e− j ω̂ − 0.25e− j2ω̂ + 0.25e− j3ω̂
y[n] = x[n] + 0.5x[n − 1] − 0.25x[n − 2] + 0.25x[n − 3]
)
eb
(b) By definition, the impulse response is:
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
h[n] = δ[n] + 0.5δ[n − 1] − 0.25δ[n − 2] + 0.25δ[n − 3]
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
(c) We need to find the values of ω̂ such that H (e j ω̂ ) = 0. The factored form shows that this can be true for values of
of or stu e o tat
ity s w g us d S
ω̂ such that 1 + e− j2ω̂ = 0 or ω̂ = ±π/2 rad.
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.10 DSP First 2e
(a) Plot {3, 0, −2, 1} for 0 ≤ n ≤ 3.

(b) The first entry in the table says that the impulse response is h[n] = δ[n] − δ[n − 3]. Therefore, y[n] = x[n] − x[n − 3].
For the given input the output is
y[n] = 3δ[n] − 2δ[n − 2] + δ[n − 3] − 3δ[n − 3] + 2δ[n − 5] − δ[n − 6]

= 3δ[n] − 2δ[n − 2] − 2δ[n − 3] + 2δ[n − 5] − δ[n − 6]
)
eb
Plot {3, 0, −2, −2, 0, 2, −1} for 0 ≤ n ≤ 6.
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(c) In this case we can use the third result in the table, which says that the frequency response at ω̂ = π/3 is H (e jπ/3 ) = 2;
.
r
i.e., the phase shift is zero. Therefore, the output for this input is y[n] = 2 cos(π(n − 3)/3)
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
(d) It is FALSE because the impulse response is δ[n] − δ[n − 3], so the frequency response is H (e j ω̂ ) = 1 − e− j3ω̂
of or stu e o tat
ity s w g us d S
which is nonzero at ω̂ = π/2.
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.11 DSP First 2e
(a) The zeros marked with o are at multiples of 2π/8 with gaps at 0, ±2π, ±4π, . . ..
1
D8 (ω̂)
−1
)
eb
er or in ing
ed id n
W
no the iss tea s
−3π −2π −π 0 π 2π 3π
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
Figure P-6.11
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
(b) The period is 4π.
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
(c) The maximum value occurs at ω̂ = 0. Using the small angle approximation for the sine functions we get
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
(Lω̂/2)
ro n an o te
D L (0) ≈ =1
st f a s d s ec
de o rse de ot
L(ω̂/2)
ill le u vi pr
w sa co ro is
or eir is p rk
We get the same result using L’ Hospital’s Rule.

o
w
an his
T
th d

P-6.12 DSP First 2e
(a) From the difference equation, we can write down H (e j ω̂ ) as
H (e j ω̂ ) = 1 − 3e− j ω̂ + 3e− j2ω̂ − e− j3ω̂ = (1 − e− j ω̂ ) 3 = e− j ω̂3/2 (e j ω̂/2 − e− j ω̂/2 ) 3

= e− j ω̂3/2 (2 j) 3 (sin(ω̂/2)) 3 = 8e− j ω̂3/2 (− j)(sin(ω̂/2)) 3
= 8 sin3 (ω̂/2)e j (−π/2−3ω̂/2)
)
eb
er or in ing
ed id n
10
W
no the iss tea s
itt W tio
w
t p W em ch
|H(ej ω̂ )|
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
5
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
0
e rt ss fo U
−2π −3π/2 −π −π/2 0 π/2 π 3π/2 2π

gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
Figure P-6.12
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
(b)
or eir is p rk
o
w
an his
(c) We need to determine the frequency response at 0 and 0.5π.

T
d
√ √
These are H (e j0 ) = 0 and H (e j0.5π ) = 8(1/ 2) 3 e j (−π/2−3π/4) = 2 2e− j7π/4 . Therefore,
th
y[n] = 5H (e j0 ) + 6|H (e j0.5π )| cos(0.5π(n − 1) + ∠H (e j0.5π ))

√
= 12 2 cos(0.5π(n − 1) − 7π/4)
(d) h[n] = δ[n] − 3δ[n − 1] + 3δ[n − 2] − δ[n − 3]

(e)
y[n] = 5H (e j0 ) + 6|H (e j0.5π )| cos(0.5πn + ∠H (e j0.5π )) + 9h[n − 3]

√
= 12 2 cos(0.5πn − 7π/4) + 9h[n − 3]
where h[n] is as determined in part (d).

P-6.13 DSP First 2e
(a)
S1 : H1 (e j ω̂ ) = 1 − e− j ω̂
S2 : H2 (e j ω̂ ) = 1 + e− j2ω̂
S3 : H3 (e j ω̂ ) = e− j ω̂ + e− j2ω̂
(b) H (e j ω̂ ) = (1 − e− j ω̂ )(1 + e− j2ω̂ )(e− j ω̂ + e− j2ω̂ )
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
(c) Multiply out all the factors in H (e j ω̂ ) and then pick off the coefficients to get the difference equation.
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
H (e j ω̂ ) = (1 − e− j ω̂ + e− j2ω̂ − e− j3ω̂ )(e− j ω̂ + e− j2ω̂ ) = e− j ω̂ − e− j5ω̂
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
Therefore, y[n] = x[n − 1] − x[n − 5]
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.14 DSP First 2e
4
1 − e− j5ω̂ sin(ω̂5/2)
(a) H (e j ω̂ ) = e− j ω̂n = = e− j ω̂2
X
n=0
1−e − j ω̂ sin(ω̂/2)
j ω̂
(b) H (e ) = 0 for ω̂ = ±0.4π, ±0.8π
(c) We need the frequency response at ω̂ = 0, 0.25π, and 0.4π.
H (e j0 ) = 5
)
eb
sin(5π/8)
er or in ing
ed id n
W
no the iss tea s
H (e j0.25π ) = e− j0.5π = 2.4142e− j0.5π
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
sin(π/8)
.
r
H (e j0.4π ) = 0
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
Therefore, y[n] = 35 + 8(2.4142) cos(0.25πn − π/2)
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
(d) Since the length of the impulse response is 5, the steady-state output is attained for n ≥ 4.
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.15 DSP First 2e
The sampled signal is
x[n] = x(n/ f s ) = 7 + 8 cos(1000πn/4000) + 9 cos(1600πn/4000)

= 7 + 8 cos(0.25πn) + 9 cos(0.4πn + 0.7π)
This is the same input signal as in Problem P-6.14 and the discrete-time system is the same as in that problem. Therefore,
from the solution to Problem P-6.14(c), the output sequence is
y[n] = 35 + 8(2.4142) cos(0.25πn − π/2)
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
The output of the D-to-C converter is therefore
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
y(t) = 35 + 8(2.4142) cos(1000πn − π/2)
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.16 DSP First 2e
(a) f s > 1000 Hz.

Since h[n] = δ[n], y[n] = x[n]. In other words, the system is a C-to-D converter followed by a D-to-C converter,
so the issue is sampling above the Nyquist rate. The highest frequency in the input signal is 500 Hz, so the Nyquist
rate is 2(500) = 1000 Hz.
(b) A delay of 10 samples must correspond to 0.001 s. After sampling x(t) at a rate of f s
x[n] = 10 + 20 cos(ω0 (n/ f s ) + π/3)
)
eb
y[n] = x[n − 10] = 10 + 20 cos(ω0 (n − 10)/ f s + π/3)
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
Reconstructing y(t) from y[n] with an ideal D-to-C converter gives
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
y(t) = 10 + 20 cos(ω0 ( f s t − 10)/ f s + π/3) = 10 + 20 cos(ω0 t − 10ω0 / f s + π/3)
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
We want y(t) = x(t − 0.001) = 10 + 20 cos(ω0 (t − 0.001) + π/3) which requires
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
−10ω0 / f s = ω0 (−0.001) f s = 10, 000 Hz

ro n an o te
⇒
st f a s d s ec
de o rse de ot
ill le u vi pr
In addition, the output frequency must be the same as the input frequency, so there must be no aliasing, i.e., the
w sa co ro is
or eir is p rk
o
relationship between ω0 and f s = 10, 000 Hz must obey the Nyquist condition.
w
an his
T
d
2ω0 < 2π f s ⇒ ω0 < 2π(5000) rad/s

th
(c) To get y(t) = A, we need y[n] = constant. Since x[n] = 10 + 20 cos(ω0 (n/ f s ) + π/3), the filter must null out the
cosine term; or the cosine term must alias to DC.
sin(5ω̂/2) − j ω̂2
Nulling: The zeros of H (e j ω̂ ) = e are at ω̂ = 2π/5, −2π/5, 4π/5, −4π/5.
5 sin(ω̂/2)
For ω0 and f s = 2000 Hz, these frequencies are 2π(±400) and 2π(±800) rad/s.
The negative frequencies must be used when we include aliases. The complete list of input frequencies that alias
to the zeros of H (e j ω̂ ) is
2π(400 + 2000`) = 2π(400), 2π(2400), 2π(4400), . . .

2π(800 + 2000`) = 2π(800), 2π(2800), 2π(4800), . . .
2π(−400 + 2000`) = 2π(−400), 2π(1600), 2π(3600), . . .
2π(−800 + 2000`) = 2π(−800), 2π(1200), 2π(3200), . . .
For these input frequencies, the cosine term will be filtered out, and only the DC remains. Since H (e j0 ) = 1 and
the DC value of x(t) = 10, the output will be y(t) = A = 10.
Cosine aliases to DC: When ω0 is a frequency that aliases to DC, then ω̂0 = (ω0 + 2π` f s )/ f s = 0,
and x[n] = 10 + 20 cos(0n + π/3) = 10 + 20 cos(π/3) = 20.
The frequencies for which this happens are
2π(0 + 2000`) = 2π(0), 2π(2000), 2π(4000), . . .
In this case, the DC value of x(t) = 20 and H (e j0 ) = 1, so the output will be y(t) = A = 20.

P-6.17 DSP First 2e
(a) The constant component passes through the first system as a constant. Then the first difference system gives a zero
output for a constant input. By superposition, the output is thus equal to y1 [n], the output due to the other input
component x 1 [n].
(b) The overall frequency response function is the product of the frequency responses. Therefore,
4
1 − e− j5ω̂
!
j ω̂ − j ω̂n + − j ω̂
(1 − e− j ω̂ ) = 1 − e− j5ω̂
X
H (e )= * e (1 − e )= − j ω̂
1 − e
)
eb
, n=0
er or in ing
-
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(c) The frequency response is zero at the 5th roots of unity or ω̂ = 2πk/5 for k = 0, 1, 2, 3, 4.
.
r
or ud a uc y
w cl le tr p
(d) From the frequency response, the overall difference equation is
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
y[n] = x[n] − x[n − 5]
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.18 DSP First 2e
(a) Need a plot of h[n] = −δ[n − 2] + δ[n − 12]
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
e j5ω̂ − e− j5ω̂
th k ( de f i es
(b) H (e j ω̂ ) = −e− j2ω̂ +e− j12ω̂ = e− j7ω̂ (−e j5ω̂ +e− j5ω̂ ) = e− j7ω̂ (−2 j) = j (−2 sin(ω̂5))e− j7ω̂ Therefore,
of or stu e o tat
2j
ity s w g us d S
is
R(e j ω̂ ) = −2 sin(5ω̂) and n0 = −7.
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
(c,d) Plots of |H (e j ω̂ )| and the principal value of ∠H (e j ω̂ ) for −π < ω̂ ≤ π.
in o e r
y y p d le d
ro n an o te
st f a s d s ec
2
de o rse de ot
|H(ej ω̂ )|
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
1
an his
T
th d
0
−π −π/2 0 π/2 π
5
H(ej ω̂ )
0
6
-5
−π −π/2 0 π/2 π
ω̂

P-6.19 DSP First 2e
(a)
H (e j ω̂ ) = H1 (e j ω̂ )H2 (e j ω̂ )
= (1 + 2e− j ω̂ + e− j ω̂2 )(1 − e− j ω̂ + e− j ω̂2 − e− j ω̂3 )
= 1 + e− j ω̂ − e− j ω̂4 − e− j ω̂5
(b) h[n] = δ[n] + δ[n − 1] − δ[n − 4] − δ[n − 5]
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
(c) y[n] = x[n] + x[n − 1] − x[n − 4] − x[n − 5]

T
th d

P-6.20 DSP First 2e
(a) We need to have zeros of the frequency response at ω̂ = 0 and 0.2π. Thus,
H (e j ω̂ ) = (1 − e− j ω̂ )(1 − e− j0.2π e− j ω̂ )(1 − e j0.2π e− j ω̂ )

= (1 − e− j ω̂ )(1 − 2 cos(0.2π)e− j ω̂ + e− j ω̂2 )
= 1 − 2.6180e− j ω̂ + 2.6180e− j ω̂2 − e− j ω̂3
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
y[n] = x[n] − 2.6180x[n − 1 + 2.6180x[n − 2] − x[n − 3]
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
(b) We need to null the frequencies ω̂ = 0, ±0.2π, ±0.4π, ±0.6π, ±0.8π, π. A system whose frequency response is zero
e in nt ns co
D
th k ( de f i es
at all these frequencies is H (e j ω̂ ) = 1 − e− j ω̂10 .
of or stu e o tat
ity s w g us d S
If we set H (e j ω̂ ) = 0 we get the equation 1 − e− j ω̂10 = 0 or e− j ω̂10 = 1 or, equivalently, e j ω̂10 = 1 = e j2πk where
is
te f t ss th nite
e rt ss fo U
k = 0, 1, . . . , 9. If we then equate the exponents of the last equation we get ω̂10 = 2πk, or ω̂ = 2πk/10 = 0.2πk.
gr hi in e
th a a ly by
k
in o e r
y y p d le d
The difference equation for this system is y[n] = x[n] − x[n − 10].
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.21 DSP First 2e
(a) The two normalized frequencies are 120π/ f s = 0.12π and 240π/ f s = 0.24π.
(b) We need to place zeros at ω̂ = ±0.12π and ±0.24π. Therefore, choose
H (e j ω̂ ) = (1 − e− j0.12π e− j ω̂ )(1 − e j0.12π e− j ω̂ )(1 − e− j0.24π e− j ω̂ )(1 − e j0.24π e− j ω̂ )

= 1 − 3.3175e− j ω̂ + 4.7111e− j ω̂2 − 3.3175e− j ω̂3 + e− j ω̂4
The corresponding difference equation is therefore:
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
y[n] = x[n] − 3.3175x[n − 1] + 4.7111x[n − 2] − 3.3175x[n − 3] + x[n − 4]
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
(c) Here is the Matlab plot of the frequency response (magnitude) versus ω̂, and also versus the analog frequency f
th k ( de f i es
in hertz. The frequencies 60 Hz and 120 Hz are marked and are snown to be nulled out.
of or stu e o tat
ity s w g us d S
Note: the plot of |H (e j ω̂ )| versus ω̂ is zoomed vertically so as to see the detail of the zeros.
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
|H(ej ω̂ )|
y y p d le d
ro n an o te
st f a s d s ec
1
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
0.5
o
w
an his
0
T
d
−π −π/2 0 π/2 π
th
|H(ej2πf /fs )|
normalized frequency ω̂
20
10
0
-500 -120-60 0 60 120 500
analog frequency f
Figure P-6.21

P-6.22 DSP First 2e
(a) To find y[n] we need the frequency response evaluated at frequencies 0 and 0.5π which are from the graph:
H (e j0 ) = 0 H (e j0.5π ) = 1.4e− jπ/4
It is easy to get these analytically as well since
H (e j ω̂ ) = 1 − e− j ω̂ = je− j ω̂/2 2 sin(ω̂/2) = 2 sin(ω̂/2)e j (π/2−ω̂/2)
Thus, the output is y[n] = 10 √1 cos(0.5πn − π/4) = 7.0711 cos(0.5πn − π/4).
)
eb
er or in ing
ed id n
W
no the iss tea s
2
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(b) The discontinuity of size π radians is due to the fact that H (e j ω̂ ) is zero at ω̂ = 0 so there is a sign change from
.
r
or ud a uc y
one side of 0 to the other. This is evident in H (e j ω̂ ) equation above since the function sin(ω̂/2) is negative for
w cl le tr p
e in nt ns co
D
th k ( de f i es
−π < ω̂ < 0 and positive for 0 < ω̂ < π.
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-6.23 DSP First 2e
(a) When the input is x[n] = 10 + 10 cos(0.2πn) + 10 cos(0.5πn), we need to determine the frequency response at
frequencies 0, 0.2π and 0.5π. From the curves we get
H (e j0 ) = 1 H (e j0.2π ) ≈ 0.9e− j0.4π H (e j0.5π ) = 0
Therefore, y[n] = 10 + 9 cos(0.2πn − 0.4π).

(b) There are two kinds of discontinuities in the phase. At ω̂ = 2π(0.17), the discontinuity of size 2π is due to taking
the principal value. At ω̂ = 0.5π the frequency response is zero and the discontinuity is a phase jump of ±π due to
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
the sign change as the frequency response crosses zero.
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 7. DTFT
P-7.1 DSP First 2e
This solution requires mostly table lookup and applying the delay property for the DTFT.
DTFT
(a) x 1 [n] = 2δ[n − 3] ←→ X1 (e j ω̂ ) = 2e− j ω̂3
DTFT
(b) x 2 [n] = 3δ[n − 2] − δ[n − 3] + 3δ[n − 4] ←→ X2 (e j ω̂ ) = 3e− j ω̂2 − e− j ω̂3 + 3e− j ω̂4
7
 1≤n≤8
(c) x 3 [n] = 7u[n − 1] − 7u[n − 9] = 
0 otherwise
Note how the step sequence is used to create a finite-length sequence. Plug into the DTFT definition and set

)
eb
er or in ing
ed id n
n = m + 1 in the sum. Then apply the formula for the sum of terms of a geometric series.
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
8 7
.
r
1 − e− j ω̂8 sin(4ω̂) − j ω̂9/2
or ud a uc y
w cl le tr p
X3 (e j ω̂ ) = 7e− j ω̂n = 7e− j ω̂ e− j ω̂m = 7e− j ω̂
X X
e in nt ns co
D
=7 e
1 − e− j ω̂ sin(ω̂/2)
th k ( de f i es
of or stu e o tat
n=1 m=0
ity s w g us d S
is
te f t ss th nite
sin(0.25πn) DTFT 1
|ω̂| < 0.25π
(
←→ X4 (e j ω̂ ) =
e rt ss fo U
(d) x 4 [n] = 9
gr hi in e
th a a ly by
9πn 0 0.25π < |ω̂| ≤ π

k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 7. DTFT
P-7.2 DSP First 2e
This solution requires table lookup and use of DTFT properties

DTFT
(a) Y1 (e j ω̂ ) = 2π ←→ y1 [n] = 2πδ[n]
DTFT
(b) Y2 (e j ω̂ ) = 5e− j3ω̂ ←→ y2 [n] = 5δ[n − 3]
DTFT
(c) Y3 (e j ω̂ ) = 6 cos(3ω̂) = 3e j3ω̂ + 3e− j3ω̂ ←→ y3 [n] = 3δ[n + 3] + 3δ[n − 3]
DTFT
(d) Y4 (e j ω̂ ) = j sin(7ω̂) = 21 e j7ω̂ − 12 e− j7ω̂ ←→ y4 [n] = 12 δ[n + 7] − 12 δ[n − 7]
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 7. DTFT
P-7.3 DSP First 2e
This solution involves either evaluating the definitions of the direct and inverse DTFT or using table lookup and the
properties of the DTFT.
0.3π
1 e j0.3πn − e− j0.3πn sin(0.3πn)
Z
(a) v1 [n] = e j ω̂n d ω̂ = =
2π j2πn πn
−0.3π
9
1 − e− j ω̂10 e− j ω̂5 (e j ω̂5 − e− j ω̂5 ) sin(ω̂5) − j ω̂4.5
(b) V2 (e j ω̂ ) = e− j ω̂n =
X
= = e
1−e − j ω̂ e − j ω̂/2 (e j ω̂/2 −e − j ω̂/2 ) sin(ω̂/2)
)
eb
er or in ing
ed id n
n=0
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
sin(0.3πn)
(c) Note that V3 (e j ω̂ ) = 1 − V1 (e j ω̂ ). Thus, v3 [n] = δ[n] − v1 [n] = δ[n] −
.
r
πn
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
Alternatively, you can plug into the inverse DTFT integral as in part (a) obtaining
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
−0.3π Zπ
e rt ss fo U
1 1 sin(πn) sin(0.3πn) sin(0.3πn)

Z
gr hi in e
j ω̂n
e j ω̂n d ω̂ =
th a a ly by
v3 [n] = d ω̂ + = δ[n] −
k
e −
in o e r
y y p d le d
2π 2π πn πn πn
ro n an o te
st f a s d s ec
−π 0.3π
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
9 9
sin((ω̂ − π)5) − j (ω̂−π)4.5
o
w
(d) V4 (e j ω̂ ) = e jπn e− j ω̂n = e− j (ω̂−π)n = V2 (e j (ω̂−π) ) =

X X
e
an his
sin((ω̂ − π)/2)
T
d
n=0 n=0
th

CHAPTER 7. DTFT
P-7.4 DSP First 2e
This solution requires using the linearity and delay properties of the DTFT and the following basic DTFT pair:
sin(ω̂0 n) DTFT 1 |ω̂| < ω̂0

(
j ω̂
h[n] = ←→ H (e ) =
πn 0 ω̂ 0 < | ω̂| ≤ π
It might be helpful to sketch the different bandlimited DTFTs that are involved.
sin(0.25πn) DTFT 1 |ω̂| < 0.25π 0 |ω̂| < 0.25π
( (
j ω̂
(a) h1 [n] = 5δ[n] − ←→ H1 (e ) = 5 − 5 =
0.2πn 0 0.25π < |ω̂| ≤ π 5 0.25π < |ω̂| ≤ π
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
(b) h2 [n] =
m ld a
−
0.1πn 0.1πn
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
0 0 ≤ |ω̂| ≤ 0.1π
th k ( de f i es

1 0 ≤ |ω̂| ≤ 0.4π 1 0 ≤ |ω̂| ≤ 0.1π
of or stu e o tat

DTFT

←→ H2 (e j ω̂ ) = 10 
  
− 10  = 10  1 0.1π < |ω̂| ≤ 0.4π
ity s w g us d S


0 0.4π < |ω̂| ≤ π 0 0.1π < |ω̂| ≤ π
is
te f t ss th nite
 0 0.4π < |ω̂| ≤ π

e rt ss fo U

gr hi in e
 
th a a ly by
k

in o e r
y y p d le d
ro n an o te
st f a s d s ec
sin(0.4π(n − 8)) sin(0.1π(n − 8))

de o rse de ot
(c) h3 [n] = − = 0.1h2 [n − 8]

ill le u vi pr
π(n − 8) π(n − 8)
w sa co ro is
or eir is p rk
o
w
an his
 0 0 ≤ |ω̂| ≤ 0.1π
T
d

DTFT

j ω̂ j ω̂ − j ω̂8 − j ω̂8

←→ H3 (e ) = 0.1H2 (e )e =e  1 0.1π < |ω̂| ≤ 0.4π

th

 0 0.4π < |ω̂| ≤ π




CHAPTER 7. DTFT
P-7.5 DSP First 2e

6
Write down the system function directly from the difference equation obtaining H (e j ω̂ ) = 3 e− j ω̂k .
X
k=2
With a change of index in the sum we can apply the formula for the sum of terms in a geometric series.
sin(ω̂5/2) − j ω̂4
H (e j ω̂ ) = 3 e
sin(ω̂/2)
Thus, the answers as requested are: L = 5, α = 3, and β = 4.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 7. DTFT
P-7.6 DSP First 2e
The exponential factors suggest a delay, but we cannot use the DTFT delay property directly because the delay would be
a fraction of a sample. This means that we should use the inverse DTFT if we can evaluate it, and, of course, we can.
(a) Inverse DTFT of G1 (e j ω̂ )
0.4π 0.4π
1 1 e j0.4π(n−0.3) − e− j0.4π(n−0.3)
Z Z
DTFT − j0.3ω̂ j ω̂n
←→ g1 [n] = 5e e d ω̂ = 5eω̂(n−0.3) d ω̂ = 5
2π 2π 2π(n − 0.3)
−0.4π −0.4π
)
eb
er or in ing
sin 0.4π(n − 0.3)
ed id n
W
no the iss tea s
itt W tio
w
so g1 [n] = 5 .
t p W em ch
e
d on g. in t la
m ld a
π(n − 0.3)
.
r
This looks like it was obtained using the delay property, but remember that the delay property only holds in general
or ud a uc y
w cl le tr p
e in nt ns co
D
for integer shifts.
th k ( de f i es
of or stu e o tat
ity s w g us d S
(b) Using the approach of (a), the solution is
is
te f t ss th nite
e rt ss fo U
gr hi in e
sin π(n − 0.1) sin 0.7π(n − 0.1)

th a a ly by
k
g2 [n] = 4 −4
in o e r
y y p d le d
ro n an o te
π(n − 0.1) π(n − 0.1)

st f a s d s ec
de o rse de ot
ill le u vi pr
If the delay had been an integer, the first function would reduce to an impulse, but in this case it does not.
w sa co ro is
or eir is p rk
o
w
(c) Again using the direct evaluation of the inverse DTFT, the solution is
an his
T
d
sin 0.7π(n − 0.2) sin 0.4π(n − 0.2)

th
g3 [n] = 9 −9
π(n − 0.2) π(n − 0.2)

CHAPTER 7. DTFT
P-7.7 DSP First 2e
sin(0.25πn) sin(0.44πn) sin(0.25πn)

(a) ∗ =
9πn 4πn 36πn
The DTFT of this signal is the product of two ideal lowpass filter frequency responses. Therefore the result
of the convolution is a sinc function corresponding to the lowest cutoff frequency and scaled by the product of the
gains.
sin(0.25πn) sin(0.25πn) sin(0.25πn) sin(0.25πn)
(b) ∗ ∗ =
9πn 8πn 7πn (9 × 8 × 7)πn
)
eb
er or in ing
ed id n
W
no the iss tea s
In this case we have the convolution of three sinc functions all with the same cutoff frequency so the result is
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
the same signal but with the product of the gains.
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
(c) Considerable simplification is possible.
th k ( de f i es
of or stu e o tat
ity s w g us d S
(0.3) n u[n] ∗ (10δ[n − 2] − 3δ[n − 3])
is
te f t ss th nite
e rt ss fo U
= 10(0.3) n−2 u[n − 2] − 3(0.3) n−3 u[n − 3]

gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
| {z }
break into 2 terms
st f a s d s ec
de o rse de ot
ill le u vi pr
= 10δ[n − 2] + 10(0.3)(0.3) n−3 u[n − 3] − 3(0.3) n−3 u[n − 3]

w sa co ro is
or eir is p rk
= 10δ[n − 2]
o
w
an his
T
th d

CHAPTER 7. DTFT
P-7.8 DSP First 2e
We will need the impulse response of the system to do parts (a) and (c). We can get it by expanding H (e j ω̂ ) using the
inverse Euler relation; i.e., H (e j ω̂ ) = 12 e j ω̂3 + 21 e− j ω̂3 , so h[n] = 12 δ[n + 3] + 12 δ[n − 3].
(a) By definition, y1 [n] = h[n − 3] = 21 δ[n] + 21 δ[n − 6].

(b) Here we can use the frequency response directly if we express x 2 [n] as x 2 [n] = 1.5e j0.25πn + 1.5e− j0.25πn . Then
y2 [n] = 1.5H (e j0.25π )e j0.25πn + 1.5H (e− j0.25π )e− j0.25πn
)
eb
er or in ing
√
ed id n
W
no the iss tea s
itt W tio
w
= 1.5 cos(0.75π)e j0.25πn + 1.5 cos(−0.75π)e− j0.25πn = −1.5 2 cos(0.25πn)
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
1 0 ≤ n ≤ 8

th k ( de f i es
(c) This signal is a length-9 pulse: x 3 [n] = u[n] − u[n − 9] = 
of or stu e o tat
 0 otherwise
ity s w g us d S
is
te f t ss th nite

This one should be done by convolution because the impulse response is so simple and a frequency-domain solution
e rt ss fo U
gr hi in e
th a a ly by
k
would not be feasible.
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
y3 [n] = x 3 [n] ∗ h[n] = x 3 [n] ∗ 12 δ[n + 3] + 12 δ[n − 3]

ill le u vi pr
w sa co ro is
or eir is p rk
1
o
−3 ≤ n ≤ 2
w
2
an his



1 3≤n≤5
T
d


= 21 (u[n + 3] − u[n − 6] + u[n − 3] − u[n − 12]) = 

th
1



 2 6 ≤ n ≤ 11
0 otherwise



CHAPTER 7. DTFT
P-7.9 DSP First 2e
(a) Expand cosine with the inverse Euler relation:
e j ω̂0 n + e− j ω̂0 n
!
x[n] = a n cos(ω̂0 n)u[n] = a n u[n] = 12 e j ω̂0 n a n u[n] + 21 e− j ω̂0 n a n u[n]
2
DTFT 1
(b) Using the DTFT pair a n u[n] ←→ and the frequency-shift property of the DTFT we can write the DTFT
1 − ae− j ω̂
of x[n] as
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
1 1
e
d on g. in t la
m ld a
2 2
X (e j ω̂ ) = +

1− ae− j (ω̂−ω̂0 ) 1− ae− j (ω̂+ω̂0 )
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
1
− 12 ae− j ω̂0 e− j ω̂ + 1
− 12 ae j ω̂0 e− j ω̂
of or stu e o tat
2 2
ity s w g us d S
=
is
te f t ss th nite
(1 − ae j ω̂0 e− j ω̂ )(1 − ae− j ω̂0 e− j ω̂ )
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
1 − 12 a(e j ω̂0 + e− j ω̂0 )e− j ω̂

ro n an o te
st f a s d s ec
=
de o rse de ot
1 − a(e j ω̂0 + e− j ω̂0 )e− j ω̂ + a2 e− j2ω̂

ill le u vi pr
w sa co ro is
or eir is p rk
o
1 − a cos(ω̂0 )e− j ω̂
w
an his
=
T
d
1 − 2a cos(ω̂0 )e− j ω̂ + a2 e− j ω̂2

th

CHAPTER 7. DTFT
P-7.10 DSP First 2e
(a) The plot of |V (e j ω̂ )| 2 for −π < ω̂ ≤ π is below.

1 1
From the table, V (e j ω̂ ) = so |V (e j ω̂ )| 2 = .
1 − ae− j ω̂ 1 + a2 − 2a cos ω̂
1
A sketch of this for a = 0.95 would show a maximum at ω̂ = 0 of size = 400 and it would fall off
(1 − a) 2
1
monotonically with increasing ω̂ to a value of = 0.2630 at ω̂ = ±π.
(1 + a) 2
)
eb
(b) If V (e j ω̂ ) were a frequency response, is the filter lowpass, highpass or bandpass? LOWPASS
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(c) Using the frequency shift property of the DTFT we obtain W (e j ω̂ ) = 12 V (e j (ω̂−0.3π) ) + 12 V (e j (ω̂+0.3π) ). Therefore,
.
r
or ud a uc y
a plot of |W (e− j ω̂ )| 2 will look like two shifted versions of the plot in part (a) with everything scaled by 14 . This will
w cl le tr p
e in nt ns co
D
th k ( de f i es
place peaks at approximately ω̂ = ±0.3π with heights of approximately 100.
of or stu e o tat
ity s w g us d S
Note: the plots use a semilog scale to show details.
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d
(d) If W (e j ω̂ ) were a frequency response, is the filter lowpass, highpass or bandpass? BANDPASS

CHAPTER 7. DTFT
P-7.11 DSP First 2e
(a) Using the definition of the ideal lowpass filter we can write
 1 |ω̂| ≤ ω̂co  1 |ω̂| ≤ ω̂co  0 |ω̂| ≤ ω̂co

Hhp (e j ω̂ ) = 
  
 1 ω̂co < |ω̂| ≤ π −  0 ω̂co < |ω̂| ≤ π =  1 ω̂co < |ω̂| ≤ π
 
  
which is the frequency response of an ideal highpass filter.
Note: A plot of the function 1 − Hlp (e j ω̂ ) might make this easier to visualize.
)
eb
(b) The cutoff frequency of the ideal HPF is ω̂co .
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(c) From part (a) it follows that hhp [n] = δ[n] − hlp [n] so
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
sin(ω̂co n)
th k ( de f i es
hhp [n] = δ[n] −
of or stu e o tat
πn
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 7. DTFT
P-7.12 DSP First 2e
(a) To get a bandpass filter, we need to subtract the low frequency band from a wider lowpass filter. Thus, we can write
 0 |ω̂| < 0.3π

1 |ω̂| ≤ 0.6π 1 |ω̂| ≤ 0.3π


j ω̂
  
Hbp (e )= = 1 0.3π ≤ |ω̂| ≤ 0.6π

0 − 
0.6π < |ω̂| ≤ π 
0 0.3π < |ω̂| ≤ π 0 0.6π < |ω̂| ≤ π
 

Therefore, the cutoff frequencies of the two lowpass filters are
)
eb
ω̂lp1 = ω̂co2 = 0.6π, and ω̂lp2 = ω̂co1 = 0.3π.
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a

(b) hbp [n] = −
.
r
πn πn
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 7. DTFT
P-7.13 DSP First 2e
(a) A sketch would show a negative exponential decaying to the left for n ≤ −1.
−1
X
(b) The DTFT exists for values of b such that |x[n]| is absolutely summable; i.e., | − bn | < ∞.
−∞
This will be true for |b| > 1.
(c) By definition, the DTFT of x[n] = −bn u[−n − 1] when |b| > 1 is
)
−1 ∞ ∞
eb
er or in ing
ed id n
X (e j ω̂ ) = −bn e− j ω̂n = − b−n e j ω̂n = 1 − b−n e j ω̂n
W
X X X
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
n=−∞ n=1 n=0
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
1 −b−1 e j ω̂
th k ( de f i es
=1− =
of or stu e o tat
1 − b−1 e j ω̂ 1 − b−1 e j ω̂
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
−b−1 e j ω̂
gr hi in e
1
th a a ly by
k
= =
in o e r
y y p d le d
ro n an o te
−b−1 e j ω̂ (1 − be− j ω̂ ) 1 − be− j ω̂

st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 7. DTFT
P-7.14 DSP First 2e
0.25π
∞ 2
sin(0.25πn) 1 1 1
X Z
(a) = d ω̂ =
3πn 2π 3 12

n=−∞
−0.25π
 1/3
sin(0.25πn) DTFT  0 ≤ |ω̂| ≤ 0.25π
since ←→ 
3πn 0 0.25π < |ω̂| ≤ π

Zπ 2
(b) 4 sin(ω̂5) e− j ω̂9/2 d ω̂= 20
)
eb
er or in ing
sin(ω̂/2) π
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch

e
d on g. in t la
m ld a
−π

9
.
r
sin(ω̂5) − j ω̂9/2
or ud a uc y
DTFT
w cl le tr p
X
since 2δ[n − k] ←→ 2 e
e in nt ns co
D
sin(ω̂/2)
th k ( de f i es
of or stu e o tat
k=0
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 7. DTFT
P-7.15 DSP First 2e
(a) This one is easy with the DTFT since X1 (e j ω̂ ) = e− j ω̂4 and Cx1 x1 (e j ω̂ ) = |X1 (e j ω̂ )| 2 = 1. Therefore, cx1 x1 [n] =
δ[n]. In the index domain, it is also easy: cx1 x1 [n] = x 1 [n] ∗ x 1 [−n] = δ[n − 4] ∗ δ[−n − 4] = δ[n]
(b) This one can be done easily by discrete convolution, but let’s use the DTFT. Since X2 (e j ω̂ ) = e− j ω̂4 − e− j ω̂6 , it
follows that
Cx2 x2 (e j ω̂ ) = |X2 (e j ω̂ )| 2 = (e− j ω̂4 − e− j ω̂6 )(e j ω̂4 − e j ω̂6 ) = −e j ω̂2 + 2 − e− j ω̂2
so cx2 x2 [n] = −δ[n + 2] + δ[n] − δ[n − 2].
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(c) Let’s try this with the DTFT. First note that x 3 [n] is identical to the impulse response of a 10-point running sum

system with additional delay of 6 samples. Therefore,
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
!2
sin(ω̂5) − j ω̂4.5 − j ω̂6 sin(ω̂5)
th k ( de f i es
X3 (e j ω̂ ) = e e and Cx3 x3 (e j ω̂ ) = |X3 (e j ω̂ )| 2 = .
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
(Note that the exponential factors corresponding to delay disappear when we take the magnitude.)
e rt ss fo U
gr hi in e
th a a ly by
k
To determine cx3 x3 [n] we have to determine the inverse DTFT. We cannot evaluate the integral with what we know
in o e r
y y p d le d
ro n an o te
and we have no entry in our table of transforms for this function. Therefore, we should try convolution, which is
st f a s d s ec
de o rse de ot
ill le u vi pr
not too difficult. The result is

w sa co ro is
or eir is p rk
o
w
 10 − |n| −9 ≤ n ≤ 9
an his

cx3 x3 [n] = 
T
d
0 otherwise
th

This might be a good transform pair to know in a general case. In particular, you can show the following for any
sequence x[n] (including an autocorrelation function):
!2
 L − |n|
 |n| ≤ (L − 1) DTFT j ω̂ sin(ω̂L/2)
x[n] =  ←→ X (e )=
0 otherwise sin(ω̂/2)

sin(0.25πn) DTFT  1/9 |ω̂| < 0.25π

←→ X4 (e j ω̂ ) = 

(d) Recall the DTFT pair: x 4 [n] =
9πn 0 0.25π < |ω̂| ≤ π

This one is a clear candidate for using the DTFT since it would be impossible to evaluate the discrete-time
autocorrelation in the time domain.
We should recognize that x 4 [n] is identical to the impulse response of an ideal lowpass filter with cutoff frequency
0.25π and gain 1/9. If we take the squared-magnitude of X4 (e j ω̂ ), we get the same cutoff but the gain is squared.
sin(0.25πn)
Therefore, it follows that cx4 x4 [n] =
81πn

CHAPTER 7. DTFT
P-7.16 DSP First 2e
(a) This is accomplished by evaluating the discrete-time convolution cx x [n] = x[n] ∗ x[−n] and the result is cx x [n] =
 5 − |n| |n| ≤ 4 .


0 otherwise

(b) To do this, we don’t try to find the DTFT of cx x [n] in part (a) directly, but instead use the fact that Cx x (e j ω̂ ) =
|X (e j ω̂ )| 2 . We can look up the DTFT of x[n] in our table of DTFT pairs with L = 5. Thus
sin(ω̂5/2) 2 sin(ω̂5/2)
!2
)
j ω̂ − j ω̂(L−1)/2
eb
Cx x (e ) = e =
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a

.
r
Cx x (e j ω̂ ) is clearly real and positive because it is a squared-magnitude of the complex DTFT X (e j ω̂ ).
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
(c) Since Cy y (e j ω̂ ) = |Y (e j ω̂ )| 2 = |X (e j ω̂ )e− j ω̂2 | 2 = |X (e j ω̂ )| 2 , it follows that cy y [n] = cx x [n]. In fact the auto
of or stu e o tat
ity s w g us d S
correlation function for a delayed signal is always the same as the autocorrelation function for the signal itself.
is
te f t ss th nite
Delay always cancels out when you evaluate a correlation.
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 7. DTFT
P-7.17 DSP First 2e
(a) Plot shows details of the passband and stopband magnitudes.

1.5
)
eb
er or in ing
ed id n
W
0.5
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
0
w cl le tr p
e in nt ns co
D
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
th k ( de f i es
of or stu e o tat
ity s w g us d S
#10 -3
is
te f t ss th nite
6
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
4
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
2
w
an his
T
d
0
th
0 0.1 0.2 0.3 0.4 0.5 0.6

1.004
1.002
1
0.998
0.996
0.994
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
!/:
Figure 7-1: Solution for Problem 7-17(a).
(b) From the plots the measured parameters are ω̂ p = 0.7034π, ω̂ s = 0.4932π, δ p = 0.003 and δ s = 0.0022.
(c) A good rule of thumb is that with a Hamming window, the transition region satisfies ∆ω̂ M = 8π/M, where
M is the order of the filter. For M = 32, this formula gives ∆ω̂32 = 0.25π, whereas the measured value is
∆ω̂32 = (0.7034 − 0.4932)π = 0.2102π, so the formula over-estimates the transition width in this case.
(d) However, the formula gives us a rationale for estimating the effect of going from M = 32 to M = 80. We should
have ∆ω̂80 ≈ ∆ω̂32 32/80 = 0.2102π(32/80) = .0841π. The formula would give a value for ∆ω̂80 of 0.1π. The
measured cutoff frequencies for M = 80 are ω̂ p = (.6 + .0841/2)π = 0.642π and ω̂ s = (0.6 − .0841/2)π = 0.558π.

CHAPTER 7. DTFT
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
1
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
0.8
th k ( de f i es
of or stu e o tat
0.6
ity s w g us d S
is
te f t ss th nite
0.4
e rt ss fo U
gr hi in e
th a a ly by
k
0.2
in o e r
y y p d le d
ro n an o te
0
st f a s d s ec
de o rse de ot
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

ill le u vi pr
w sa co ro is
or eir is p rk

o
#10 -3
w
5
an his
T
d
4
th
3
2
1
0
0 0.1 0.2 0.3 0.4 0.5 0.6
1.005
1
0.995
0.99
0.985
0.98
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
!/:
Figure 7-2: Solution for Problem 7-17d. The estimates of the cutoff frequencies from part (c) are seen to be quite
accurate.

CHAPTER 7. DTFT
P-7.18 DSP First 2e

To
solve this problem use Matlab to design the filter using filterdesign. Save the impulse response to the work space
as h24 for use. The solution to each of the following parts requires that we evaluate the frequency response at a given
frequency. The frequencies are 0, 0.5π and 0.75π. This can be done in many ways, but the most straightforward way is to
use Matlab’s freqz( ) function as follows:
>> H24 = freqz(h24,1,[0,0.5*pi,0.75*pi])

H24 =
Columns 1 through 2
)
eb
0.000134504777373 + 0.000000000000000i 0.041265977941552 + 0.000000000000000i
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
Column 3
e
d on g. in t la
m ld a
-0.996401749465568 - 0.000000000000001i
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
We will need the magnitude and angle at these three frequencies so we do the following:
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
>> zprint(H24)
e rt ss fo U
gr hi in e
Z = X + jY Magnitude Phase Ph/pi Ph(deg)
th a a ly by
k
in o e r
y y p d le d
ro n an o te
0.0001345 0 0.0001345 0.000 0.000 0.00

st f a s d s ec
de o rse de ot
0.04127 7.164e-17 0.04127 0.000 0.000 0.00

ill le u vi pr
w sa co ro is
-0.9964 -1.142e-15 0.9964 -3.142 -1.000 -180.00

or eir is p rk
o
w
an his
(a) If the input is x 1 [n] = 10 for all n, the output is y1 [n] = 10H24 (e j0 ), so we use Matlab as above to find
T
d
H24 (e j0 ).H32=sum(h32)=0.0001345, so y1 [n] = 0.001345. This is a highpass filter so the DC is greatly at-
th
tenuated.
(b) When the input is a zero-phase cosine, the output is y2 [n] = 10|H24 (e j0.5π )| cos(0.5πn + ∠H24 (e j0.5π )), so from
the Matlab analysis, y2 [n] = 0.4127 cos(0.5πn + 0). This frequency is in the transition zone near the stopband so
its amplitude is much reduced.
(c) As in part (b) the output in this case is y3 [n] = 10|H24 (e j0.75π )| cos(0.75πn + ∠H24 (e j0.75π )), so from the Matlab
analysis, y3 [n] = 9.964 cos(0.75πn − π). This frequency is in the passband so the amplitude is barely reduced.

CHAPTER 7. DTFT
P-7.19 DSP First 2e

A
Matlab program to make the requested plots is as follows:
% solution to problem 7-19

% assumes that h37 has been saved in workspace by filterdesign GUI
% h37 = fir1(37,0.3,ones(1,38);
n=0:99;
x1=10*ones(1,100);
x2=10*cos(0.5*pi*n);
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
sn=x1+x2;
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
snOutput=filter(h37,1,sn);
th k ( de f i es
%
of or stu e o tat
ity s w g us d S
subplot(411)
is
te f t ss th nite
stem(n,y1,’filled’);hold on; plot([37,37],[-9,14.5],’r--’);hold off
e rt ss fo U
gr hi in e
th a a ly by
k
title(’Output y_1[n] due to x_1[n] = 10’)
in o e r
y y p d le d
ro n an o te
st f a s d s ec
text(20,1,5,’Transient region’);hold off

de o rse de ot
ill le u vi pr
subplot(412)
w sa co ro is
or eir is p rk
stem(n,y2,’filled’);hold on; plot([37,37],[-10,15],’r--’);hold off

o
w
an his
title(’Output y_2[n] due to x_2[n] = 10cos(0.5 \pi n)’)

T
d
subplot(413)
th
stem(n,y1+y2,’filled’);hold on; plot([37,37],[-10,15],’r--’);hold off

title(’Sum of Outputs’)
subplot(414)
stem(n,snOutput,’filled’)
title(’Output Due to Sum of Inputs’)
xlabel(’time index n’)
hold on
han=stem(n,snOutput-y1-y2);hold on; plot([37,37],[-10,15],’r--’);hold off
The resulting plot is shown in the figure above.
(a) When the input x 1 [n] is DC with a DC level of 10, the output level y1 [n] converges to 10H (e j0 ) = 10(1) = 10 after
a brief transient of less than 37 samples.
(b) When the input is the sinusoid x 2 [n] = 10 cos(0.5πn) we must evaluate H (e j ω̂ ) at ω̂ = 0.5π rad. We obtain
H (e j0.5π ) = 0.02814e j0.75π , so the output would be y2 [n] = 0.2814 cos(0.5πn + 0.75π).
(c) Note that the bottom two figures are identical, thereby showing that superposition holds. This is also shown by the
open circles, which show the difference between y1+y2 and snoutput.

CHAPTER 7. DTFT
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
Output y [n] due to x [n] = 10
t p W em ch
e
d on g. in t la
m ld a
1 1
.
r
or ud a uc y
w cl le tr p
10
e in nt ns co
D
th k ( de f i es
of or stu e o tat
0
ity s w g us d S
Transient region is n=0 to 37
is
te f t ss th nite
−10
e rt ss fo U
gr hi in e
th a a ly by
k
0 10 20 30 40 50 60 70 80 90 100
in o e r
y y p d le d
ro n an o te
Output y [n] due to x [n] = 10cos(0.5 π n)

st f a s d s ec
de o rse de ot
2 2
ill le u vi pr
w sa co ro is
2
or eir is p rk
o
w
1
an his
T
d
0
th
−1
0 10 20 30 40 50 60 70 80 90 100
Sum of Outputs
15
10
5
0
0 10 20 30 40 50 60 70 80 90 100
Output Due to Sum of Inputs
15
10
5
0
0 10 20 30 40 50 60 70 80 90 100
Time Index (n)

CHAPTER 7. DTFT
P-7.20 DSP First 2e
(a) Use the GUI to find this frequency to be approximately 0.712π rad. The unwrapped phase at this frequency is
ϕ ≈ −10.5 rad. The principal value of the phase is −10.5 + 2π ≈ 2.01 rad.
(b) Here again, you can use the GUI to find the frequencies where |H (e j ω̂ )| = 1.
These are approximately {0.71π, 0.788π, 0.854π, 0.914π, 0.972π}
(c) Use the GUI to find the frequencies where |H (e j ω̂ )| = 0.
These are approximately {0.03π, 0.091π, 0.151π, 0.212π, 0.272π, 0.334π, 0.398π, 0.486π}.
)
eb
er or in ing
The following figure shows the places where the frequency response is one and zero more accurately than is possible
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
with the GUI. This plot can be constructed by saving the impulse response to the work space and then using freqz

to compute the frequency response.
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
1.5
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
1
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
0.5
an his
T
th d
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
#10 -3
6
0
0 0.1 0.2 0.3 0.4 0.5 0.6
1.002
1
0.998
0.996
0.994
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
!/:

CHAPTER 8. DFT
P-8.1 DSP First 2e
(a) The DTFT of x 0 [n] is X0 (e j ω̂ ) = 1 for all ω̂. Therefore, X0 [k] = X0 (e j (2πk /10) ) = 1 for k = 0, 1, . . . , 9. In Matlab
>> X0=fft([1,zeros(1,9)])
X0 = 1 1 1 1 1 1 1 1 1 1
9
1 − e− j ω̂10
(b) The DTFT of x 1 [n] is X1 (e j ω̂ ) = e− j ω̂n =
X
. Therefore,
1 − e− j ω̂
)
eb
n=0
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
1 − e− j2πk  10 k=0

X1 [k] = X1 (e j (2π/10)k ) =

= k = 0, 1, . . . , 9

.
r
or ud a uc y
1−e (2π/10)k 0 k = 1, 2, . . . , 9
w cl le tr p
− j
e in nt ns co
D
th k ( de f i es

of or stu e o tat
ity s w g us d S
1 − e− j2πk
Note that the closed form X1 [k] = is indeterminant when we try to evaluate for k = 0. L’ Hospital’s
is
te f t ss th nite
1 − e− j (2π/10)k
e rt ss fo U
gr hi in e
rule doesn’t work here because k is an integer variable. Therefore we need to go back to the sum form where we
th a a ly by
k
in o e r
y y p d le d
9
ro n an o te
st f a s d s ec
X
see that X1 [0] = X1 (e j (2π/10)0 ) = 1 = 10.
de o rse de ot
ill le u vi pr
w sa co ro is
n=0
or eir is p rk
o
w
an his
T
d
>> X1 = fft(ones(1,10))
th
X1 = 10 0 0 0 0 0 0 0 0 0
(c) The DTFT of x 2 [n] is X2 (e j ω̂ ) = e− j ω̂4 so X2 [k] = e− j (2π/10)k4 k = 0, 1, . . . , 9.

>> X2 = fft([zeros(1,4),1,zeros(1,5)])
X2 = 1.0000 + 0.0000i -0.8090 - 0.5878i 0.3090 + 0.9511i
0.3090 - 0.9511i -0.8090 + 0.5878i 1.0000 + 0.0000i
-0.8090 - 0.5878i 0.3090 + 0.9511i 0.3090 - 0.9511i
-0.8090 + 0.5878i
9 9
1 − e− j (ω̂−2π/5)10
(d) The DTFT of x 3 [n] is X3 (e j ω̂ ) = e j2πn/5 e− j ω̂n = e− j (ω̂−2π/5)n =
X X
.
n=0 n=0
1 − e− j (ω̂−2π/5)
Therefore,
 0 k = 0, 1
1 − e− j ((2π/10)k−2π/5)10 1 − e− j2π(k−2)


j (2π/10)k

X3 [k] = X3 (e )= = =  10 k=2


1 − e− j ((2π/10)k−2π/5) 1 − e− j (2π/10)(k−2) 0 k = 3, 4, . . . , 9


As in part (b), we have to be careful in evaluating this expression for k = 2.
>> X3 = fft(exp(j*2*pi*(0:9)/5))
X3 = -0.0000 + 0.0000i -0.0000 + 0.0000i 10.0000 - 0.0000i
0.0000 + 0.0000i -0.0000 + 0.0000i 0.0000 + 0.0000i
-0.0000 + 0.0000i 0.0000 + 0.0000i 0.0000 + 0.0000i
-0.0000 + 0.0000i

CHAPTER 8. DFT
P-8.2 DSP First 2e
(a) Substituting into the inverse DFT definition we get,

9
1 X
x a [n] = δ[k]e j (2π/10)k n = 0.1 n = 0, 1, . . . , 9
10 k=0
>> xa = ifft([1,zeros(1,9)])
xa = 0.1000 0.1000 0.1000 0.1000 0.1000 0.1000
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
0.1000 0.1000 0.1000 0.1000
t p W em ch
e
d on g. in t la
m ld a
(b) This one is the first entry in Table 8-1, so we can write the answer down by inspection: x b [n] = δ[n]. Alternatively,
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
you can plug into the inverse DFT definition and get
th k ( de f i es
of or stu e o tat
9
ity s w g us d S
1 X j (2π/10)k n 1 − e j (2π/10)10n 
1 n=0
x b [n] = e = = = δ[n]
is
te f t ss th nite
10 k=0 1 − e j (2π/10)n 0 n = 1, 2, . . . , 9
e rt ss fo U
gr hi in e
th a a ly by
k

in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
>> xb = ifft(ones(1,10))
or eir is p rk
o
xb = 1 0 0 0 0 0 0 0 0 0
w
an his
T
d
(c) Since there are only two nonzero terms, we can write down the answer directly and manipulate it if possible.
th
x c [n] = 0.1(e j (2π/10)3n + e j (2π/10)7n )

= 0.1(e j (2π/10)3n + e j (2π/10)(10−3)n )
= 0.1(e j (2π/10)3n + e j (2π/10)10n e− j (2π/10)3n )
= 0.1(e j (2π/10)3n + e− j (2π/10)3n )
= 0.2 cos((2π/10)3n)
>> xc = ifft([0,0,0,1,0,0,0,1,0,0])
xc = 0.2000 -0.0618 -0.1618 0.1618 0.0618 -0.2000
0.0618 0.1618 -0.1618 -0.0618
(d) We can observe that
X d [k] = cos(2πk/5)X b [k]
= 21 e j (2π/10)2k X b [k] + 21 e− j (2π/10)2k X b [k]
= 12 e j (2π/10)2k X b [k] + 21 e j (2π/10)(10−2)k X b [k]
= 21 e j (2π/10)2k X b [k] + 21 e j (2π/10)8k X b [k]
so by the time-shift property of Table 8-2,
x d [n] = 21 x b [n − 2] + 21 x b [n − 8] = 12 δ[n − 2] + 12 δ[n − 8]
>> xd = ifft(cos(2*pi*(0:9)/5))
xd = -0.0000 + 0.0000i -0.0000 + 0.0000i 0.5000 + 0.0000i
0.0000 - 0.0000i -0.0000 - 0.0000i 0.0000 + 0.0000i
-0.0000 + 0.0000i 0.0000 + 0.0000i 0.5000 - 0.0000i
-0.0000 - 0.0000i

CHAPTER 8. DFT
P-8.3 DSP First 2e
(a) Express (−1) n as e jπn = e j (2π/12)6n and substitute into the DFT definition.
11 11
X X 1 − e− j2π(k−6)
Y0 [k] = 3 e j (2π/12)6n e− j (2π/12)k n = 3 e− j (2π/12)(k−6)n = 3
n=0 n=0
1 − e− j (2π/12)(k−6)



 0 k = 0, 1, . . . , 5
= 36 k=6


)
0

k = 7, 8, . . . , 11
eb

er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch

e
d on g. in t la
m ld a
Note that the numerator of the closed form is zero for all integer k, but when k = 6 we have an indeterminant form
.
r
or ud a uc y
that evaluates to 3 × 12 = 36.
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
>> Y0 = 3*fft((-1).^(0:11))
e rt ss fo U
gr hi in e
th a a ly by
k
Y0 = 0 0 0 0 0 0 36 0 0 0 0 0
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
(b) This can be looked up in Table 8-1. It is the third entry with L = 4 and N = 12 so
ill le u vi pr
w sa co ro is
or eir is p rk
sin(πk/3) − jπk /4
o
w
Y1 [k] = e
an his
sin(πk/12)
T
th d
>> Y1 = fft([ones(1,4),zeros(1,8)])
Y1 = 4.0000 + 0.0000i 2.3660 - 2.3660i 0.0000 - 1.7321i
0.0000 + 0.0000i 1.0000 + 0.0000i 0.6340 - 0.6340i
0.0000 + 0.0000i 0.6340 + 0.6340i 1.0000 + 0.0000i
0.0000 + 0.0000i 0.0000 + 1.7321i 2.3660 + 2.3660i
(c) In this problem, the even samples are 1 and the odd indexed samples are zero. This solution is facilitated if we
11
1
X
j (2π/12)6n

write y2 [n] as y2 [n] = 2 1 + e δ[n − m]. Therefore, it follows from the linearity and frequency-shift
m=0
properties in Table 8-2 and the third entry in Table 8-1 with L = N = 12 that
1 sin(πk) − jπk11/12 1 sin(π(k − 6)) − jπ(k−6)11/12

Y2 [k] = 2 e +2 e
sin(πk/12) sin(π(k − 6)/12)
The first term is zero except at k = 0 where its value is Y2 [0] = 6. The second term is zero except when k = 6 where
its value is Y2 [6] = 6. Thus
6
 k = 0, 6
Y2 = 
0
 k = 1, 2, . . . , 5, 7, 8, . . . , 11
We can get another closed form expression by summing the DFT expression only over the even indices using a
index m = 2n, n = 0, 1, . . . , 5. This gives us
5
X 1 − e− j (2π/12)6k 1 − e− jπk
Y2 [k] = e− j (2π/12)2k m = =
m=0
1 − e− j (2π/12)2k 1 − e− j (2π/6)k

CHAPTER 8. DFT
11
X
In this case, the expression is indeterminant for k = 0, 6. The value at these DFT indices is Y2 [0] = y2 [n] = 6,
n=0
and because the odd-indexed samples are zero,
11
X 11
X
Y2 [6] = y2 [n]e− j (2π/12)6n = y2 [n]e− jπn = 6
n=0 n=0
)
eb
>> Y2 = fft([1,0,1,0,1,0,1,0,1,0,1,0])
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
Y2 = 6 0 0 0 0 0 6 0 0 0 0 0
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 8. DFT
P-8.4 DSP First 2e
(a) The DFT of y[n] is

5
X 1 − e− j2πk
Y [k] = 7e− j (2π/6)k n − 3W [k] = 7 − 3W [k] = 7(6)δ[k] − 3W [k]
n=0
1 − e− j (2π/6)k
Therefore, Y [k] = {39, 0, −3 j, 0, +3 j, 0}

(b) If we write v[n] = 4(−1) n w[n] as v[n] = 4e j (2π/6)3n w[n], we see that the new DFT is V [k] = 4W [k − 3] where the
)
eb
shift is periodic with period 6.
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
Therefore, V [k] = {0, −4 j, 0, 4, 0, 4 j}
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 8. DFT
P-8.5 DSP First 2e
255
X 255
X
(a) X[0] = x[n]e− j2π (0)n/256 = x[n] is real since all x[n] are real-valued.
n=0 n=0
255
X 255
X
X[128] = x[n]e− j2π(128)n/256 = x[n](−1) n is real since all x[n] are real-valued.
n=0 n=0
− j (2π/4)k − j (2π/4)2k
(b) X[k] = 3 − e + 3e − e− j (2π/4)3k = {4, 0, 8, 0}
)
3
eb
1 X
er or in ing
ed id n
W
no the iss tea s
V [k]e j (2π/4)(2)k = 41 (3 + (5 j)(−1) + 0 + (−5 j)(−1)) = 3
itt W tio
(c) v[2] =
w
t p W em ch
e
d on g. in t la
m ld a
N k=0 4
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
9 9
th k ( de f i es
X X
(d) Y [0] = y[n] = 0 Y [5] = 7(−1) n e− j (2π/10)5n = 70
of or stu e o tat
ity s w g us d S
n=0 n=0
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 8. DFT
P-8.6 DSP First 2e
(a) Sketch should show x[n] = {0, 1, 2, 3, 4, 0, 0, 0} starting at n = 0.

(b) Sketch should show y[n] = {2, 3, 4, 0, 0, 0, 0, 1} because y[n] = x[n − 6] evaluated with a period of 8.
Problem 8.6(a), x[n]
4
3
)
eb
er or in ing
2
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
1
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
0
th k ( de f i es
of or stu e o tat
0 1 2 3 4 5 6 7
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
Problem 8.6(b), y[n]

th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
4
de o rse de ot
ill le u vi pr
w sa co ro is
3
or eir is p rk
o
w
an his
2
T
d
1
th
0
0 1 2 3 4 5 6 7
Time Index (n)

CHAPTER 8. DFT
P-8.7 DSP First 2e
(a) Plot y[n] = {2, 1, 0, 0, −2, −1} starting at n = 0.
Problem 8.7, y[n]

2
1
0
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
−1
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
−2
th k ( de f i es
of or stu e o tat
0 1 2 3 4 5
ity s w g us d S
is
te f t ss th nite
Using the method of synthetic polynomial multiplication to evaluate the convolution, we get
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
1 1 1 1
de o rse de ot
ill le u vi pr
2 -1 -1 so y[n] = 2δ[n] + δ[n − 1] − 2δ[n − 4] − δ[n − 5]

w sa co ro is
or eir is p rk
2 1 0 0 -2 -1
o
w
an his
T
(b) Applying the definition of the DFT and writing out the individual terms gives
th d
X[k] = 1 + e− j (2π/N )k + e− j (2π/N )k (2) + e− j (2π/N )k3 + e− j (2π/N )k (4)

H[k] = 2 − e− j (2π/N )k − e− j (2π/N )k (2)
(c) Forming the product Y [k] = H[k]X[k] is the same as doing the polynomial multiplication as in part (a) so we get
Y [k] = 2 + e− j (2π/N )k − 2e− j (2π/N )k (4) − e− j (2π/N )k (5) .
(d) For N = 6, the DFT is Y [k] = 2 + e− j (2π/6)k − 2e− j (2π/6)k (4) − e− j (2π/6)k (5) . All we need to do is pick off the
coeficients of Y [k] to obtain y[n] = {2, 1, 0, 0, −2, −1}
(e) We need to adjust the exponents of W [k] so that any power of e− j (2π/4)k (n+4) is replaced by e− j (2π/4)k n .
Thus, Y [k] = 2 + e− j (2π/4)k − 2e− j (2π/4)k (4) − e− j (2π/4)k (5) is equivalent to
Y [k] = 2 + e− j (2π/4)k − 2e− j (2π/4)k (0) − e− j (2π/4)k = 0.
(f) Since the length of the convolution is L + M − 1, we need to be able to represent the sequence of this length by the
DFT. Therefore, N ≥ L + M − 1 is required.

CHAPTER 8. DFT
P-8.8 DSP First 2e
In general, the 100-point DFT would be

sin(12.5(2π/100)k)
X[k] = X (e j (2π/100)k ) = 7e− j (2π/100)18k
sin(πk/100)
If this is evaluated for k = 13 we get X[13] = X (e j (2π/100)13 ) = 12.46 e j0.32π = 12.46 e j1.005
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 8. DFT
P-8.9 DSP First 2e
(a) From Q(e j ω̂ ) it is clear that the peaks will occur around DFT indices corresponding to the frequencies ω̂0 and
2π − ω̂0 ; i.e. when (2π/32)k1 = ω̂0 and (2π/32)(32 − k1 ) = 2π − ω̂0 . Therefore, ω̂0 = (2π/32)10 = 0.625π.
(b) The second peak occurs when (2π/32)(32 − k1 ) = (2π/32)k2 or when k2 = 22. The third nonzero value occurs at
k 3 = 0, corresponding to the DC component of 0.1 in s[n]. The DFT of that DC component is 0.1N δ[k], which is
zero for all k except k = 0. The Dirichlet components are also zero except at k1 = 10 for the first one and k 2 = 22
for the second one.
(c) The peak magnitude at k = 10 is 12 ( A)(32) = 50 so A = 100/32 = 3.125.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
>> x=0.1+(100/32)*cos(2*pi*10*(0:31)/32);
e in nt ns co
D
th k ( de f i es
>> X=fft(x);
of or stu e o tat
ity s w g us d S
>> abs(X)
is
te f t ss th nite
ans = 3.2000 0.0000 0.0000 0.0000 0.0000 0.0000
e rt ss fo U
gr hi in e
th a a ly by
k
0.0000 0.0000 0.0000 0.0000 50.0000 0.0000
in o e r
y y p d le d
ro n an o te
0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

st f a s d s ec
de o rse de ot
0.0000 0.0000 0.0000 0.0000 50.0000 0.0000

ill le u vi pr
w sa co ro is
0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

or eir is p rk
o
w
0.0000 0.0000
an his
T
th d

CHAPTER 8. DFT
P-8.10 DSP First 2e
(a) N = 32 + 48 = 80.
(b) k peaks = 8, and −8 → (80 − 8 = 72) because (2π/80)k peaks = ±0.2π
The peaks will occur around DFT indices corresponding to ω̂ = ±0.2π, or when (2π/80)k = 0.2π or k = 8 and
when (2π/80)(80 − 8) = 2π − 0.2π or k = 72.
(c) max{|R[k]|} = 118.4553. Evaluate in Matlab via
k
r=1+7*cos(0.2*pi*(0:31)); R=fft(r,80); [mx,ix]=max(abs(R))
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 8. DFT
P-8.11 DSP First 2e

sin(0.15πn)
The impulse response h[n] = corresponds to a frequency response
5πn
 1/5 |ω̂| ≤ 0.15π

H (e j ω̂ ) = 

0 0.15π < |ω̂| ≤ π

The input signal has fundamental frequency ω̂0 = 0.08π. Thus, only DC and the first harmonic are below the cutoff
of the filter. Since the gain is 1/5, we get three terms in the output
)
eb
er or in ing
ed id n
W
no the iss tea s

y[n] = (1/5) 1 + 2e j0.08πn + 2e− j0.08πn = 0.2 + 0.4 cos(0.08πn)
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 8. DFT
P-8.12 DSP First 2e
(a) In order to show the spectrum only for positive frequencies, we need to convert the frequencies by adding 2πn to
the exponent of each of the components having negative frequency. Therefore,
x[n] = 3 + 2e j0.2πn + 2e j (2π−0.2π)n − 7 je j0.7πn + 7 je j (2π−0.7π)n

= 3 + 2e j0.2πn + 2e j1.8πn − 7 je j0.7πn + 7 je j1.3πn
Therefore, the spectrum plot is as follows:
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
7j −7 j 7j
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
3 3
ity s w g us d S
2 2 2
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
π
-
−.2π 0 .2π .7π 1.3π 1.8π 2π ω̂
st f a s d s ec
−π −.7π
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
The dashed lines show negative frequencies that differ by 2π from frequencies in the range 0 ≤ ω̂ < 2π.
o
w
an his
T
d
(b) If we carry out the multiplication and adjust any resulting negative frequencies as we did in part (a), we get
th
x 1 [n] = e j0.4πn x[n] = 3e j0.4πn + 2e j0.6πn + 2e0.2πn − 7 je j1.1πn + 7 je j (−0.3π)n

= 3e j0.4πn + 2e j0.6πn + 2e0.2πn − 7 je j1.1πn + 7 je j1.7πn
This is an example of the frequency shift property since by adding 0.4π to each frequency we shift the spectrum to
the right by that amount. The resulting spectrum is
−7 j 7j −7 j 7j
3
2 2
π 1.1π
-
−.9π −.3π 0 .2π .4π .6π 1.7π 2π ω̂
(c) Now we can write x 2 [n] = (−1) n x[n] = e jπn x[n] and we see that by the same process as in part (b), each of
the original frequencies will be increased by π radians. If we choose the alias frequencies that are in the range
0 ≤ ω̂ < 2π, we can write
x[n] = 3e jπn + 2e j1.2πn + 2e j0.8πn − 7 je j1.7πn + 7 je j0.3πn
Therefore the spectrum plot is that of part (a) shifted to the right by π radians.

CHAPTER 8. DFT
−7 j 7j −7 j
3 3
2 2 2
π 1.2π
-
−π −.8π −.3π 0 .3π .8π 1.7π 2π ω̂
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 8. DFT
P-8.13 DSP First 2e
(a) The sampled signal is
x[n] = x(n/60) = 6 cos(42πn/60) + 4 cos(18πn/60 − 0.5π)

= 6 cos(2π(7/20)n) + 4 cos(2π(3/20)n − 0.5π)
= 6 cos(0.7πn) + 4 cos(0.3πn − 0.5π)
Since 7 and 3 have no common factors, the period is N = 20 and the fundamental frequency is ω̂0 = 2π/20.
)
eb
(b) x[n] = 2e− j0.5π e j (2π/20)3n + 2e j0.5π e− j (2π/20)3n + 3e j (2π/20)7n + 3e− j (2π/20)7n
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(c) From the answer to part (b), the DFS coefficients are c3 = 2e− j0.5π , c−3 = 2e j0.5π , c7 = 3, c−7 = 3.
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
(d) Here is the spectrum plot of the DFS:
th k ( de f i es
of or stu e o tat
3 3 ity s w g us d S 3 3
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
2j −2 j 2j −2 j
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
π
-
−π −.7π −.3π 0 .3π 1.3π 1.7π 2π 2.3π 2.7π 3π ω̂
T
th d
Dashed lines are aliases of the frequencies in the range −π < ω̂ ≤ π.

CHAPTER 8. DFT
P-8.14 DSP First 2e
sin(4ω̂) − j ω̂7/2
(a) The frequency response of this filter is H (e j ω̂ ) = e . The zeros of this function are at frequencies
sin(ω̂/2)
where 4ω̂ = πr where r is an integer; i.e., where ω̂ = 0.25π, 0.5π, 0.75π, π. The equivalent values of DFT index
k are when (2π/N )k = 0.25πr or k = Nr/8. For N = 512, these values are multiples of 64.
Problem 8.14
8
)
eb
6
er or in ing
Magnitude
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
4
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
2
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
0
gr hi in e
th a a ly by
k
0 32 64 96 128 160 192 224 256
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
pi/2
or eir is p rk
o
w
an his
T
Phase (rad)
0
th
−pi/2
−pi
0 32 64 96 128 160 192 224 256
DFT Frequency Index (k)
(b) Note that the given frequencies are all multiples of 0.25π so all the frequencies except DC fall at the zeros of
H (e j ω̂ ). Therefore, these frequencies do not appear in the output. The gain of the filter at ω̂ = 0 is H (e j0 ) = 8, so
the output is y[n] = 3(8) = 24 for all n. You can pick all these values off the plot, or evaluate H (e j ω̂ ) at individual
frequencies with freqz.

CHAPTER 8. DFT
P-8.15 DSP First 2e
(a) We have two constraints. To avoid aliasing we need 2(1000) < f s . To meet the frequency spacing condition, we
need f s /N ≤ 5, where N = 2ν .
The combined constraints are given by 2000 < f s ≤ 5 · 2ν , or equivalently 400 < f s /5 ≤ ·2ν .
Thus we require N = 512 = 29 . If we fix N at 512, the condition on f s is 2000 < f s ≤ 2560 Hz.
(b) As discussed in Section 8-7.4.1, the width of the main lobe of the Hann and Hamming windows is approximately
∆ω̂ = 8π/L where L is the window length. Therefore, for L = N/2 = 256, we should expect a main lobe width
of about 8π/256 = π/32 rad. This is equivalent to analog radian frequency of ∆ω = (π/32) f s or using analog
)
eb
er or in ing
ed id n
W
no the iss tea s
frequency in hertz, ∆ f = f s /64.
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
Using the constraints on f s obtained in part (a), we get 2000/64 < ∆ f ≤ 2560/64, or 31.25 < ∆ f ≤ 40.
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 8. DFT
P-8.16 DSP First 2e
(a) Here’s a Matlab plot of the ideal spectrogram. A sketch should look like this.
Problem 8-16a: Ideal Spectrogram
2000
1800
1600
1400
frequency
1200
)
eb
er or in ing
ed id n
1000
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
800
.
r
or ud a uc y
600
w cl le tr p
e in nt ns co
D
th k ( de f i es
400
of or stu e o tat
ity s w g us d S
200
is
te f t ss th nite
e rt ss fo U
0
gr hi in e
th a a ly by
0 0.2 0.4 0.6 0.8 1 1.2
k
in o e r
y y p d le d
time axis
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
(b) The width of the main lobe of the DTFT of the Hann window is approximately ∆ω̂ = 8π/L, where L is the
w sa co ro is
or eir is p rk
o
window length. Therefore the normalized main lobe width for the L = 200 window is ∆ω̂ = 0.04π. In terms of
w
an his
analog cyclic frequency, this would be equivalent to ∆ f = ∆ω̂ f s /(2π) = 0.04π8000/(2π) = 160 Hz. Thus, the
T
d
spectrogram sketch should show bars about 160 Hz wide and there should be blurred regions of approximately 200
th
samples (or equivalently, 200/8000 = 0.025 s) duration at the beginning and end of each sinusoidal component.
The following is a Matlab spectrogram of the signal.
Problem 8-16b: Spectrogram with 200-point Window
2000
1800
1600
1400
frequency
1200
1000
800
600
400
200
0
0 0.2 0.4 0.6 0.8 1 1.2
time axis

CHAPTER 8. DFT
P-8.17 DSP First 2e
Note that there are 8 equal duration notes in (approximately) 2 seconds of time. Therefore, each note has duration
2/8 s or 250 ms.
Spectrogram of C Major scale, L D 500; fs D 4 kHz
600
550
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
500
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
450
e in nt ns co
D
Frequency (Hz)
th k ( de f i es
of or stu e o tat
ity s w g us d S
400
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
350
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
300
w sa co ro is
or eir is p rk
o
w
an his
250
T
th d
200
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Time (s)
Figure P-8.17

CHAPTER 8. DFT
P-8.18 DSP First 2e
The ratio of the window lengths is 256/100 ≈ 2.5.

The sketch will look similar to the spectrogram in Fig. P-8.17, but it will have bars that are 2.5 times as wide in the
frequency dimension. The frequency width of the horizontal bars in Fig. P-8.17 is approximately 16 Hz, so it will change
to 40 Hz in the new figure.
In the time dimension, the fuzzy regions at the time of frequency change will be shorter—their duration will be 40%
of the duration of those in Fig. P-8.17 which are approximately 0.03 s. Thus, in the new figure the duration of the fuzzy
regions will be approximately 0.01 s.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 8. DFT
P-8.19 DSP First 2e
(a) The Hamming window has a main lobe width of approximately ∆ω̂ = 8π/L, where L is the window length. For the
given parameters, this would give a main lobe width equivalent to ∆ f = 4 f s /L = 4(8000)/256 = 125 Hz. Thus,
two sinusoids whose frequencies are farther apart than 125 Hz will show up as two distinct peaks, but if the two
frequencies are closer that 125 Hz they will tend to blend together in the spectrum.
(b) The window length is equivalent to 256/8000 =32 ms so if there are abrupt changes in the waveform, the spectral
characteristics will look blurred over approximately 32 ms around the abrupt change.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

CHAPTER 8. DFT
P-8.20 DSP First 2e
(a) The Hamming window has a main lobe width of approximately ∆ω̂ = 8π/L, where L is the window length. In
terms of analog cyclic frequency this corresponds to ∆ f = 4 f s /L, where L is the window length. For the given
parameters, we want 40000/L = 250 or L = 160.
(b) Here is the plot of the DTFT magnitude:
DTFT of Windowed Sinusoids at Frequencies 3000 and 3250 Hz
)
eb
45
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
40
e
d on g. in t la
m ld a
35
.
r
or ud a uc y
w cl le tr p
30
e in nt ns co
D
th k ( de f i es
25
of or stu e o tat
ity s w g us d S
20
is
te f t ss th nite
15
e rt ss fo U
gr hi in e
th a a ly by
10
k
in o e r
y y p d le d
ro n an o te
5
st f a s d s ec
de o rse de ot
0
ill le u vi pr
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
w sa co ro is
analog frequency in Hz
or eir is p rk
o
w
an his

T
th d

CHAPTER 8. DFT
P-8.21 DSP First 2e
(a) The notes are in the octave above A-440, so the frequencies are given by the equation f = 440(2n/12 ) with
n = 0, 1, . . . , 12 covering one octave from A-440 to A-880. Here are the notes and their theoretical frequencies.
(Subscript denotes octave number.)
note A4 B4 [ B4 C5 C5 ] D5 D5 ] E5 F5 F5 ] G5 G5 ] A5
f 440 466 494 523 554 587 622 659 698 740 784 831 880
Now the first 9 notes of Für Elise are { E5 , D5 ], E5 , D5 ], E5 , B4 , D5 , C5 , A4 }. Therefore, the theoretical frequencies
for the first nine notes are { 659, 622, 659, 622, 659, 494, 587, 523, 440 }.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
Spectrogram of Fur Elise (9 notes), L D 1000; fs D 8 kHz
e
d on g. in t la
m ld a
800
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
750
th k ( de f i es
of or stu e o tat
700
ity s w g us d S
is
te f t ss th nite
650 E
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
600
ro n an o te
D
st f a s d s ec
de o rse de ot
Frequency (Hz)
ill le u vi pr
550
w sa co ro is
C
or eir is p rk
o
500
w
B
an his
T
d
450
A
th
400
350
E
300
250
200
6 6.25 6.5 6.75 7 7.25 7.5 7.75 8 8.25 8.5 8.75 9
Time (s)
(b) The spectrogram in the figure above is marked with the true frequencies. Observe that the passage of interest starts
at t = 6.4 s (approximately). It is difficult to obtain very accurate estimates of the frequencies by measuring on the
spectrogram because the spectral peaks have a width of about 10 Hz. However, careful examination of the figure
above confirms that the theoretical note frequencies are close to the center of the spectral peaks.
(c) Observe that the passage of interest starts at t = 6.4 s (approximately). From the score in Fig. 3-22, the first eight
notes are 16th notes while note nine is an eighth note, which should have twice the duration.
The first five notes are easy to distinguish, but overlap in time due to the sliding window of the spectrogram. Their
total duration is 7.7 − 6.4 = 1.3 s, so each 16th note has a duration of approximately 0.26 s. The sixth note (B4 )
appears shorter (≈ 0.15 s), but might be softer so its second half is hard to see; likewise, for note eight (C5 ). The
seventh note (D5 ) is barely visible, but there is evidence of a spectral peak in the 0.2 s gap between B4 and C5 . The
ninth note (A4 ) is easy to distinguish, and its duration seems a bit longer (≈ 0.3 s), but not twice as long.

CHAPTER 8. DFT
P-8.22 DSP First 2e
(a) From the spectrogram it appears that ω̂0 = 0.7π and ω̂1 = 0.75π. To estimate the chirp parameter α, recall that the
instantaneous frequency is ω̂i [n] = 2αn. From the spectrogram we see that ω̂i [n] starts at 0 and increases linearly
to ω̂i [2000] = 0.5π = 2α(2000). Therefore, α = π/8000.
(b) The Hann window has a main lobe width of approximately ∆ω̂ = 8π/L, where L is the window length. Thus, if we
multiply the window length by 4, we divide the main lobe width by 4. Thus, the spectrogram with L = 400 will
like Fig. P-8.22 except that the bars will be much narrower—they will have 1/4 the width.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-9.1 DSP First 2e
X1 (z) = 7, X2 (z) = z −1, X3 (z) = 5z −4, X4 (z) = 7 − z −1 + 5z −4
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-9.2 DSP First 2e
Y (z) = X (z) − z −1 X (z) = (1 − z −1 )X (z) = H (z)X (z)
where H (z) = 1 − z −1 is the z-transform of h[n] = δ[n] − δ[n − 1], the impulse response of the system.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-9.3 DSP First 2e
(a) y[n] = x[n] + 5x[n − 2] − 3x[n − 3] + 2x[n − 5] + 4x[n − 7],

(b) h[n] = δ[n] + 5δ[n − 2] − 3δ[n − 3] + 2δ[n − 5] + 4δ[n − 7],
Plot {1, 0, 5, −3, 0, 2, 0, 4} for 0 ≤ n ≤ 8.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-9.4 DSP First 2e
(a) H (z) = 0.1(1 − z −1 + z −2 )

(b) The roots of H (z) are e±π/3 so we can write
(z − e jπ/3 )(z − e− jπ/3 )

H (z) = 0.1(1 − e jπ/3 z −1 )(1 − e− jπ/3 z −1 ) = 0.1
z2
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
0.5
w
Imaginary Part
t p W em ch
e
d on g. in t la
m ld a
2
0
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
-0.5
of or stu e o tat
ity s w g us d S
-1
is
te f t ss th nite
e rt ss fo U
-1 -0.5 0 0.5 1
gr hi in e
th a a ly by
k
Real Part
in o e r
y y p d le d
ro n an o te
(c) H (e j ω̂ ) = 0.1(1 − e− j ω̂ + e− j ω̂2 )

st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
0.3
or eir is p rk
o
w
0.2π
an his
T
d
0.1
th
0
−π −0.8π −0.6π −0.4π −0.2π 0 0.2π 0.4π 0.6π 0.8π π
5
H(ej ω̂ )
0
6
-5
−π −0.8π −0.6π −0.4π −0.2π 0 0.2π 0.4π 0.6π 0.8π π
(d) ω̂
(e) For this we need the frequency response evaluated at ω̂ = 0, π/4, 2π/3.
H (e j0 = 1 H (e jπ/4 ) = 0.4142e− jπ/4 H (e j2π/3 ) = 2e jπ/3
Therefore, the output is
y[n] = 9 − 8(0.4142) cos[0.25π(n − 1) − π/4] + 7(2) cos[(2π/3)n + π/3]

= 9 − 3.3136 cos[0.25π(n − 2)] + 14 cos[(2π/3)n + π/3]

P-9.5 DSP First 2e
(a) We have to multiply out the factors to get a polynomial in z −1 .
H (z) = 1 + 2z −1 + 2.64z −2 + 2.28z −3 + 1.28z −4 + 0.64z −5
Then the difference equation is:
y[n] = x[n] + 2x[n − 1] + 2.64x[n − 2] + 2.28x[n − 3] + 1.28x[n − 4] + 0.64x[n − 5]
)
eb
(b) There are five poles at z = 0 and the five zeros are at z = −1, ± j0.9, and e± j2π/3 .
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
1
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
0.5
Imaginary Part
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
5
0
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
-0.5
or eir is p rk
o
w
an his
T
d
-1
th
-1 -0.5 0 0.5 1
Real Part
(c) We need to find the frequencies where H (e j ω̂ ) = 0.

These frequencies are the angles of the zeros that are on the unit circle; i.e., (1 + z −1 ) = 0, (1 − e j2π/3 z −1 ) = 0 and
(1 − e− j2π/3 z −1 ) = 0, or ω̂ = π, ±2π/3.

P-9.6 DSP First 2e
(a)
H (z) = H2 (z)H1 (z)

= 14 (1 + z −1 + z −2 + z −3 ) 13 (1 + z −1 + z −2 )
1
= 12 (1 + 2z −1 + 3z −2 + 3z −3 + 2z −4 + z −5 )
(b)
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
Y2 (z) = X (z)H2 (z)
d on g. in t la
m ld a
.
= (1 − z −1 ) 31 (1 + z −1 + z −2 )
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
= 13 (1 − z −3 )
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
W (z) = Y2 (z)H1 (z)

gr hi in e
th a a ly by
k
in o e r
y y p d le d
= 31 (1 − z −3 ) 14 (1 + z −1 + z −2 + z −3 )
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
1
= 12 (1 + z −1 + z −2 − z −4 − z −5 − 3z −3 )
w sa co ro is
or eir is p rk
o
w
an his
T
d
1
(c) y[n] = 12 (x[n] + 2x[n − 1] + 3x[n − 2] + 3x[n − 3] + 2x[n − 4] + x[n − 5])
th
Clearly, the samples are not weighted equally although the sum of the weights (coefficients) is equal to one.
(d) There are five poles at z = 0, and the five zeros are at z = −1, ± j, and e± j2π/3
0.5
Imaginary Part
5
0
-0.5
-1
-1 -0.5 0 0.5 1
Real Part

P-9.7 DSP First 2e
>> z=roots([1,-.5,.5,-1])
z =
1.0000
-0.2500 + 0.9682i
-0.2500 - 0.9682i
>> magz=abs(z)
magz =
)
eb
er or in ing
1.0000
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
1.0000

1.0000
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
>> angz=angle(z)/pi
th k ( de f i es
of or stu e o tat
angz =
ity s w g us d S
0
is
te f t ss th nite
e rt ss fo U
0.5804
gr hi in e
th a a ly by
k
in o e r
y y p d le d
-0.5804
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
Thus, the zeros are at z = 1, e± j0.5804π and there is one pole at z = 0 for each of the zeros.
w sa co ro is
or eir is p rk
o
w
an his
1
T
th d
0.5
Imaginary Part
3
0
-0.5
-1
-1 -0.5 0 0.5 1
Real Part

P-9.8 DSP First 2e
(a) Impulse response:

4
X
h[n] = δ[n] − δ[n − 1] + δ[n − 2] − δ[n − 3] + δ[n − 4] = (−1) k δ[n − k]
k=0
4 4
X X 1 − (−z −1 ) 5 1 + z −5 z5 + 1
(b) H (z) = (−1) k z −k = (−1z −1 ) k = = = 4
1 − (−z )
−1 1+z −1 z (z + 1)
)
k=0 k=0
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
(c) Plot the poles and zeros of H (z) in the complex z-plane.
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
1
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
0.5
is
te f t ss th nite
Imaginary Part
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
4
ro n an o te
0
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
-0.5
o
w
an his
T
th d
-1
-1 -0.5 0 0.5 1
Real Part
1 + e− j ω̂5 cos(ω̂2.5)
(d) H (e j ω̂ ) = = e− j ω̂2
1 + e− j ω̂ cos(ω̂/2)
(e) Sketch of the frequency response (magnitude and phase) , or Matlab plot.
6
|H(ej ω̂ )|
0
−π −0.6π −0.2π 0 0.2π 0.6π π
5
H(ej ω̂ )
0
6
-5
−π −0.6π −0.2π 0 0.2π 0.6π π
ω̂
Figure P-9.8
(f) We need the frequency response evaluated at ω̂ = 0, 0.5π, 0.6π. These values are H (e j ω̂ ) = 1, 1, 0.
Therefore, y[n] = 5 + 4 cos(0.5πn).

P-9.9 DSP First 2e
(a) H (z) = (1 − z −1 ) 5
(b) Plot of the poles and zeros of H (z) in the z-plane.
0.5
Imaginary Part
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
5 5
w
t p W em ch
e
d on g. in t la
0
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
-0.5
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
-1
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
-1 -0.5 0 0.5 1
st f a s d s ec
de o rse de ot
ill le u vi pr
Real Part
w sa co ro is
or eir is p rk
o
(c)
w
an his
T
d
5
H (e j ω̂ ) = (1 − e− j ω̂ ) 5 = e− j ω̂5/2 = 2e jπ/2 e− j ω̂/2 sin(ω̂/2)

th
= 32e (π−ω̂)5/2 sin5 (ω̂/2)
(d)
H (z) = (1 − z −1 ) 5 = 1 − 5z −1 + 10z −2 − 10z −3 + 5z −4 − z −5

h[n] = δ[n] − 5δ[n − 1] + 10δ[n − 2] − 10δ[n − 3] + 5δ[n − 4] − δ[n − 5]

P-9.10 DSP First 2e
(a) We can find N1 as the lowest power of z −1 of the product X (z)H (z). That will be N1 = 0 + 6 = 6. Similarly, N2 is
the highest power of z −1 if the product X (z)H (z). That is N2 = 9 + 9 = 18.
(b) Again, we can use the polynomials of X (z) and H (z) to get the answer. y[6] is the product of the coefficients of the
two lowest order terms or y[6] = 2 × 1 = 2. Similarly, y[18] is the product of the coefficients of the highest order
terms, or y[18] = −1 × 3 = −3.
(c) y[n] = 2δ[n − 6] − 4δ[n − 7] + 8δ[n − 8] + δ[n − 9] − 6δ[n − 10] + 12δ[n − 11] − 11δ[n − 12]
+16δ[n − 13] − 32δ[n − 14] + 11δ[n − 15] − 6δ[n − 16] + 12δ[n − 17] − 3δ[n − 18]
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-9.11 DSP First 2e
(a) H (z) = 1 + z −1 + z −2 = (1 − e j2π/3 z −1 )(1 − e− j2π/3 z −1 ), so there are zeros at z = e± j2π/3 .

(b) To solve this part we need the frequency response evaluated at ω̂ = π/6. That is,
sin(ω̂3/2)
H (e j ω̂ ) = 1 + e− j ω̂ + e− j ω̂2 = e− j ω̂
sin(ω̂/2)
so H (e jπ/6 ) = 2.7321e− jπ/6 and, therefore, the formula for the steady-state output is
)
eb
er or in ing
ed id n
y[n] = 2.7321 cos((π/6)(n − 1) − π/4)
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
(c) omega = 2*pi/3 gives zero for the steady-state output after n ≥ 2. (Until the last two samples.)
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-9.12 DSP First 2e
(a) y[n] = x[n] − x[n − 4]

(b) H (e j ω̂ ) = 1 − e− j ω̂4 = 2e j (π/2−ω̂2) sin(ω̂2)
π/2 − 2ω̂ 0 < ω̂ < π/2
(
j ω̂ j ω̂
(c) |H (e )| = 2| sin(ω̂2)| and ∠H (e )=
π/2 − 2ω̂ + 2π π/2 < ω̂ < π
(d) The frequencies nulled are the angles of the zeros of H (z); ω̂0 = 0, ±π/2, π.
(e) y[n] = |H (e jπ/3 )| cos(πn/3 + ∠H (e jπ/3 )) = 1.7321 cos(πn/3 − π/6)
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-9.13 DSP First 2e
We should find the output as the sum of the response to DC plus the response to an impulse of size -70 plus the re-
sponse to the sinusoid of frequency 0.5π. We need the impulse response and the frequency response function to do
this.
From H (z) in expanded form, h[n] = δ[n] − 5δ[n − 2] + 4δ[n − 4] and
H (e j ω̂ ) = (1 − e− j ω̂2 )(1 − 4e− j ω̂2 ) = 1 − 5e− j ω̂2 + 4e− j ω̂4
so H (e j0 ) = 0 and H (e j0.5π ) = 1 − 5e− j2π/2 + 4e− j4π/2 = 1 − 5e−π + 4e− j4π/2 = 10. Therefore,
)
eb
er or in ing
ed id n
W
no the iss tea s
y[n] = −70δ[n] + 350δ[n − 2] − 280δ[n − 4] + 300 cos(0.5πn + π/4)
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-9.14 DSP First 2e
After sampling, we get the sequence x[n] = 4 + cos(1000πn/2000 − π/4) − 3 cos(500πn/2000). Thus, we need to
evaluate the frequency response at ω̂ = 0, π/2, π/4 to find the outputs due to each component of the input. These values
are H (e j0 ) = 0, H (e jπ/2 ) = 0, and H (e jπ/4 ) = 1.0824e jπ/8 . Therefore,
y[n] = −3(1.0824 cos(πn/4 + π/8) = 3.2472 cos(πn/4 − π + π/8) = 3.2472 cos(πn/4 − 2.7489)
and therefore the output of the D-to-C converter with f s = 2000 is
y(t) = 3.2472 cos(π2000t/4 − 2.7489) = 3.2472 cos(500πt − 2.7489)
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-9.15 DSP First 2e
(a) α = 4, β = 10
From H (z) we get H (e j ω̂ ) = 5(1 + .8e− j ω̂ + e− j ω̂2 . By factoring e− j ω̂ out and using the inverse Euler relation we
get
H (e j ω̂ ) = e− j ω̂ (4 + 10 cos(ω̂))
(b) In the plot the DC value is +14, H (e jπ ) = −6, and H (e j ω̂ ) = 0 at ω̂ = 0.631π = 1.9823.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
15

|H(ej ω̂ )|
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
10
th k ( de f i es
of or stu e o tat
ity s w g us d S
5
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
0
in o e r
y y p d le d
−π π
ro n an o te
−0.8π −0.6π −0.4π −0.2π 0 0.2π 0.4π 0.6π 0.8π

st f a s d s ec
de o rse de ot
2
ill le u vi pr
H(ej ω̂ )
w sa co ro is
or eir is p rk
o
w
an his
0
T
d
6
th
-2
−π −0.8π −0.6π −0.4π −0.2π 0 0.2π 0.4π 0.6π 0.8π π
ω̂
Figure P-9.15
(c) We want the system function to have the form H (z) = (1−e j ω̂0 z −1 )(1−e− j ω̂0 z −1 ) = 1−2 cos ω̂0 z −1 +z −2 . Therefore,
we must set b1 = −2 cos ω̂0 . The normalized frequency that we want to null is ω̂0 = 2π f 0 / f s = 2π(1600/8000) =
0.4π. Thus, we need to set b1 to b1 = −0.618.

P-9.16 DSP First 2e
(a) Factoring the two second-order factors gives
H (z)(1 − z −1 )(1 + z −1 )(1 − 0.8e jπ/4 z −1 )(1 − 0.8e− jπ/4 z −1 )(1 + e jπ/2 z −1 )(1 + e− jπ/2 z −1 )
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
0.5

Imaginary Part
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
6
th k ( de f i es
0
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
-0.5
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
-1
w sa co ro is
or eir is p rk
o
w
-1 -0.5 0 0.5 1
an his
T
Real Part
th d
Figure P-9.16
(b) To satisfy the condition, we must group the zeros at ±1 and e± jπ/2 together to form H1 (z) as in
H1 (z) = (1 − z −2 )(1 + z −2 ) = (1 − z −4 )
. That leaves the remainder for H2 (z) as
H2 (z) = (1 − 0.8e jπ/4 z −1 )(1 − 0.8e− jπ/4 z −1 ) = 1 − 1.6 cos(π/4)z −1 + 0.64z −2
.
(c) y[n] is the output of H2 (z) when the input is y1 [n] so from H2 (z) we get
y[n] = y1 [n] − 1.6 cos(π/4)y1 [n − 1] + 0.64y1 [n − 2]

P-9.17 DSP First 2e
(a) h[n] = 7δ[n] − 14.35δ[n − 1] + 22.4175δ[n − 2] − 14.35δ[n − 3] + 7δ[n − 4]

We need to determine the expanded version of H (z) with an additional gain factor.
H (z) = A(1 − 0.8e jπ/3 z −1 )(1 − 0.8e− jπ/3 z −1 )(1 − 1.25e jπ/3 z −1 )(1 − 1.25e− jπ/3 z −1 )
= (1 − 0.8z −1 + 0.64z −2 )(1 − 1.25z −1 + 1.5625z −2 )
Clearly, when expanded out, the H (z) polynomial will be of the form
H (z) = A + . . . + A(0.8) 2 (1.25) 2 z −4 , so if h[0] = 7 then A = 7.
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(b) H (z) = 1.81406 − 3.71882z −1 + 5.80952z −2 − 3.71882z −3 + 1.81406z −4
.
r
or ud a uc y
If we expand H (z) into a polynomial with unknown gain we get
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
H (z) = A(1 − 2.05z −1 + 3.2025z −2 − 2.05z −3 + z −4 )
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
The DC value of the frequency response is

th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
H (e j0 ) = A(1 − 2.05 + 3.2025 − 2.05 + 1) = 1.1025A = 2

de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
so A = 1.8141 and
o
w
an his
T
d
H (z) = 1.8141(1 − 2.05z −1 + 3.2025z −2 − 2.05z −3 + z −4 )

th

P-9.18 DSP First 2e
(a) H (1) = b0 + b1 − b1 − b0 = 0
(b)
H (e j ω̂ ) = b0 + b1 e− j ω̂ − b1 e− j ω̂3 − b0 e− j ω̂4 = e− j ω̂2 (b0 e j ω̂2 + b1 e j ω̂ − b1 e− j ω̂ − b0 e− j ω̂2 )

= [2b0 sin(2ω̂) + 2b1 sin(ω̂)]e j (π/2− j ω̂ (from the inverse Euler formula)
(c)
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
H (z −1 ) = b0 + b1 z − b1 z 3 − b0 z 4 = z 4 (b0 z −4 + b1 z −3 − b1 z −1 − b0 )
e
d on g. in t la
m ld a
.
r
= −z 4 H (z)
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
(d) In the general case with bk = −b M−k , the condition H (1/z) = −z M H (z) holds for M even or odd.
ity s w g us d S
is
te f t ss th nite
The frequency response has the form
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
M /2
st f a s d s ec
de o rse de ot
H (e j ω̂ ) = 2e j (π/2−ω̂ M /2)
X
b M /2−k sin(ω̂k) M an even integer
ill le u vi pr
w sa co ro is
or eir is p rk
k=1
o
w
and
an his
T
d
(M+1)/2
th
H (e j ω̂ ) = 2e j (π/2−ω̂ M /2)
X
b(M+1)/2−k sin(ω̂(2k − 1)/2) M an odd integer
k=1

P-9.19 DSP First 2e
(a) H1 (z) = H2 (z) = 1 + z −1 + z −2 + z −3

(b) H (z) = H1 (z)H2 (z) = (1 + z −1 + z −2 + z −3 ) 2
(c) H (z) = 1 + 2z −1 + 3z −2 + 4z −3 + 3z −4 + 2z −5 + z −6
so h[n] = δ[n] + 2δ[n − 1] + 3δ[n − 2] + 4δ[n − 3] + 3δ[n − 4] + 2δ[n − 5] + δ[n − 6]
(d) y[n] = x[n] + 2x[n − 1] + 3x[n − 2] + 4x[n − 3] + 3x[n − 4] + 2x[n − 5] + x[n − 6]
(e) We can use the formula for the sum of N terms of a geometric series to write H1 (z), or H2 (z) as
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
1 − z −4

H1 (z) = H2 (z) = 1 + z −1 + z −2 + z −3 =
.
r
or ud a uc y
w cl le tr p
1 − z −1
e in nt ns co
D
th k ( de f i es
of or stu e o tat
(f) From part (e) we see that (1 + z −1 + z −2 + z −3 ) 2 so all the zeros at {−1, ± j} are double zeros, and there are six poles
ity s w g us d S
is
te f t ss th nite
at z = 0.
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
2
de le o rse ide rot
1
u v p
sa co ro is
or eir is p rk
o
nd s w
0.5
a hi
ImaginarythPart
T
2 6
ill
0
w
-0.5
2
-1
-1 -0.5 0 0.5 1
Real Part
Figure P-9.19
(g) From the result in (e), we can write
1 − e− j ω̂4 e− j ω̂2 (e j ω̂2 − e− j ω̂2 ) sin(2ω̂) − j3ω̂/2

H1 (e j ω̂ ) = H2 (e j ω̂ ) = = = e
1 − e− j ω̂ e− j ω̂/2 (e j ω̂/2 − e− j ω̂/2 ) sin(ω̂/2)
(h) Note that H (e j ω̂ ) = H2 (e j ω̂ )H1 (e j ω̂ ) = [H1 (e j ω̂ )]2 = [H2 (e j ω̂ )]2 .
!2
sin(2ω̂)
H (e j ω̂ ) = H1 (e j ω̂ )H2 (e j ω̂ ) = e− j3ω̂
sin(ω̂/2)
Since the squared-term is nonnegative, it is the magnitude, and the system has linear phase with time delay equal
to 3 samples.

)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
20
de o rse de ot
ill le u vi pr
|H(ej ω̂ )|
w sa co ro is
or eir is p rk
o
w
an his
10
T
th d
0
−π −0.8π −0.6π −0.4π −0.2π 0 0.2π 0.4π 0.6π 0.8π π
5
H(ej ω̂ )
0
6
-5
−π −0.8π −0.6π −0.4π −0.2π 0 0.2π 0.4π 0.6π 0.8π π
ω̂
Figure P-9.19

P-10.1 DSP First 2e
(a) The system function is

1 1
H (z) = =
1− z −1 − z −2 (1 − p1 z −1 )(1 − p2 z −1 )
√
where, by the quadratic formula, p1,2 = (1 ± 5)/2 = 1.618, −0.618. The partial fraction expansion of H (z) is
√ √
p1 p2 5+1 5−1
)
eb
√ √
er or in ing
ed id n
p1 − p2 p2 − p2 2 5 2 5
W
no the iss tea s
itt W tio
w
t p W em ch
H (z) = + = +
e
d on g. in t la
m ld a
1 − p1 z −1 1 − p2 z −1 1 − p1 z −1 1 − p2 z −1
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
√ √ n √ √ n
5 + 1+ *1 + 5+ 5 − 1+ *1 − 5+
is
te f t ss th nite
h[n] = √ u[n] + √ u[n]
e rt ss fo U
* *
, 2 5 -, 2 - , 2 5 -, 2 -
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
Try a few terms of this closed form expression to see if it generates the Fibonacci sequence, 1, 1, 2, 3, 5, 8, 13, ...
de o rse de ot
ill le u vi pr
√
w sa co ro is
(b) The poles determined in part (a) are p1,2 = (1 ± 5)/2 = 1.618, −0.618. Since one of the poles is outside the unit
or eir is p rk
o
w
circle, this is an unstable system.

an his
T
th d

P-10.2 DSP First 2e
First, find the system function by inspection as
z −1 1
!
H (z) = = z −1 = G(z)z −1
1 − 0.9z −1 + 0.9z −2 1 − 0.9z −1 + 0.9z −2
The z −1 factored out suggests that the impulse response will be zero prior to n = 1. Then we can find the partial fraction
expansion of the rational function part and write G(z) as
0.568e− j0.4942 0.568e j0.4942
)
eb
G(z) = +
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
1 − 0.9487e 1 − 0.9487e− j1.0766 z −1
t p W em ch
j1.0766 z −1
e
d on g. in t la
m ld a
and
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
n n
g[n] = 0.568e− j0.4942 0.9487e j1.0766 u[n] + 0.568e j0.4942 0.9487e− j1.0766 u[n]
of or stu e o tat
ity s w g us d S
is
te f t ss th nite
= 1.136(0.9487) n cos(1.0766n − 0.4942)u[n]
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d

ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
h[n] = g[n − 1] = 1.136(0.9487) n−1 cos(1.0766(n − 1) − 0.4942)u[n − 1]

w sa co ro is
or eir is p rk
o
w
which, by virtue of u[n − 1], is the desired formula showing h[n] being zero for n < 1 and being a decaying sinusoid for
an his
T
d
n ≥ 1.
th
It should be noted that we could have made a partial fraction expansion of H (z) directly. The resulting inverse
transform would be a formula that is valid from n = 0 since the complex exponentials would be multiplied by u[n].
However, when the formula is evaluated at n = 0, the result would be zero.

P-10.3 DSP First 2e
The system function is
5z −7
H (z) =
1 − 0.5z −1
Us the z-transform table and then the delay property,
h[n] = 5(0.5) n−7 u[n − 7]
)
eb
Note that h[n] = 0 for n < 7. It decays exponentially as (0.5) n for n ≥ 7.
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
6
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
4
ity s w g us d S
h[n]
is
te f t ss th nite
2
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
0
st f a s d s ec
de o rse de ot
0 5 10 15 20 25 30
ill le u vi pr
w sa co ro is
time index n
or eir is p rk
o
w
Figure P-10.3
an his
T
th d

P-10.4 DSP First 2e
The impulse response is h[n] = 5(0.5) n u[n], so the output for the given input is
y[n] = 3h[n − 1] − 2h[n − 2] + 3h[n − 4] − h[n − 5]

= 15(0.5) n−1 u[n − 1] − 10(0.5) n−2 u[n − 2] + 15(0.5) n−4 u[n − 4] − 5(0.5) n−5 u[n − 5]
Evaluate with Matlab in order to make a plot.
20
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
10
y[n]
.
r
or ud a uc y
w cl le tr p
0
e in nt ns co
D
th k ( de f i es
of or stu e o tat
ity s w g us d S
-10
is
te f t ss th nite
0 5 10 15 20 25 30
e rt ss fo U
time index n
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
Figure P-10.4
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-10.5 DSP First 2e
The easiest way to work this is to write the output in terms of shifted impulse responses as
y[n] = h[n] + βh[n − 2] + h[n − 4]

√
and then examine y[n] for n ≥ 4. Define a = 1/ 2 so h[n] = 10a n u[n]. Then for n ≥ 4, the output is
y[n] = 10a n + 10 βa n−2 + 10a n−4 = 10a n (1 + βa−2 + a−4 )
To make y[n] = 0 for n ≥ 4, we need 1 + βa−2 + a−4 = 0. Solving this equation for β gives β = −(1 + a4 )/a2 , or for the
)
eb
er or in ing
given value of a, β = −2.5. If you use this value of β, the output looks like the following
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
5
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
0
ity s w g us d S
y[n]
is
te f t ss th nite
e rt ss fo U
-5
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
-10
de o rse de ot
0 5 10 15 20 25 30
ill le u vi pr
w sa co ro is
time index n
or eir is p rk
o
w
Figure P-10.5
an his
T
th d

P-10.6 DSP First 2e
−1 + 3z −1 − 2z −2
(a) The system function is H (z) =
1 − 13 z −1 − 21 z −2
The zeros are at z = 1, 2, and the poles at z = 0.8931, −0.5598
2
1
Imaginary Part
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
0
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
th k ( de f i es
of or stu e o tat
-1
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
-2
ro n an o te
st f a s d s ec
de o rse de ot
-2 -1 0 1 2
ill le u vi pr
w sa co ro is
Real Part
or eir is p rk
o
w
−1 + 3z −1 + 2z −2
an his
(b) The system function is H (z) =

T
d
1 − 13 z −1 + 21 z −2
th
The zeros are now at z = 3.5616, −0.5616, and the poles at z = 0.7071e± j1.3329 .
4
2
Imaginary Part
-2
-4
-4 -2 0 2 4
Real Part

P-10.7 DSP First 2e
(a) y[n] = 13 y[n − 1] − 21 y[n − 2] − x[n]
−1 −z 2
H (z) = =
1 − 13 z −1 + 12 z −2 z 2 − 13 z + 1
2
=⇒Poles: z = 0.7071e± j1.3329 Zeros: z = 0, 0
(b) y[n] = 13 y[n − 1] − 21 y[n − 2] − x[n − 2]
)
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
−z −2

−1
H (z) = =
.
r
or ud a uc y
1 −1 1 −2
− 13 z + 1
w cl le tr p
1− 3z + 2z z2
e in nt ns co
D
2
th k ( de f i es
of or stu e o tat
=⇒Poles: z = 0.7071e± j1.3329 Zeros: z = ∞, ∞
ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
(c) y[n] = 13 y[n − 1] − 21 y[n − 2] − x[n − 4]

k
in o e r
y y p d le d
ro n an o te
st f a s d s ec
de o rse de ot
−z −4
ill le u vi pr
−1
H (z) = =
w sa co ro is
1 −1
+ 12 z −2 z 2 (z 2 − 31 z + 12 )
or eir is p rk
1− 3z
o
w
an his
T
d
=⇒Poles: z = 0.7071e± j1.3329, 0, 0 Zeros: z = ∞, ∞, ∞, ∞

th
In (c) there are four zeros at infinity balanced by two additional poles at the origin.
(a) Two zeros at origin (b) Two zeros at z = ∞ (c) Four zeros at z = ∞
1 1 1
Imaginary Part
Imaginary Part
Imaginary Part
0.5 0.5 0.5

(2) (2)
0 0 0
-0.5 -0.5 -0.5
-1 -1 -1
-1 0 1 -1 0 1 -1 0 1
Real Part Real Part Real Part

P-10.8 DSP First 2e
(a) By inspection,
1 z
H (z) = 1 −1
= 1
1− 2z z− 2
1
H (z) has a pole at z = 2 and a zero at z = 0. The corresponding impulse response is h[n] = (0.5) n u[n].
(b) Write the input as x[n] = δ[n] + δ[n − 2] + δ[n − 5]. Therefore, it follows that for this input the output is y[n] =
)
h[n] + h[n − 2] + h[n − 5]. If we substitute for h[n] we get
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
y[n] = (0.5) n u[n] + (0.5) n−2 u[n − 2] + (0.5) n−5 u[n − 5]
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
Taking the step sequence into account we get
th k ( de f i es
of or stu e o tat
ity s w g us d S
is
0 n<0
te f t ss th nite
e rt ss fo U

n = 0, 1
gr hi in e
(0.5) n

th a a ly by
k

y[n] = 

in o e r
y y p d le d
5(0.5) n n = 2, 3, 4
ro n an o te
st f a s d s ec


37(0.5) n n≥5

de o rse de ot

ill le u vi pr

w sa co ro is
or eir is p rk
o
w
an his
T
th d

P-10.9 DSP First 2e
−15 + 20z −1 −15(z − 4/3)

(a) H (z) = =
1 − 0.75z −1 z − 0.75
(b) Need plot. Pole at z = 0.75, zero at z = 4/3.
−15 + 20e− j ω̂
(c) H (e j ω̂ ) =
1 − 0.75e− j ω̂
(d) |H (e j ω̂ )| 2 = 400
)
eb
er or in ing
conjugate
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
!z }| {!
−15 + 20e− j ω̂ −15 + 20e j ω̂

j ω̂ 2 j ω̂ ∗ j ω̂
|H (e )| = H (e )H (e ) =
.
r
or ud a uc y
w cl le tr p
1 − 0.75e− j ω̂ 1 − 0.75e j ω̂
e in nt ns co
D
th k ( de f i es
of or stu e o tat
625 − 600 cos(ω̂)
ity s w g us d S
=
is
te f t ss th nite
1.5625 − 1.5 cos(ω̂)
e rt ss fo U
gr hi in e
th a a ly by
k
400(1.5625 − 1.5 cos(ω̂))
in o e r
y y p d le d
= = 400
ro n an o te
st f a s d s ec
1.5625 − 1.5 cos(ω̂)

de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
w
an his
T
th d

(a) By inspection,
1 1
H (z) = = √ √
1 + 0.8z −2
(1 − 0.8e jπ/2 z −1 )(1 − 0.8e− jπ/2 z −1 )
z2
= √ √
(z − 0.8e jπ/2 )(z − 0.8e− jπ/2 )
√
)
Zeros at z = 0, 0 and poles at z = 0.8e± jπ/2 .
eb
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(b) To find the output for the given input, determine the impulse response.
.
r
or ud a uc y
w cl le tr p
1 0.5 0.5
e in nt ns co
D
H (z) = = +
th k ( de f i es
√ √
of or stu e o tat
1 + 0.8z −2 1 − 0.8e jπ/2 z −1 1 − 0.8e− jπ/2 z −1
ity s w g us d S
is
te f t ss th nite
so
e rt ss fo U
gr hi in e
th a a ly by
k
√ √ √
in o e r
y y p d le d
h[n] = 12 ( 0.8e jπ/2 ) n u[n] + 12 ( 0.8e− jπ/2 ) n u[n] = 0.8 cos( π2 n)u[n]
ro n an o te
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
Note: h[n] = 0 for n even, which could have been predicted from the fact that the initial H (z) has only z −2 in the
or eir is p rk
o
denominator—an even power.

w
an his
T
d
Therefore,
th
y[n] = 5h[n] + 2.5h[n − 1] + 4h[n − 2] + 2h[n − 3]

√ √
= 5 0.8 cos( π2 n)u[n] + 2.5 0.8 cos( π2 (n − 1))u[n − 1]
√ √
+ 4 0.8 cos( π2 (n − 2))u[n − 2] + 2 0.8 cos( π2 (n − 3))u[n − 3]

(a) By inspection,
1 z5
H (z) = =
1 − 0.97z −5 z 5 − 0.97
(b) There are 5 zeros at z = 0, and the poles are at the 5th roots of 0.97, z k = (0.97) 0.2 e j2πk /5 = (0.9939)e j2πk /5 .
)
eb
er or in ing
ed id n
W
no the iss tea s
1
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
.
r
or ud a uc y
w cl le tr p
0.5
e in nt ns co
D
Imaginary Part
th k ( de f i es
of or stu e o tat
ity s w g us d S
5
is
te f t ss th nite
0
e rt ss fo U
gr hi in e
th a a ly by
k
in o e r
y y p d le d
ro n an o te
-0.5
st f a s d s ec
de o rse de ot
ill le u vi pr
w sa co ro is
or eir is p rk
o
-1
w
an his
T
d
-1 -0.5 0 0.5 1
th
Real Part
Figure P-10.11
(c) The z-transform of the output is

1
Y (z) = H (z)X (z) = (1 + βz −p )
1 − 0.97z −5
If we choose β = −0.97 and p = 5, we get Y (z) = 1 or y[n] = δ[n], which satisfies the condition y[n] = 0 for n ≥ 5.

(a) By using the delay and linearity properties,
ha [n] = (−0.77) n u[n] − (−0.77) n−1 u[n − 1]

1.77/0.77
Ha (z) = −1/.77 + ⇐⇒ x a [n] = −1/0.77 + (1.77/0.77) (−0.77) n u[n]
1 + 0.77z −1 | {z } | {z }
)
2.2987
eb
−1.2987
er or in ing
ed id n
W
no the iss tea s
itt W tio
w
t p W em ch
e
d on g. in t la
m ld a
(b) By using the delay and linearity properties,
.
r
or ud a uc y
w cl le tr p
e in nt ns co
D
hb [n] = (0.9) n u[n] + 0.8(0.9) n−1 u[n − 1]
th k ( de f i es
of or stu e o tat
or, by making the partial fraction expansion, ity s w g us d S
is
te f t ss th nite
e rt ss fo U
gr hi in e
th a a ly by
k
1.7/0.9
in o e r
y y p d le d
Ha (z) = −0.8/0.9 + ⇐⇒ x a [n] = −0.8/0.9 + (1.7/0.9) (0.9) n u[n]

ro n an o te
st f a s d s ec
1 − 0.9z −1
de o rse de ot
| {z } | {z }
ill le u vi pr
−0.8889 1.8889
w sa co ro is
or eir is p rk
o

(N - A) James H. McClellan, Ronald W.schafer, Mark A. Yoder - DSP First Solution Manual-Pears Education (2015)

Uploaded by

Copyright:

Available Formats

(N - A) James H. McClellan, Ronald W.schafer, Mark A. Yoder - DSP First Solution Manual-Pears Education (2015)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(N - A) James H. McClellan, Ronald W.schafer, Mark A. Yoder - DSP First Solution Manual-Pears Education (2015)

Uploaded by

Copyright:

Available Formats

What are sinusoids and how are they represented mathematically?

What are sinusoids and how are they represented mathematically?

What properties allow sinusoids to be added or multiplied?

What properties allow sinusoids to be added or multiplied?

T

J.H. McClellan, R.W. Schafer, M.A. Yoder

P-2.1 DSP First 2e

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.2 DSP First 2e

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.3 DSP First 2e

(a) Plot of cos θ

(b) Plot of cos(20πt)

(c) Plot of cos(2π/T0 + π/2)

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.4 DSP First 2e

( jθ) 2 ( jθ) 3 ( jθ) 4 ( jθ) 5

Thus, e jθ = cos θ + j sin θ

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.5 DSP First 2e

(a) Real part of complex exponential is cosine.

cos(θ 1 + θ 2 ) = < e j (θ1 +θ2 ) = < e jθ1 e jθ2

= < {(cos θ 1 + j sin θ 1 )(cos θ 2 + j sin θ 2 )}

cos(θ 1 + θ 2 ) = cos θ 1 cos θ 2 − sin θ 1 sin θ 2

(b) Change the sign of θ 2 .

cos(θ 1 − θ 2 ) = cos θ 1 cos θ 2 + sin θ 1 sin θ 2

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.6 DSP First 2e

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.7 DSP First 2e

(a) 3e jπ/3 + 4e−jπ/6 = 5e j0.12 = 4.9641 + j0.5981

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.8 DSP First 2e

The variable zz defines z(t), and xx defines x(t) = <{z(t)}.

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.9 DSP First 2e

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.10 DSP First 2e

(b) x(t) = <{z(t)} = <{2.646e−j1.761 e jω0 t }

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.11 DSP First 2e

Add complex amps: e−jπ + e jπ/3 + 2e−jπ/3 = |

Here is the Matlab plot of the vectors.

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.12 DSP First 2e

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.13 DSP First 2e

Three periods of the signal will be 3(1/250) = 12 ms.

(b) Plot q(t) = <{ dt

Time t (ms) rld

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.14 DSP First 2e

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.15 DSP First 2e

Need to add complex amps: 2e j5 + 8e j9 + 4e j0 = 3.051e j2.673

Here is the plot of vectors representing the complex amplitudes:

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015

P-2.16 DSP First 2e

©J. H. McClellan, R. W. Schafer, & M. A. Yoder November 22, 2015