Digital Signal Processing Reference
In-Depth Information
P
(
j
ω
)
baseband
signal
(a)
P
-
P
+
ω
−σ
σ
0
cosine
modulation
(b)
P
-
P
+
P
-
P
+
ω
−ω
c
ω
c
0
sine
modulation
(c)
jP
-
jP
+
ω
c
ω
−ω
c
0
−
jP
-
−
jP
+
Figure 2.24
. Fourier transforms of (a) the baseband pulse signal
p
(
t
), (b) its cosine
modulated version, and (c) its sine modulated version.
Assume the real baseband signal
p
(
t
) has a bandlimited Fourier transform as
in Fig. 2.24(a). Since
cos
ω
c
t
=
e
jω
c
t
+
e
−jω
c
t
2
sin
ω
c
t
=
e
jω
c
t
e
−jω
c
t
2
j
−
and
,
the cosine modulated version has the Fourier transform
P
(
j
(
ω
−
ω
c
)) +
P
(
j
(
ω
+
ω
c
))
2
,
and the sine modulated version has the Fourier transform
P
(
j
(
ω
−
ω
c
))
−
P
(
j
(
ω
+
ω
c
))
=
−
jP
(
j
(
ω
−
ω
c
)) +
jP
(
j
(
ω
+
ω
c
))
2
,
2
j
as indicated in Figs. 2.24(b) and 2.24(c). Since the cosine and sine modulated
versions carry different messages
, we cannot eliminate any
of the side bands. Thus we need to retain both side bands, and the QAM signal
(2.40) is a DSB signal. The QAM signal therefore requires
twice as much band-
width
as a PAM/SSB signal. But, since the QAM signal carries two messages
s
c
(
n
)and
s
s
(
n
) at the same rate, the bandwidth used per real symbol is identical
for the two systems. One advantage of a QAM system is that it does not require
the Hilbert transformer. It is very easy to generate the QAM signal by using
Eq. (2.40), as shown schematically in Fig. 2.25.
{
s
c
(
n
)
}
and
{
s
s
(
n
)
}
Search WWH ::
Custom Search