15

Discrete-Time
Modulation
The modulation property is basically the same for continuous-time and dis-
crete-time signals. The principal difference is that since for discrete-time sig-
nals the Fourier transform is a periodic function of frequency, the convolution
of the spectra resulting from multiplication of the sequences is a periodic con-
volution rather than a linear convolution.
While the mathematics is very similar, the applications are somewhat dif-
ferent. In continuous time, modulation plays a major role in communications
systems for transmission of signals over various types of channels. That appli-
cation usually is inherently a continuous-time application. However, in many
modern communication systems, signals may go through various stages and
types of modulation as they move from one channel to another, and often this
conversion from one modulation system to another is best implemented digi-
tally. In their digital form the signals are discrete-time signals, and such trans-
modulation systems are based on modulation properties associated with dis-
crete-time signals.
In addition to digital modulation systems, the concepts of discrete-time
modulation (and, for that matter, continuous-time modulation also) are useful
in the context of filtering, particularly when it is of interest to implement fil-
ters with a variable center frequency. It is often simpler in such situations to
implement a fixed filter (either continuous time or discrete time) and through
modulation shift the signal spectrum in relation to the fixed filter center fre-
quency rather than shifting the filter center frequency in relation to the signal.
For discrete-time signals, for example, from the modulation property it fol-
lows that multiplying a signal by (- 1)' has the effect of interchanging the
high and low frequencies. Consequently, by alternating the algebraic sign of
the input signal, processing with a lowpass filter, and then alternating the al-
gebraic sign of the output signal, a highpass filter can be implemented.
In discussing continuous-time modulation in Lecture 13 and discrete-
time modulation in the first part of this lecture, the emphasis is on a carrier
signal that is a complex exponential or sinusoidal signal. Another important
and useful class of carrier signals is periodic pulse trains that are constant for
some fraction of the period and zero for the remainder. In effect, then, either
15-1
Signals and Systems
15-2
in continuous time or discrete time, modulation with such a pulse train con-
sists of extracting time slices of the modulating signal. In data representation
or transmission, this permits, for example, a type of multiplexing referred to
as time division multiplexing since during the "off" part of the pulse train,
time slices from signals in other channels can be inserted. Somewhat amaz-
ingly, the original modulating signal can be recovered exactly after pulse-am-
plitude modulation provided only that the fundamental frequency of the carri-
er pulse train is greater than twice the highest frequency in the modulating
signal. The modulating signal can then theoretically be recovered exactly by
filtering the pulse-amplitude-modulated signal with an ideal lowpass filter.
Furthermore, this ability to exactly reconstruct the original signal is indepen-
dent of the "duty cycle" of the carrier, i.e., it is theoretically possible no matter
how narrow the "on" time of the pulse train is made. If for the continuous-
time case, the "on" time of the carrier pulse train is made arbitrarily small,
with the amplitude increasing proportionately, the carrier then corresponds
to an impulse train. For discrete time, the pulse train with the smallest "on"
time would likewise correspond to a periodic train of impulses or unit sam-
ples. In both cases, then, modulation with the impulse train carrier would cor-
respond to sampling the modulating (input) signal. This leads to an extremely
important concept, referred to as the sampling theorem. The sampling theo-
rem states that a bandlimited signal can be exactly reconstructed from equal-
ly spaced time samples provided that the fundamental frequency of the sam-
pler (i.e., the impulse train carrier) is greater than twice the highest frequency
in the signal to be reconstructed. This fundamental and important result, to be
explored further in Lecture 16, provides a major bridge between continuous-
time and discrete-time signals and systems.
Suggested Reading
Section 7.5, Discrete-Time Amplitude Modulation, pages 473-479
Section 7.4, Pulse Amplitude Modulation and Time-Division Multiplexing,
pages 469-473
Discrete-Time Modulation
15-3
e Cormplex exponentiQL
Carrier
XE"3
Ctn]
. Sin\usoidal C av-rier
. ScroNous rrodulat.ion-.
Asinchronous
,modulakion
_
x E I (e)c(n-e)d 0
CoMP ex exponentla L
C , 3(n -r' en+c)
S' nusod'
Carrier
C [r= Cos(en.+ec)
. S I le
Side baNd
Pulse Carrier
x[n] c[n] X()C(2 ) dO
2i7
X(2)
I III\I
-27r -M 0 M 2r E2
C(92)
2w 2w 21r
27r -2w+ fl 0 _ 27r 2w + 92
Y()
TRANSPARENCY
15.1
Spectra associated
with discrete-time
amplitude modulation
with a complex
exponential carrier.
27r -27r + 2(c2c - EZ ) Se (Oc + EM) 27r
MARKERBOARD
15.1
c c
Signals and Systems
15-4
TRANSPARENCY
15.2
Spectra associated
with discrete-time
amplitude modulation
with a sinusoidal
carrier.
c[n] = cos 2cn
iei cn +
1
e cjn
2 2
X(2)
i I
C (2)
I I I
-T -2c 2c 2
Y(Q)
A1A
_r-2 iT 2r
TRANSPARENCY
15.3
Spectra associated
with demodulation
of an amplitude-
modulated signal with
a sinusoidal carrier.
cos
2n
y[nI w[n]
Y(M)
l
2
W(W)
F
1
2
4
2&2r nT S 22C
Discrete-Time Modulation
15-5
(-1
)
x~n] y[n]
x
Lowpass filter
H (2)
o E2 oE
(-1)
y
1
p [n] XhpE]
x
TRANSPARENCY
15.4
System and spectra for
sinusoidal amplitude
demodulation.
TRANSPARENCY
15.5
The use of amplitude
modulation to
implement highpass
filtering with a
lowpass filter.
Signals and Systems
15-6
TRANSPARENCY
15.6
Spectra associated
with the use of
modulation and
demodulation to
implement highpass
filtering using a
lowpass filter.
X(92)
-2 -f f 2n 7
C(2)
21r
-31r -I
31T 2
Y(Q)
-Ir - 2
xI (E2)
NA it NA
p(t)
TRANSPARENCY
15.7
Continuous-time
amplitude modulation
with a pulse carrier.
x (t) y(t)
x (t)
y(t)
iI I j I I I
Q . 7r
Discrete-Time Modulation
15-7
y[n]
p[n]
p[n]
0 1* * - M N (N + M)
TRANSPARENCY
15.8
Discrete-time
amplitude modulation
with a pulse carrier.
TRANSPARENCY
15.9
Transparencies 15.9
and 15.10 illustrate
spectra associated
with the system of
Transparency 15.7.
Shown here are the
input spectrum and
spectrum of a pulse
carrier.
x[n]
fiT
X(W)
- WM
WM
27rA/T
-W w
Sam see
Signals and Systems
15-8
TRANSPARENCY
15.10
Input spectrum and
output spectrum.
TRANSPARENCY
15.11
Time division
multiplexing using
amplitude modulation
with a pulse carrier.
X(W)
-wM WM
Y(w)
- ~-LWM WM L p
t
X
1
(t)
1
(t)
X
t
x
2
(t)
Y
2
(t)
t
y (t)
Y3 (t) x3 (t)
Discrete-Time Modulation
15-9
p(t)
x(t)
y(t)
x (t)
y(t)
ri
0 t
TRANSPARENCY
15.12
Time waveforms
associated with
amplitude modulation
with a pulse carrier.
X(W)
~WM WM c
Y(W)
A/T
-LO
~WM M
p
TRANSPARENCY
15.13
Spectra associated
with amplitude
modulation with a
pulse carrier.
Signals and Systems
15-10
TRANSPARENCY
15.14
Amplitude modulation
with a pulse carrier
with the pulses chosen
to have unit area.
TRANSPARENCY
15.15
Amplitude modulation
with a pulse carrier in
the limit as the pulse
width approaches zero
and the pulse area
remains unity. This
corresponds to
amplitude modulation
with an impulse train
carrier.
p (t)
x(t)
x(nT)
I I I I I IT
t
r
Discrete-Time Modulation
15-11
X(w)
P(w)
2r
T
2-, - oS W, 2o, o
T
wM(WM , jM
TRANSPARENCY
15.16
Transparencies 15.16
and 15.17 illustrate
spectra associated
with impulse train
modulation. Here, the
frequency of the
modulating
impulse
train is chosen large
enough so that the
individual replications
of the input spectrum
do not overlap.
TRANSPARENCY
15.17
The carrier
fundamental
frequency is chosen
such that the
individual replications
of the input spectrum
overlap.
M C M (A
Signals and Systems
15-12
p(t)=
TRANSPARENCY
x(t)
15.18
Modulation and
demodulation with an
impulse train carrier.
MARKERBOARD
15.2
Samkin Theoreix%
etually space4 Samples
oS x(t) x(Kr) n%=o,ti,tz,...
X( =0&D lI > LO~A
Then. X(t) Aniquelk recoerable
with a loupass filter
MIT OpenCourseWare
http://ocw.mit.edu
Resource: Signals and Systems
Professor Alan V. Oppenheim
The following may not correspond to a particular course on MIT OpenCourseWare, but has been
provided by the author as an individual learning resource.
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.