18 Discrete-Time

Processing of
Continuous-Time
Signals
Recommended
Problems
P18.1
Considerthe systemin Figure P18.1-1 for discrete-time processing ofa continuous-
time signalusing sampling period T,where theC/D operation isas shown in Figure
P18.1-2 and theD/C operation isas shown in Figure P18.1-3.
Xc(t) C/D xG(9) ----- D/C yPc (t)
FigureP18.1-1
y[n]
Discrete-time sequence
to impulse train with
spacing T
-
T
H(w)
T
T
Co
1
(t)
FigureP18.1-3
P18-1
SignalsandSystems
P18-2
The filter G(Q) isthe lowpass filter shown inFigure P18.1-4.
G(Gi)
IT iT
Figure P18.1-4
The Fouriertransform ofxc(t),Xc(w) isgiven in Figure P18.1-5.
Xc(w)
1
-2nTX 4kHz 27X 4kHz
Figure P18.1-5
The sampling frequency is 8 kHz. Sketch accurately thefollowing transforms.
(a)X,(w)
(b) X(Q)
(c) Y(Q)
(d) Ye(w)
P18.2
Considerthe continuous-time frequency response in Figure P18.2.
H(o)
-500T 500n7r
Figure P18.2
We wantto implement thiscontinuous-time filterusing discrete-time processing.
(a) What is themaximumvalue of thesampling period Trequired?
Discrete-Time Processing ofContinuous-Time Signals / Problems
P18-3
(b) What isthe required discrete-time filter G(Q) forTfound inpart(a)?
(c) Sketch thetotalsystem.
P18.3
Thesystem inFigure P18.3 issimilartothat demonstrated inthelecture. Note that,
as inthe lecture, thereis no anti-aliasing filter.
xc(t) C|D x[]YLj D|C Noy(t)
20000 20000
Figure P18.3
Forthefollowing signals, drawXc(w), X(Q), andYc(w).
(a) xc(t) = cos(2r - 5000t)
(b) xc(t) = cos(2xr 27000t)
(c) xc(t) = cos(2r - 17000t)
P18.4
Suppose we want to design a variable-bandwidth, continuous-time filterusing the
structure inFigure P18.4-1.
Find, interms ofwc, the valueofthesamplingperiod To and thecorrespondingvalue
co, such that the total continuous-time filter has the frequency response shown in
Figure P18.4-2.
Figure P18.4-2
Signalsand Systems
P18-4
P18.5
Considerthe systemin Figure P18.5-1.
Let H(Q) be as giveninFigure P18.5-2 andX(co) as given inFigure P18.5-3.
x(t) C/D H(Q) y-[nw]
T= To
Figure P18.5-1
--
H(Q)
1
IT
3
Figure P18.5-2
it
-3
X(W)
7T 7T
To
To
Figure P18.5-3
(a) SketchX(Q) andY(Q).
(b) Suppose we replace the system in Figure P18.5-1 by the
P18.5-4. Find G(w) suchthat y[n] = z[n].
system in Figure
x(t) G(w) C/D
T
=
To
Figure P18.5-4
P z [n]
Discrete-Time Processing ofContinuous-Time Signals / Problems
P18-5
Optional
Problems
P18.6
Suppose we are given the systemin Figure P18.6-1.
yIn] Convert
xIn]
y
K
x / to y(t)
-1007r 1007r T=To train -100ir 100T
T= To
FigureP18.6-1
(a)Find the appropriate values of the sampling period To to avoid aliasing. Also
find the proper value forK so that the overall system has a gain of unity at
w = 0 (i.e., no overall dc gain).
(b) Suppose To is halved, but the anti-aliasing and reconstruction filters are not
modified.
(i) IfX(w) is as given in Figure P18.6-2, find Y(Q).
X(W)
-1007r 100ff
FigureP18.6-2
(ii) IfY(Q) is as giveninFigure P18.6-3, find Y(w).
Y(92)
1-
IT I
3 3
FigureP18.6-3
SignalsandSystems
P18-6
P18.7
Figure P18.7 shows a system that processes continuous-time signalsusing adigital
filter.The digital filterh[n]is linearand causalwith difference equation
y[n] = -y[n - 1]+x[n]
For inputsignals that are bandlimited so thatXe(w) = 0 for I l > ir/T,thesystem
is equivalent to a continuous-time LTI system. Determine the frequency response
He(w) oftheequivalent overall system with inputxc(t) and outputyc(t).
x,(Conversion of x[]yn
Conversion ofa y I(t) l owpass
xc t) X impulse train x-n h _n] sequence y t) fequtenr / yt)
sequence impulsetrain and gain
p(t) = I (t- nT)
Figure P18.7
P18.8
Figure P18.8-1 depicts a system for which the input and output are discrete-time
signals. The discrete-time inputx[n]is convertedtoacontinuous-time impulsetrain
x,(t).The continuous-time signalx,(t) is then filtered by anLTI system toproduce
the output yc(t), which is then converted to the discrete-time signal y[n]. The LTI
system with inputxc(t) and outputyc(t) is causal and is characterized by thelinear
constant-coefficient difference equation
dt
2
+4
dt
+ 3yc(t) = x(t) C
I 5(t-nT)
xP(t = xx[n] 6It - nT)
yp(t) = y,(t) 6(t - nT)
y[n] = y,(nT)
FigureP18.8-1
Discrete-Time Processing ofContinuous-Time Signals / Problems
P18-7
The overall system is equivalent to acausal discrete-time LTI system,as indicated
in Figure P18.8-2. Determine the frequency response H(Q) of the equivalent LTI
system.
h[n]; H(92)
x[n] 1 equivalent P y[n]
LTI system
FigureP18.8-2
P18.9
We wish to design a continuous-time sinusoidal signal generator that is capable of
producing sinusoidal signalsat any frequency satisfying wi : W5 W
2
, wherew,and
W2 are positive numbers.
Our design is totakethe following form.We have storeda discrete-time cosine
wave ofperiod N; thatis, we havestoredx[O], . . ., x[N - 1], where
x[k] = cos (271.k
N)
Every Tseconds we output animpulse weighted by a value ofx[k], where we pro-
ceed through thevalues ofk = 0, 1, ... ,N - 1in acyclic fashion. Thatis,
y,(t) = ( x[k modulo N]S(t - kT)
k= -
(ttkT) = cos N -
(a)Showthatby adjusting Twecan adjustthefrequency ofthecosinesignalbeing
sampled. Specifically, showthat
y,(t) = (cos wat) ('btt- kT),
k= -o
wherewo = 21r/NT.Determinearange ofvalues forTsothaty,(t)canrepresent
samples ofacosine signal with afrequency that is variableover the fullrange
(b) Sketch Y,(w).
The overall system forgenerating acontinuous-time sinusoid is depicted in Figure
P18.9-1. H(w) is an ideal lowpass filterwithunity gain in itspassband:
H(w) = othwis
0, otherwise
Signalsand Systems
P18-8
x[O]
e T yp(t) H(w) y(t)
x[N -1] D
Figure P18.9-1
The parameter we is to bedetermined such thaty(t) is acontinuous-time cosine
signal in the desired frequency band.
(c) Consider anyvalue ofTin therange determined inpart(a). Determine themin-
imum value ofN and some value forw, such that y(t) is acosine signal in the
range wi : O o0
2
.
(d) The amplitude of y(t) will varydepending onthevalue ofwchosen between wi
and W
2
.Determine the amplitude ofy(t) as afunction ofwand as afunction of
N.
P18.10
In many practical situations,a signal is recorded in thepresence of an echo, which
we would like to remove by appropriate processing. Forexample, Figure P18.10-1
illustrates a system in which a receiver receives simultaneously a signal x(t) and
anechorepresented by anattenuated delayedreplication ofx(t).Thus,thereceiver
output is s(t) = x(t) + ax(t - TO), where jal < 1. The receiver output is to be
processed torecoverx(t) byfirst converting toasequence andusing anappropriate
digital filterh[n]as indicated inFigure P18.10-2.
/( X
0at
- TO)
x(t)
Receiver output
s(t)=x(t) +ax(t- TO)
Figure P18.10-1
Discrete-Time Processing ofContinuous-Time Signals / Problems
P18-9
Ideal lowpass
filter
se(tM x(t) + ax(t- TO)
H
SP(t) Conversion of s[n] y[n] Conversion of y,(t) A Y, t)
X impulseatrain - ---0 h[n] sequence ---- 0
sequence impulse train 7
T T
p(t)= I (t - kT)
k =-
FigureP18.10-2
Assumethatx(t)is bandlimited,i.e., X(w) = 0 for IwI > wm, and that al < 1.
(a)IfTo < lr/M and the sampling period is taken equal to To (i.e., T = TO), deter-
mine the difference equation for the digital filter h[n] so that yc(t) is propor-
tional tox(t).
(b)With the assumptions of part (a), specify thegainA ofthe ideal lowpass filter
so that yc(t) = x(t).
(c) Now suppose that 2r/wm < To <
2
7r/WM. Determine a choice for the sampling
period T,thelowpass filter gainA, and the frequency response forthe digital
filterh[n]suchthatyc(t) is equal tox(t).
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.