ECE 322 Principles of Communication Systems Rutgers: Spring 2017

Prof. Anand D. Sarwate Due: January 24, 2017

Homework 1: Pre-req check and intro

Objective: The goal of this homework is to refresh your memory about the basics from the prerequisites
to this course. You should be able to understand these problems and solve them based on what you have
learned already. If the meanings of the questions are unclear or the material seems very unfamiliar then
you should (a) check back in with your notes/book/etc. from previous courses and (b) check with yourself
if this is really the course you want to take.

Note: This review homework consists of problems drawn from a related class taught at the University
of Illinois at Urbana-Champaign.

Mathematical notes: Because it’s more “the style” in talking about communication systems, this class
will use Fourier transforms with respect to the frequency variable f (measured in Hertz) rather than the
radian frequency variable ω = 2πf (radians per second). This may or may not be an adjustment for
you depending on when you took the prerequisite signals and systems class. The continuous-time Fourier
transform in (CTFT) pairs are therefore
Z ∞
X(f ) = x(t) exp(−j2πf t) dt, (1)
−∞
Z ∞
x(t) = X(f ) exp(j2πf t) df. (2)
−∞

Parseval’s identity is
Z ∞ Z ∞
|x(t)|2 dt = |X(f )|2 df. (3)
−∞ −∞

Note also that the unit rectangular pulse rect(·) is

(
1, |t| < 21 ,
rect(t) = (4)
0, |t| > 12 ,

and the sinc function sinc(·) is


 sin(πt)
, t 6= 0,
sinc(t) = πt (5)
1, t = 0.

The sinc function has the useful property that sinc(0) = 1 while sinc(n) = 0 for nonzero integers n .

Problem 1. (Fourier transforms and linear systems recap)

(a) Show that the Fourier transform of rect(t/T ) , a pulse of duration T , is T · sinc(f T ) .

1
 (b) What is the Fourier transform of sinc(t/T ) ? Note that sinc(t/T ) passes through 0 at (nonzero)
integer multiples of T .
(c) Use the modulation theorem to show that

2T cos(πf T )
cos(πt/T ) rect(t/T ) ↔ · . (6)
π 1 − (2f T )2
R∞
(d) Find the value of −∞
sinc(t/T ) dt .
(e) What is the output y(t) of an ideal lowpass filter with bandwidth W when its input is x(t) =
sinc(t/T ) ? The answer may depend on the relationship between W and T .

Problem 2. (Exercises on Gaussian random variables) Let XN (0, 1) and Y N (0, 1) denote indepen-
dent standard Gaussian random variables.

(a) What is the joint pdf fX,Y (u, v) of X and Y ?

(b) Sketch the u - v plane and indicate on it the region over which you need to integrate the joint pdf
in order to find P(X 2 + Y 2 > 2α2 ) . Compute P(X 2 + Y 2 > 2α2 ) .
(c) Let Z = X 2 + Y 2 . What is the pdf of Z ?

(d) Express P(|X| > α) in terms of the complementary unit Gaussian CDF function Q(x) , and use
this to write P(|X| > α, |Y | > α) in terms of Q(x) . (Remember commas mean intersections).
(e) On your sketch of part (b), show the region over which you must integrate the joint pdf to find
P(|X| > α, |Y | > α) . Use your sketch to prove the following result:

P(|X| > α, |Y | > α) < P(X 2 + Y 2 > 2α2 ) for α > 0. (7)

1
(f) Show that inequality of part (e) implies that Q(x) < 2 exp(−x2 /2) for x > 0 .

(g) On your sketch of parts (b) and (d), show the region over which you must integrate to find P(|X| <
α, |Y | < α) , and prove that

P(X 2 + Y 2 ≤ α2 ) < P(|X| < α, |Y | < α) < P(X 2 + Y 2 ≤ 2α2 ). (8)

Use these inequalities to deduce the lower bound Q(x) > 14 exp(−x2 ) for x > 0 . Note that at
x = 0 , equality holds in the upper bound of part (f) but not in this lower bound.

Problem 3. (Connecting probability with linear systems) A signal x(t) = exp(−πt2 ), −∞ < t < ∞ , is
the input to an ideal low-pass filter of bandwidth 1 Hz, that is, H(f ) = rect(f /2) .
Let y(t) denote the output of the filter. Find the numerical value of y(0) . Hint: X(f ) = exp(−πf 2 ), −∞ <
f < ∞.

Problem 4. (Variance of the product of two independent random variables) If X and Y are independent
random variables with finite means E[X] and E[Y ] respectively and finite variances Var(X) and Var(Y )
respectively, find Var(XY ) in terms of these four parameters.

2
Problem 5. (Analyzing communication using probability) A message is sent simultaneously over two
separate parallel channels connecting a transmitter to a receiver. However, because of delays in the
channels, the message sent via channel 1 arrives X minutes after the transmission while the message
sent via channel 2 arrives Y minutes after the transmission. Assume that X and Y are independent
random variables uniformly distributed on [0, 1] . The message is said to have been received as soon as it
first arrives (doesn’t matter on which channel it arrives) and to have been verified when it arrives (some
time later) on the other channel.

(a) What is the probability that the message is verified less than fifteen seconds after it was trans-
mitted? What is the probability that the message is received less than fifteen seconds after it was
transmitted?

(b) What is the average delay until the message is verified ? What is the average delay until the message
is received ?
(c) What is the probability that the message is verified less than fifteen seconds after it is received ?
(d) What is the probability that the message is verified less than fifteen seconds after it arrives on
channel 1? Be careful here: the arrival on channel 1 might be the first arrival of the message, or it
might be the second arrival, i.e., be the verifier.