Lahore University of Management Sciences

School of Science and Engineering, Department of Electrical Engineering

EE564 - Stochastic Systems
Fall 2019
Homework 1

Date Assigned: September 18th , 2019 Date Due: Monday, September 30th , 2019
(In-class)

Instructions

• Some of the problems on this homework may be challenging - that’s why the term “homework
problems”. The objective is to maximize your learning experience, so start early!

• Your solutions must be submitted before the end of class on Monday, Sept 30th. No submissions will
be accepted after the deadline. Do not expect any deadline extensions either.

• While discussions with your peers is strongly encouraged, please keep in mind that the homework is
to be attempted on an individual level. Plagiarism can have dire consequences.

• In order to gain the maximum benefit, try to perform solo brainstorming before any consultations.
The thought process behind searching for the correct approach all on your own will be invaluable.

1. Let A, B be two events. Prove the following:

(a) If A and B are statistically independent, A and Bc must also be statistically independent.
(b) If A and B are mutually exclusive, they cannot be statistically independent.
(c) If A ⊆ B, then P (A) ≤ (B).

2. A maintenance firm has gathered the following information regarding the failure mechanisms for air
conditioning systems:

Evidence of Gas Leakage

YES NO
Evidence of Electrical Failure YES 55 10
NO 32 3
For instance, out of a total of 100 failure events, 55 presented evidence of both Gas Leakage and
Electrical Failure. Similarly, 3 failure events presented neither an evidence of gas leakage, nor of
electrical failure. If this is a representative sample of AC failure, find the probability

(a) That a failure involves a gas leak.

(b) That there is evidence of an electrical failure given that there was a gas leak.
(c) That there is evidence of a gas leak given that there is evidence of an electrical failure.
3. A computer reserves a path in a network for 10 minutes. To extend the reservation, the computer
must successfully send a ref resh message before the expiry time. However, messages are lost with
probability 1/2. Suppose that it takes 10 seconds to send a refresh request and receive an acknowl-
edgement. When should the computer start sending refresh messages in order to have a 99% chance
of successfully extending the reservation time?

4. Consider a best of seven squash fixture between two players A and B. The tournament terminates
as soon as either player wins four games. Let X be the total number of games played (which range
from 4 to 7). Assuming that A and B are equally matched and that the games are independent,
evaluate the probability mass function of X.

5. Consider a supermarket that has only two checkout counters: one of them is a regular counter (open
to all), and the second is an express counter (for less than 10 items). The number of customers
serviced at each counter per minute can be well modelled as Poisson. In particular, If XR and XE
are the random variables corresponding to the number of customers serviced per minute at the regular
and express counter, respectively, their pmfs are given by

λkR e−λR
PXR (k) = , k = 0, 1, 2, 3, . . .
k!
λkE e−λE
PXE (k) = , k = 0, 1, 2, 3, . . .
k!
where λR and λE are the average number of per-minute customer services at the two counters. Let
X = XE + XR be the total number of customers serviced by the supermarket per minute. Determine
the pmf of X. You may assume that the number of customers serviced at the two counters are
independent of each other. Is X also Poisson?

6. Consider two students Khurram and Talha attempting an EE564 exam consisting of 10 MCQ-type
problems. For each problem, a student could either (a) answer it correctly, (b) answer it incorrectly,
or (c) leave it unanswered. Suppose you are given a graded exam that does not have any student
name on it. You observe that the questions answered correctly are Q1, Q3, Q4, Q7, and Q10. The
questions answered incorrectly are Q2 and Q5, while the remaining questions Q6, Q8, and Q9 are
unanswered. Assume that you are given the following probabilistic model.

• The probability that Khurram (a) answers a question correctly is 0.8, (b) answers a question
incorrectly is 0.1, and (c) leaves a question unanswered is 0.1,
• The probability that Talha (a) answers a question correctly is 0.6, (b) answers a question
incorrectly is 0.1, and (c) leaves a question unanswered is 0.3,
• Both students answer all questions independently.

(a) Use probability theory to determine whether the exam belongs to Khurram or Talha.
(b) What is the probability with which your decision in part (a) is wrong?

2
7. (Modified from Starks and Woods) A particular TV model is manufactured in three different plants,
say plants A, B, and C of the same company. Because of the different levels of expertise at the three
plants, the quality of the sets differs from plant to plant. Let X denote the failure time of the sets
in years. The failure time is modeled by an exponential random variable, with means of 5 years, 7
years, and 10 years for plants A, B, and C, respectively. We also know that plant A produces three
times as many sets as plant B, which produces twice as many as plant C.

(a) Suppose you buy a TV set at random. What is the probability the TV set will fail in the first
five years?
(b) Suppose you bought a TV set five years ago. If the TV is still working, what is the probability
that it was manufactured by (i) plant A, (ii) plant B, and (iii) plant C?

8. Let the probability density function of a continuous random variable X be given by

(x−2)2
(
− 8
fX (x) = e −2 ≤ x < b
0 otherwise

for some constant b > −2. Determine the value of b.

Note: You may need to use the “qfunc” and the “qfuncinv” function in Matlab.

9. A Gaussian random variable has a pdf of the form:

r
2 −2(x+1)2
fX (x) = e
π
Compute the following probabilities in terms of the Q-function (with positive arguments) and also
give numerical evaluations:
S
(a) P r (| X + 1 |> 3 | X − 1 |< 2)
T
(b) P r (| X + 1 |> 3 | X − 1 |< 2)

3
10. Willy Wonka’s chocolate factory produces two quality grades of a certain type of candy: Grade A
and Grade B. The two candy grades have exactly the same wrappers and are stored separately in
identical containers (containing a million candies each) that are clearly labeled. Playing mischief,
Wonka’s Oompa Loompas mix the contents of the two containers so that 20% of the Grade A
candies are transferred to the container B (and perfectly mixed) and 20% of the Grade B candies are
transferred to container A (and perfectly mixed). Next, they remove the container labels so that the
two containers are completely indistinguishable from each other. Detecting the mischief, Wonka tasks
Charlie (after firing the Oompa Loompas) to identify the original container labels. Realizing that
the only way to distinguish the two grades is through tasting, he decides on the following strategy:
Randomly pick five candies from one of the two containers and eat them all (lucky Charlie!). If
majority (three or more) of the candies taste Grade A, the container from which the candies were
picked will most likely be the Grade A container; otherwise it will most likely be the Grade B
container (this decision rule, while very intuitive, is optimally derived from an intelligent use of
Bayes’ theorem - try deriving).

(a) What is the probability that Charlie makes a wrong decision? Explicitly outline any and all
assumptions you make and clearly justify the reasoning behind your answer.
(b) Wonka is not too happy with the probability of making a wrong decision associated with tasting
five candies. In order to lower this probability, Charlie decides to taste N candies (where N is
always odd), with a majority-based decision outlined above, i.e., Decide Grade A if (N + 1)/2
or more candies taste like Grade A; otherwise make a Grade B decision. What is the minimum
number of candies Charlie must taste so as to ensure that he makes a wrong decision with
probability of at most 0.01?
Note: You may need to write a Matlab program for this.
(c) Suppose that Charlie had chosen to taste six candies with the following decision rule: (a) If more
than three candies are Grade A, the decision is A; (a) If fewer than three candies are Grade
A, the decision is B; (c) If exactly three candies are Grade A, then flip a fair coin and make a
random decision between A and B. What is the probability of error in this case?
(d) Charlie decides to modify his decision rule of part (b) to the following: (a) If more than three
candies are Grade A, the decision is A; (a) If fewer than three candies are Grade A, the decision
is B; (c) If exactly three candies are Grade A, the decision is always A. What is the probability
of error in this case?