Scribd
Upload
99 views

Ps 2

This document contains the problems for Problem Set 2 of the course 6.432 Stochastic Processes, Detection and Estimation at MIT. It includes 8 problems related to probability density functions, Gaussian random variables, linear transformations, approximation of random vectors, binary hypothesis testing, and binary communication systems. Students are asked to find probability distributions, decision rules, error probabilities, and solve other statistical signal processing problems. The problem set is due on February 19, 2004.

Uploaded by

Anonymous bOvvH4
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
99 views

Ps 2

This document contains the problems for Problem Set 2 of the course 6.432 Stochastic Processes, Detection and Estimation at MIT. It includes 8 problems related to probability density functions, Gaussian random variables, linear transformations, approximation of random vectors, binary hypothesis testing, and binary communication systems. Students are asked to find probability distributions, decision rules, error probabilities, and solve other statistical signal processing problems. The problem set is due on February 19, 2004.

Uploaded by

Anonymous bOvvH4
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

Massachusetts Institute of Technology

Department of Electrical Engineering and Computer Science


6.432 Stochastic Processes, Detection and Estimation
Problem Set 2
Spring 2004
Issued: Tuesday, February 10, 2004

Due: Thursday, February 19, 2004

Reading: For this problem set: Chapter 2, through Section 2.5.1


Next: Chapter 2, Chapter 3 through Section 3.2.4
Problem 2.1
Let x and y be statistically independent random variables with probability density
functions
1
1
px (x) = (x + 1) + (x 1)
2
2
and

y2
1
exp 2 ,
py (y) =
2
2 2
and let z = x + y , and w = xy .
(a) Find pz (z), the probability density function for z .
(b) Find the conditional probability density functions pz |x (z | x = 1) and pz |x (z |x =
1).
(c) Find the mean values x = mx and y = my , the variances y2 and w2 , and the
covariance yw . Are y and w uncorrelated random variables? Are y and w
statistically independent random variables?
(d) Are y and w Gaussian random variables? Are they jointly Gaussian? Explain.
Problem 2.2
Let x1 and x2 be zero-mean jointly Gaussian random variables with covariance matrix

34 12
x =
.
12 41
(a) Verify that x is a valid covariance matrix.
(b) Find the marginal probability density for x1 . Find the probability density for
y = 2x1 + x2 .

(c) Find a linear transformation dening two new variables


x1
x1
= P

x2
x2
so that x1 and x
2 are statistically independent and so that
PPT = I,
where I is the 2 2 identity matrix.
Problem 2.3 (practice)
Let x be an N -dimensional zero-mean random vector whose covariance matrix has
eigenvalues
1 > 2 > > N ,
and corresponding eigenvectors
1 , 2 , , N .
Suppose we wish to approximate x as a scalar random variable b times a deterministic
N -dimensional vector a ( i.e., ba ). This problem concerns nding the best b and
a.
(a) Let

bopt = arg min x ba,

bR

where is the Euclidean norm for RN , i.e., z2 = zT z. Find an explicit


expression for bopt in terms of x and a. Note that bopt is a random variable.
(b) Let bopt be as dened in part (a). Now dene
aopt = arg min E[x bopt a2 ].
aRN

Show that

aopt = arg max


a

var [aT x]

.
aT a

(c) Determine

var [aT x]

a
aT a

and indicate the value(s) of a for which the maximum is achieved.

max

(d) Repeat part (c) when we impose the constraint that

a i (i.e., aT i = 0) for i = 1, 2, , k 1,

for some k 2. Again, indicate the value(s) of a for which the maximum is
achieved.
Problem 2.4
Suppose x and y are random variables. Their joint density, depicted below, is con
stant in the shaded area and 0 elsewhere.

p x,y(x ,y )

-2

-1

2
x

-1
-2

(a) Let H = H0 when x 0, and let H = H1 when x > 0. Determine P0 =


Pr [H = H0 ] and P1 = Pr [H = H1 ], and make fully labelled sketches of py |H (y |H0 )
and py |H (y |H1 ).

(b) Construct a rule H(y)


for deciding between H0 and H1 given an observation
y = y that minimizes the probability of error. Specify for which values of y
your rule chooses H1 , and for which values it chooses H0 . That is, determine
the regions

Z0 = {y | H(y)
= H0 }

Z1 = {y | H(y)
= H1 }.
What is the resulting probability of error?
3

Problem 2.5
In the binary communication system shown in Fig. 5-1, messages m = 0 and m = 1
occur with a priori probabilities 1/4 and 3/4 respectively. Suppose that we observe
r,
r = n + m,
where n is a continuous valued random variable with the pdf shown in Fig. 5-2. The
random variable n is statistically independent of whether message m = 0 or m = 1
occurs.
source

m
m=0 or 1
message

^
m

receiver

received
signal

^
m=0
or 1

Figure 5-1

pn (n)
2/3

3/4

-3/4

Figure 5-2
(a) Find the minimum probability of error detector, and compute the associated
= m].
probability of error, Pr [m
(b) (practice) Suppose that the receiver does not know the a priori probabilities,
so it decides to use a maximum likelihood (ML) detector. Find the ML detector
and the associated probability of error. Is the ML detector unique? Justify your
answer. If your answer is no, nd a dierent ML receiver and the associated
probability of error.
Problem 2.6
Consider the binary discrete-time communication system shown in Fig. 6-1. Assume
that m = 0 and m = 1 are equally likely to occur. Under hypothesis Hm (m = 0, 1),
the received signal r [n] is given by
r [n] = ym [n] + w [n] = sm [n] h[n] + w [n],

for all n,

where denotes convolution, and where the w [n]s are zero-mean statistically independent Gaussian random variables with variance 2 . The two signals, s0 [n] and
4

s1 [n] are shown in Figs. 6-2 and 6-3, respectively. The parameter in Fig. 6-3 satises
0 1. The impulse response of the linear time invariant lter h[n] is shown in
Fig. 6-4.
sm [n]

ym [n]

h[n]

r[n]

m = 0 or 1

w[n]

Figure 6-1
1

s0 [n]

Figure 6-2

s1 [n]

Figure 6-3

h[n]

Figure 6-4
We wish to obtain a rule for making a decision about which hypothesis was used,
based on the observed sequence r [n].
(a) Which samples of r [n] provide information in making the decision? Justify your
reasoning.
(b) Find the minimum probability of error decision rule, based on observation of
r [n]. Simplify your processor as much as possible to minimize computation.

(c) Obtain an expression for Pr [] in terms of and Q (), where Q () is dened as



1
2
Q (x) =
e /2 d.
2 x
(d) Find the values of (where 0 1) that minimize Pr [] for the detector
obtained in part (b).
Problem 2.7 (practice)
Binary frequency shift keying (FSK) on a Rayleigh fading channel can be modeled in
terms of a 4-dimensional observation vector

y1
y 2

y=
y 3 ,
y4
which is given by y = x + z, where z N (0, 2 I) and z is independent of x. Under
H = H0 ,

x1
x 2

x=
0,
0
whereas under H = H1 ,


0
0

x=
x 3 .
x4

The xi are independent, identically-distributed (IID) N (0, 2 ) random variables. Fur


thermore, the two hypotheses are equally likely.
(a) Show that the ML receiver calculates v0 = y12 + y22 and v1 = y3
2 + y42 , and chooses
= H0 if v0 v1 , and chooses H

= H1 otherwise.
H
(b) Find pv0 |H (v0 | H0 ) and pv1 |H (v1 |H0 ).
(c) Let u = v0 v1 , and nd pu|H (u|H0 ).
(d) Show that

Pr [|H = H0 ] =

,
2
2+ 2

=
where denotes the error event H
H . Explain why this is also the uncondi
tional probability of an incorrect decision.
6

Problem 2.8
In the binary communications system shown below, messages m = 0 and m = 1 occur
with a priori probabilities 41 and 43 respectively. The random variable n is independent
from m and takes on the values -1, 0, 1 with probabilities 81 , 43 , 81 respectively
source

m
m=0 or 1
message

receiver

received
signal

^
m
^
m=0
or 1

Figure 8-1
Find the receiver which achieves the maximum probability of correct decision.
Compute the probability of error for this receiver.

You might also like