Scribd
Upload
20 views6 pages

EEE539 Spring2024 MT Sol

Uploaded by

selinturhan22
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
20 views6 pages

EEE539 Spring2024 MT Sol

Uploaded by

selinturhan22
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/ 6

1

Bilkent University

EEE-539 DETECTION AND ESTIMATION THEORY

Spring 2024 - Midterm

March 19, 2024

Duration: 120 minutes

Surname
Name
ID #
Signature

Question-1 (25 pts)


Question-2 (25 pts)
Question-3 (25 pts)
Question-4 (25 pts)
TOTAL (100 pts)
2

1) Consider a binary hypothesis-testing problem in which observation Y has the following probability
density functions under the hypotheses:
( (
1 − |y|, if |y| ≤ 1 0.25(2 − |y|), if |y| ≤ 2
H0 : p0 (y) = H1 : p1 (y) =
0, if |y| > 1 0, if |y| > 2
Assume that the cost assignment is given as follows: C00 = C11 = 0, C10 = N , and C01 = 2N , where N
is a positive number. Let π1 denote the prior probability of H1 , which is a known number in (0, 1).
a) Derive the Bayes rule as a function of π1 in the simplest forms for all possible values of π1 ∈ (0, 1).
b) Calculate the minimum Bayes risk in the simplest form assuming that π1 = 0.5.
2) Consider a binary hypothesis-testing problem in which the observations are modeled as

H0 : Yk = k + Nk , k = 1, 2
H1 : Yk = 2k + Nk , k = 1, 2
   
0 2 1
where N1 and N2 are jointly (multivariate) Gaussian with mean vector and covariance matrix .
0 1 3
a) Find an α-level Neyman-Pearson test and calculate its detection probability in the simplest forms.
b) Considering uniform cost assignment (UCA), obtain the minimax decision rule in the simplest form
(no need to calculate the minimax risk).
3) Consider a binary hypothesis-testing problem in which observation Y has the following probability
density functions under the hypotheses:

H0 : p0 (y) = 0.5 e−|y| H1 : p1 (y) = 0.5 e−|y−θ|

where y ∈ R and θ > 0.


a) For α ∈ (0, 0.5), obtain the α-level locally most powerful (LMP) test and calculate its power function
in the simplest forms.
b) Derive the generalized likelihood ratio test (GLRT) for this problem and simplify it as much as
possible (no need to specify the threshold in this case).
4) This question has two independent parts, a) and b).
a) Consider a simple binary hypothesis-testing problem between hypotheses H0 and H1 based on the
observation of a realization of a random vector Y . For a given realization y of the random vector Y , the
likelihood ratio is defined as p1 (y)
L(y) =
p0 (y)
where p0 (y) and p1 (y) denote the joint probability density function of Y under H0 and under H1 ,
respectively. (You can assume continuous random variables.)
i) For any integrable function g : R → R, prove that E{(L(Y ))n+1 g(Y ) | H0 } = E{(L(Y ))n g(Y ) | H1 },
where n is a positive integer.
ii) Express Var{L(Y ) | H0 } in terms of E{L(Y ) | H1 }.
b) Let X1 , . . . , Xn represent independent and identically distributed observations, each of which has
uniform density over the open set of (0, θ) for θ > 0. For the following binary hypothesis-testing problem

H0 : θ = 2
H1 : θ 6= 2

find an uniformly most powerful test (UMP) of level α for a generic α ∈ (0, 1).

You might also like