Scribd
Upload
55 views1 page

HW 4

This document outlines the homework problems for ELEC 531. It includes 6 problems related to maximum likelihood estimation, efficient estimators, the Cramer-Rao lower bound, and the bias-variance decomposition for vector parameters. Problem 1 asks to show the MLE of θ is efficient given the Fisher information matrix. Problem 2 asks for the MVUE and efficiency when data follows an autoregressive model with Gaussian noise. Problem 3 finds the CRLB for estimating r from observations with Gaussian noise.

Uploaded by

Hira Majid
Copyright
© Attribution Non-Commercial (BY-NC)
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
55 views1 page

HW 4

This document outlines the homework problems for ELEC 531. It includes 6 problems related to maximum likelihood estimation, efficient estimators, the Cramer-Rao lower bound, and the bias-variance decomposition for vector parameters. Problem 1 asks to show the MLE of θ is efficient given the Fisher information matrix. Problem 2 asks for the MVUE and efficiency when data follows an autoregressive model with Gaussian noise. Problem 3 finds the CRLB for estimating r from observations with Gaussian noise.

Uploaded by

Hira Majid
Copyright
© Attribution Non-Commercial (BY-NC)
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/ 1

ELEC 531 HW 4

Due Wednesday, Sept. 23, beginning of class

1. (5 points) Consider X N (H, ) where H and are known. Assume X is N 1 and is p 1. Vector calculus can be used to show that the Fisher information matrix is I() = HT 1 H. Show that the MLE of is ecient. 2. (5 points) The data x(n) = Arn + w(n), n = 0, . . . , N 1, are observed where w(n) is Gaussian white noise with variance 2 and r > 0 is known. What is the MVUE? Is it ecient? What happens to the variance as N for various values of r? 3. (7 points) If x(n) = rn + w(n), n = 0, . . . , N 1, are observed, where w(n) is Gaussian white noise with variance 2 (known) and r > 0 is to be estimated, nd the CRLB. Does an ecient estimator exist and if so nd its variance? 4. (8 points) Consider the problem of line tting to noisy observations: xn = A + Bn + wn , n = 1, . . . , N

where wn N (0, 2 ). In this case, the parameter is the vector = [A B]T . a. Determine the Fisher Information Matrix for and the CRLBs for A and B. b. Compare the CRLB of A in this case, with the CRLB for A when B is known. 5. (5 points) Extend the bias-variance decomposition of the MSE to a vector parameter. 6. (Challenge-Optional) Consider the setting of problem 1.a on HW 3. Find the CRLB. Does an ecient estimator exist? HINT 1: If Z 2 , then E[Z] = and Var(Z) = 2. HINT 2: The sample mean and sample variance of IID Gaussian data are independent. HINT 3: If S 2 = 1/(N 1) (xi x)2 is the sample variance, then (N 1)S 2 / 2 2 1 . N

You might also like