MLE - Detection and Estimation(Spring 2009 MLE Maximum...

This preview shows pages 1–10. Sign up to view the full content.

Detection and Estimation (Spring , 2009) MLE NCTU EE P.1 Maximum Likelihood Estimation ~ MLE 9 A “turn-the-crank” procedure 9 The most popular approach to obtaining practical estimators 9 It is approximately the MVU estimator (actually approximately efficient) for large enough data records Example: DC Levels in WGN x[n] = A + w[n] n = 0, 1, …, N-1 A: unknown, A > 0 w[n] WGN with unknown variance A

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Detection & Estimation (Spring, 2009) MLE NCTU EE P.2 RBLS Approach Sufficient statistic: Assume T(x) is complete. We need to find a function g such that But How to choose g is not obvious! Alternatively, since x[0] is an unbiased estimator, the MVU estimator would take the form too difficult to get!!
Detection & Estimation (Spring, 2009) MLE NCTU EE P.3 We propose an estimator that is approximately optimal. As N Asymptotically unbiased. Asymptotically efficient Consider The estimator is biased. However, as N The estimator is said to be a consistent estimator . where To find the mean and variance of A ˆ as N , we linearize the estimator about u 0 = E(U) = A + A 2 .

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Detection & Estimation (Spring, 2009) MLE NCTU EE P.4 Asymptotically unbiased. var(x 2 [n]) = 4A 3 +2A 2 , CRLB!! A ˆ is asymptotically efficient!! Moreover, by the central limit theorem, is Gaussian as N . A ˆ is a linear function of A ˆ is Gaussian.
Detection & Estimation (Spring, 2009) MLE NCTU EE P.5 ~ Finding the MLE The MLE for a scalar parameter is defined to be the value of θ that maximizes p( x ; θ ) for x fixed. Example: DC Level in WGN x[n] = A + w[n] n = 0, 1, …, N-1 A: unknown, A > 0 w[n] WGN with unknown variance A Since A > 0, we choose

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Detection & Estimation (Spring, 2009) MLE NCTU EE P.6 Sometimes, the MLE yields an efficient estimator for finite data records. Example: DC Level in WGN x[n] = A + w[n], n = 0, 1, …, N-1 w[n]: WGN with known variance σ 2 Remark: If an efficient estimator exists, the maximum likelihood procedure will produce it!!
Detection & Estimation (Spring, 2009) MLE NCTU EE P.7 ~ Properties of the MLE a asymptotically Example: DC Level in WGN x[n] = A + w[n] n = 0, 1, …, N-1 A: unknown, A > 0 w[n] WGN with unknown variance A How large must N be for asymptotic results to apply? Use Monte Carlo computer simulation. Remark: In practice, an analytical expression for the pdf of the MLE is difficult to derive and a computer simulation is usually required.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Detection & Estimation (Spring, 2009) MLE NCTU EE P.8 Remark: Usually, N does not have to be very large. Example: Phase estimation Assume we wish to estimate the phase φ of a sinusoid embedded in WGN.
MLE NCTU EE P.9 To find the MLE, we maximize or minimize For f 0 not near 0 or 1/2, Remark: The MLE is a function of the sufficient statistics. Explanation:

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 07/21/2009 for the course EE IEE5703 taught by Professor Sheng-jyhwang during the Spring '09 term at National Chiao Tung University.

Page1 / 34

MLE - Detection and Estimation(Spring 2009 MLE Maximum...

This preview shows document pages 1 - 10. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online