Scribd
Upload
54 views20 pages

Lecture 5 ParametricMethod

This document discusses parametric methods for spectrum estimation, which incorporate available information about the signal generation process to estimate the spectrum more accurately than non-parametric methods. It describes selecting an appropriate model such as autoregressive (AR), moving average (MA), or autoregressive moving average (ARMA) and estimating the model parameters from the data. The autoregressive power spectrum is presented based on the AR coefficients, and methods for estimating the coefficients like Yule-Walker, covariance, and modified covariance are outlined. Model order selection and examples are also provided.

Uploaded by

con.chim.non.s17
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
54 views20 pages

Lecture 5 ParametricMethod

This document discusses parametric methods for spectrum estimation, which incorporate available information about the signal generation process to estimate the spectrum more accurately than non-parametric methods. It describes selecting an appropriate model such as autoregressive (AR), moving average (MA), or autoregressive moving average (ARMA) and estimating the model parameters from the data. The autoregressive power spectrum is presented based on the AR coefficients, and methods for estimating the coefficients like Yule-Walker, covariance, and modified covariance are outlined. Model order selection and examples are also provided.

Uploaded by

con.chim.non.s17
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/ 20

Da Nang University of Science and Technology

Spectrum estimation
Parametric method
Ho Phuoc Tien
(hptien@yahoo.com)
Overview
 Non-parametric methods: not incorporate information
about the process
 Parametric methods incorporate available information
about the process can help to estimate spectrum more
accurately and with higher resolution
 The steps of parametric methods:
 Select an appropriate model (AR, MA, ARMA…)
 Estimate the model parameters from given data
 Incorporate the estimated parameters to the parametric form of
the spectrum 2
Overview
 Parametric methods can significantly improve spectrum
resolution
 Important: Appropriate model must be chosen !

(Generally, the parametric approach should be examined after signal


modeling, see more in Hayes’s book)

3
Autoregressive (AR) spectrum
estimation
 An AR process can be represented as the output of an
all-poles filter that is driven by unit variance white noise
 The power spectrum of a p-order AR process is

4
AutoRegressive (AR) spectrum
estimation
 If the coefficients b(0) and ap(k) are estimated, the
power spectrum may be estimated as follows:

 The accuracy of the estimated spectrum depends on


the accuracy of the estimated parameters and the
appropriateness of the chosen model
5
AR model estimation

 There are several methods to estimate AR models


 The autocorrelation method
 The covariance method
 The modified covariance method
 The Burg algorithm

6
The autocorrelation method
 Find out the coefficients ap(k) by solving the normal
equations (Yule-Walker method)

where
b(0) is computed as

7
The covariance method
 Find out the coefficients ap(k) by solving a set of linear
equations

where

8
The modified covariance
method
 Like the covariance method, but replace the estimation
of the autocorrelation coefficients

Forward and backward method

9
Selecting the model order

 Small order p: spectrum is smooth and hence, poor


resolution
 Large order p: spurious peaks (spectral line splitting)
 Criterion: incorporating a penalty function that
inscreases with order p

10
Example

11
Example

12
Example

13
Moving Average spectrum
estimation
 A moving average process may be generated by
filtering unit variance white noise with a FIR filter

 The power spectrum

14
Moving Average spectrum
estimation
 The power spectrum can also be represented in term of
autocorrelations

where

15
Moving Average spectrum
estimation
 Two methods for the spectrum estimation
 Using an estimate of autocorrelation (Blackman-Tukey)

 Using the estimated parameters b(k) (Durbin’s method)

16
Example

17
Example

18
Autoregressive Moving Average
spectrum estimation
 The power spectrum of an ARMA process is given as

(consider an ARMA filtering model)

19
Autoregressive Moving Average
spectrum estimation
 The spectrum estimation

(the coefficients a(k) can be first estimated, and then the


coefficients b(k))

20

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy