ebook img

DTIC ADA249623: Maximum Likelihood Estimation of Fractional Brownian Motion and Markov Noise Parameters PDF

140 Pages·4.6 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview DTIC ADA249623: Maximum Likelihood Estimation of Fractional Brownian Motion and Markov Noise Parameters

ADFA249 623 I I I i I I PAGE form Approved MB No. 0704-0188 S jthe, ¢OPl.c n r" c rff brAtrI' , C-t het 'q SnUrge gde stLoA,rP ' aOO .o. . n/e d. .Ii .Q,A U0I:. n ", OurO,p P. .t.: ,%ds~h,r.iO.n,a toOnie ta cc't,,Oartremrs ~tOns >e,rev-tacv . CDO'J iaf me c't orartefo oerM o - jrtINer, iastbOo umnr dei O Sr %tt aelros B &i Rnetp .Athje,, 12 d1.e5 l cJe fO~tr tTWh,-' 1. AGENCY USE ONLY (Leave blank; 2. REP1O9R9T1 DATE 3. REPOTRHTE TsYI PsEA NAVNDO ODART ES CXOIVERED 4. TITLE AND SUBTITLE 5. FUNDING NUMBERS Maximum Likelihood Estimation of Fractional Brownian Motion and Markov Noise Parameters 6. AUTHOR(S) Matthew E. Skeen, 2d Lt 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT NUMBER AFIT Student Attending: Massachusettes Institute of AFIT/CI/CIA- 91-132 Technology 9. SPONSORING /MONITORING AGENCY NAME(S) AND ADDRESS(b T ICAGE1N0. SCPOYNSO RING / MONITORING REPORT NUMBER AFIT/C Wright-Patterson AFB OH 45433-6583 EIECTE 11 1S.PL M N AY N T S C _MAY 7 19921 , 12a, DISTRIBUTION/ AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE Approved for Public Release IAW 190-1 Distributed Unlimited ERNEST A. HAYGOOD, Captain, USAF Executive Officer 13. ABSTRACT (Maximum2 00 words) 14. SUBJECT TERMS 15. NUMBER OF PAGES 138 16 , ,CE CODE 17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT OF REPORT OF THIS PAGE OF ABSTRACT NSN 7540-01-280-5500 Stardard Porm 298 (Rev 2-89) Title: Maximum Likelihood Estimation of Fractional Brownian Motion and Markov Noise Parameters Author: Matthew Edward Skeen Military Rank: Second Lieutenant Service Branch: Air Force Date: 1991 Number of Pages: 138 Degree Awarded: Master of Science in Aeronautics and Astronautics Institution: Massachusettes Institute of Technology ABSTRACT Maximum likelihood estimation and power spectral density analysis are developed as tools for the analysis of stochastic processes. Some useful results from the theory of Markov stochastic processes are then presented followed by the introduction of fractional Brownian motion and fractional Gaussian noise as non- Markov models for systems with power spectral density proportional to f0, where -3 < P < -1 and -1 < P < 1 over all frequencies. Maximum likelihood system identification is applied to estimating the unknown parameters in a Markov model which approximates fractional Brownian motion. The algorithm runs a Kalman filter on states and a maximum likelihood estimator on parameters. Results are presented from estimating trend, white noise, random walk, and exponentially correlated noise parameters from fits to simulated and real test data. Maximum likelihood estimation is applied to the batch estimation of parameters in the non-Markov fractional Brownian motion model. New in this thesis is the use of partial derivatives to minimize the resulting likelihood function, and the capability to estimate the unknown parameters of additional trend and Markov noise processes. Results are presented from fits to computer simulated sample paths. PRIMARY REFERENCES Lundahl, Torbjijrn, William J. Ohley, Steven M. Kay and Robert Siffert, "Fractional Brownian Motion: A Maximum Likelihood Estimator and Its Application to Image Texture," IEEE Transactions on Medical Imaging, Vol. MI-5, No. 3, September 1986, pp. 152-161. Mandelbrot, Benoit B. and John W. Van Ness, "Fractional Brownian Motions, Fractional Noises and Applications," SIAM Review, Vol. 10, 1969, pp. 422-437. Sandel, Nils R. Jr. and Khaled I. Yared, "Maximum Likelihood Identification of State Space Models for Linear Dynamic Systems," M.I.T. Electronic Systems Laboratory 92-11977 Report ESL-R-814, Cambridge, MA, April 1978. 91r 5 01 010 MAXIMUM LIKELIHOOD ESTIMATION OF FRACTIONAL BROWNIAN MOTION AND MARKOV NOISE PARAMETERS by 4'f: MATTHEW E. SKEEN B.S., Astronautical Engineering, United States Air Force Academy Colorado Springs, Colorado (1990) Ace w34o& For Submitted to the Department of Aeronautics and Astronautics NT IS Qgi" in Partial Fulfillment of the Requirements for the Degree of 971 tA- MASTER OF SCIENCE in AERONAUTICS AND ASTRONAUTICS ttat at the __ MASSACHUSETTS INSTITUTE OF TECHNOLOGY February 1992 -VB±1 , Diet ropoojal © Matthew E. Skeen, 1991. All Rights Reserved The author hereby grants to M.I.T. permission to reproduce and to distribute copies of this thesis document in whole or in part. Signature of Author Department of Aeronautics and Astronautics 20 December 1991 Approved by Dr. Michael E. Ash Principal Member Technical Staff, Charles Stark Draper Laboratory Thesis Supervisor Certified by Professor Wallace E. Vander Velde Department of Aeronautics and Astronautics Thesis Advisor Accepted by Professor Harold Y. Wachman Chairman, Department Graduate Committee £ 2 MAxM ULIKEuHOOD ESTIMATION OF FRACTIONAL BROWNIAN MOTION AND MARKOV NOISE PARAMETERS MATTHEW EDWARD SKEEN Submitted to the Department of Aeronautics and Astronautics on December 20, 1991 in partial fulfillment of the requirements for the degree of Master of Science in Aeronautics and Astronautics ABSTRACT Maximum likelihood estimation and power spectral density analysis are developed as tools for the analysis of stochastic processes. Some useful results from the theory of Markov stochastic processes are then presented followed by the introduction of fractional Brownian motion and fractional Gaussian noise as non-Markov models for systems with power spectral density proportional to f0, where -3 < 3 <-1 and -1 < 3 < 1 over all frequencies. Maximum likelihood system identification is applied to estimating the unknown parameters in a Markov model which approximates fractional Brownian motion. The algorithm runs a Kalman filter on states and a maximum likelihood estimator on parameters. Results are presented from estimating trend, white noise, random walk, and exponentially correlated noise parameters from fits to simulated and real test data. Maximum likelihood estimation is applied to the batch estimation of parameters in the non-Markov fractional Brownian motion model. New in this thesis is the use of partial derivatives to minimize the resulting likelihood function, and the capability to estimate the unknown parameters of additional trend and Markov noise processes. Results are presented from fits to computer simulated sample paths. Thesis Supervisor: Dr. Michael E. Ash Principal Member Technical Staff, C. S. Draper Laboratory Thesis Advisor: Professor Wallace E. Vander Velde Department of Aeronautics and Astronautics 3 4 Acknowledgements My sincerest gratitude is extended to my thesis supervisor, Dr. Michael E. Ash, not only for his enthusiastic assistance in my research, but also for his interest in my overall education. I would like to thank my thesis advisor, Prof. Wallace E. Vander Velde, for his guidance in the preparation of this document and for the care with which he reviewed the drafts. I thank my friends at Draper Laboratory for all the help that they have given me and for making my time at MIT enjoyable. I give special thanks to Tanya for her love, support, and perseverance over the last 17 months. 5 This thesis was researched and written at the Charles Stark Draper Laboratory under Corporate Sponsored Research Project C68. Publication of this thesis does not constitute approval by the laboratory of the findings or conclusions contained herein. It is published for the exchange and stimulation of ideas. I hereby assign my copyright of this thesis to the Charles Stark Draper Laboratory, Inc., of Cambridge, Massachusetts. Matthew E. Skeen Permission is hereby granted by the Charles Stark Draper Laboratory, Inc. to the Massachusetts Institute of Technology to reproduce and to distribute copies of this thesis document in whole or in part. 6 Table of Contents Ab stract ..................................................................................................................... 3 Acknowledgements .............................................................................................. 5 Table of Contents .................................................................................................. 7 List of Illustrations ................................................................................................. 13 List of T ables ........................................................................................................... . 13 N otation ................................................................................................................... 15 Chapter 1 Introduction and Summary ............................................................. 19 1.1 Modeling Noise Processes and Estimating Noise Parameters.. 19 1.2 Background Material .......................................................................... 21 1.2.1 Maximum Likelihood Estimation .................... 21 1.2.2 Power Spectral Density Analysis ............................................. 22 1.2.3 Markov Noise Processes ............................................................ 22 1.2.4 Fractional Brownian Motion ................................................... 24 1.3 Summary of Maximum Likelihood Fits to Real and Simulated Data ......................................................................... 24 1.3.1 Maximum Likelihood System Identification for Markov Pr ocesses .......................................2.4............................................. 1.3.2 Maximum Likelihood Estimation of Fractional Brownian and Other Parameters .................... 25 7 Chapter 2 Maximum Likelihood Estimation ................................................. 27 2.1 Parameter Estimation ........................................................................ 27 2.2 Likelihood Function and Negative Log-Likelihood Function 27 2.3 Fisher Information Matrix ................................................................ 28 2.4 Cramer-Rao Lower Bound ................................................................ 29 2.5 Properties of Maximum Likelihood Estimates .............. 31 2.6 Iterative Determination of Maximum Likelihood Estimates... 32 2.7 Relation to Least Squares Estimates ................................................ 34 2.8 Relation to Other Estimators ............................................................ 36 Chapter 3 Power Spectral Density Analysis ..................................................... 39 3.1 Stochastic Processes ............................................................................ 39 3.2 Autocorrelation Function of a Stochastic Process ....................... 40 3.3 Estimation of the Autocorrelation Function ................ 41 3.4 Power Spectral Density of a Stationary Stochastic Process ......... 43 3.5 Estimation of the Power Spectral Density .................. 43 3.5.1 Continuous Data ........................................................................ 43 3.5.2 Interpretation of the PSD as a Power Spectrum ................... 45 3.5.3 D iscrete D ata ................................................................................ . 47 3.6 Frequency Averaging ........................................................................ 49 Chapter 4 Markov Stochastic Processes ........................................................... 53 4.1 Martingale and Markov Processes .................................................. 53 4.2 Weiner Brownian Motion Process ................................................. 54 4.3 White Noise PSD Slope ..................................................................... 56 4.4 PSD Slope of Random Walk, Trend, and Quantization N oise .......................................................................................... . 57 4.5 Ito Stochastic Integral ........................................................................ 61 8

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.