EECE-595, Section II, Adaptive Filtering

Term: Spring 2003

Instructor: Balu Santhanam

Pre-requisites: EECE-539, EECE-541, knowledge of MATLAB


Announcements :

o Please check here often because announcements about the course will be posted here.

Course Materials:

o Flier for the course
o Course Outline/Syllabus

Review material:

Class Notes :

Preliminaries :

o Lecture I notes
o Lecture II notes
o Hilbert Space View of Random Signals
o On Signals with Rational Power Spectra
o Power Spectrum Factorization
o On Autoregressive Processes
o On Linear Prediction and Autoregressive Processes

LMS Algorithm and Variants:

o Steepest Descent: AR(2) Example
o Steepest Descent Versus Newton's Algorithm
o Lecture Notes on the LMS Algorithm
o Lecture Notes on the NLMS Algorithm
o NLMS: Minimum Norm/SVD solution
o AR(2) Example: (a) Average Tap-weights and (b) Learning Curve
o Lecture Notes on Affine Projection Algorithm
o Lecture Notes on Variants of the LMS

RLS Algorithm and Variants:

o On Least Squares Inversion
o On the Least Squares Algorithm
o Exponentially Weighted RLS Algorithm
o RLS Algorithm: Design Guidelines
o AR(2) Example: RLS Tap-weights

Kalman Filter and Variants:

o Discrete Kalman Filter
o Relation Between the DKF and RLS
o DKF AR(2) Prediction Example: o State estimate o Kalman gain vector o MMSE learning curve
o On Wiener and Kalman Filters
o Extended Kalman Filter (EKF)
o Iterated Extended Kalman Filter (IEKF)

Order Recursive Adaptive Filters:

o Gradient Adaptive Lattice
o Least Squares Lattice

Problem Sets :

o Problem Set # 1.0
o Solutions to Problem Set # 1.0
o Problem Set # 2.0
o Sample output from Problem Set # 2.0
o Solution to Problem Set # 2.0

MATLAB Files:

o LMS Algorithm
o Normalized LMS Algorithm
o Recursive Least Squares (RLS) Algorithm
o Script for AR(2) example : I (NLMS)
o Script for AR(2) example : II (RLS)
o Script for AR(2) example : III (DKF)
o Discrete Kalman Filter
o EKF for Tracking Example