Electrical and Computer Engineering ETDs

Publication Date

8-22-1979

Abstract

A new element, the adaptive delay, is developed as a means for synchronizing two signals that are highly correlated. An LMS error function derived and gradient techniques are used to determine an adaptive delay algorithm. Like Widrow's adaptive filter coefficient algorithm, this adaptive delay algorithm does not explicitly utilize a priori statistics on the input signals. However, unlike the LMS error function for the adaptive filter, the LMS error function for the adaptive delay is not quadratic. Thus, the error function will be required to be unimodal in order to assure convergence of the adaptive delay to its optimum value. Further analysis of the adaptive algorithm determines when the gradient estimate is biased and gives bounds for the convergence parameter. Applications presented include delay-lock loops, frequency tracking, and seismic noise­cancelling. This new adaptive element is significant in that, in certain situations, it can either replace entire adaptive filters or be used with a small adaptive filter and still minimize the error. In some configurations, the adaptive delay yields information not easily discernable from either adaptive filters or Wiener filters. Furthermore, while the simple adaptive algorithm requires a small amount of memory storage, it does not require squaring, differentiating, averaging, or the use of transforms, thus making it very attractive for use in microprocessor-based systems.

Document Type

Dissertation

Language

English

Degree Name

Electrical Engineering

Level of Degree

Doctoral

Department Name

Electrical and Computer Engineering

First Committee Member (Chair)

Daniel Paul Peterson

Second Committee Member

Samuel Dow Stearns

Third Committee Member

Shlomo Karni

Fourth Committee Member

Richard H. Williams

Fifth Committee Member

Lambert Herman Koopmans

Download

Share

COinS