Finite precision error control and array implementation of signal processing algorithms
Posted on:1998-12-25
Degree:Ph.D
Type:Dissertation
University:University of California, Los Angeles
Candidate:Yang, Tung-chiang
Full Text:PDF
GTID:1468390014476971
Subject:Engineering
Abstract/Summary:
Forgetting factor based systems and fixed sized sliding window systems have their own advantages and disadvantages in parameter estimation and channel identification problems of many signal processing applications. Numerical errors caused by the mismatches of the updating and the downdating of a certain data in finite precision hardwares are observed, and a stabilization factor concept is proposed for stabilizing parameters with certain properties. An array architecture for implementing the ULLVD algorithm for rank revealing purposes is also presented.