Abstract

An equalizer is an adaptive filter that compensates for the non-ideal characteristics of a communication channel by processing the received signal. The adaptive algorithm searches for the inverse impulse response of the channel, and it requires knowledge of a training sequence, in order to generate an error signal necessary for the adaptive process. There are practical situations where it would be highly desirable to achieve complete adaptation without the use of a training sequence, hence the the term "blind". Examples of these situations are multipoint data networks, high-capacity line-of-sight digital radio, and reflection seismology. A blind adaptive algorithm has been developed, based on simplified equalization criteria. These criteria are that the second- and fourth-order moments of the input and output sequences are equalized. The algorithm is entirely driven by statistics, only requiring knowledge of the variance of the input signal. Because of the insensitivity of higher-order statistics to Gaussian processes, the algorithm performs well when additive white Gaussian noise is present in the channel. Simulations are presented in which the new blind equalizer developed is compared to other equalization algorithms.

Library of Congress Subject Headings

Equalizers (Electronics); Algorithms; Signal processing--Computer simulation

Publication Date

8-1-1994

Document Type

Thesis

Department, Program, or Center

Electrical Engineering (KGCOE)

Advisor

Dianat, S.

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: TK7872.E7D53 1994

Campus

RIT – Main Campus

Download

Share

COinS