Sequence and channel estimation for channels with memory /

Sequence and channel estimation for communication channels with memory are addressed in this work. The slowest descent method as a novel approach to discrete multi-parameter estimation is introduced. It is shown that, this method can achieve close-to-optimal performance for sequence estimation in mu...

Full description

Bibliographic Details
Main Author: Spasojevic, Predrag, 1966-
Format: Thesis Book
Language:English
Published: [Place of publication not identified] : [publisher not identified] ; 1999.
Subjects:
Online Access:http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=731686271&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD
Description
Summary:Sequence and channel estimation for communication channels with memory are addressed in this work. The slowest descent method as a novel approach to discrete multi-parameter estimation is introduced. It is shown that, this method can achieve close-to-optimal performance for sequence estimation in multi-user detection, intersymbol interference, and Gaussian interference channels. An expectation maximization (EM) based sequence estimator is demonstrated to achieve close to optimal performance for the finite-dimensional and reduced-rank interference models. In the case of an interference channel, the generalized sequence detector is proven optimal for the case of large variance of all relevant components in the Gaussian interference model. Sufficient (non-)convergence conditions for discrete parameter estimation using the EM algorithm are derived. For the interference channel, by reduction in the rank of the channel model, one can ensure the convergence of the EM algorithm. A realizable iterative sequence estimator based on the slowest descent method that allows for resolution of unresolved path components is derived for the quasi-static multi-path fading channel. Additionally, optimal pilot symbol sequence pairs for channel estimation in quasi-static ISI channels are derived. New almost-complementary periodic sequences of odd length are introduced. It is shown that pilot and zero-terminated sequences can eliminate or significantly reduce the number of sequences ambiguous to a maximum-likelihood sequence estimator in a quasi-static fading ISI channel. It is also demonstrated that in some cases avoiding narrow-band interference can improve the system's performance significantly. This is in particular true for the overlay systems where the interference of the wide-band system to the narrow-band system degrades the system's performance significantly.
Item Description:Vita.
"Major Subject: Electrical Engineering".
Physical Description:xiv, 176 leaves : illustrations ; 28 cm.
Issued also on microfiche from University Microfilm Inc.
Bibliography:Includes bibliographical references (leaves 151-161).