Files in this item

FilesDescriptionFormat

application/pdf

application/pdfRiedl_Thomas.pdf (425kB)
(no description provided)PDF

Description

Title:A Graphical Model for the Communications Channel
Author(s):Riedl, Thomas J.
Advisor(s):Singer, Andrew C.
Contributor(s):Singer, Andrew C.
Department / Program:Electrical & Computer Eng
Discipline:Electrical & Computer Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:M.S.
Genre:Thesis
Subject(s):communications
Markov random field
channel estimation
channel modeling
gaussian message passing
Abstract:We consider the problem of channel modeling and channel estimation. The widely used wide sense stationary uncorrelated scattering model for the communications channel neglects correlations between different multipath arrivals, but this seems to oversimplify the real channel in many cases. One example is the underwater acoustic channel, whose impulse response is fairly continuous in delay and hence indeed exhibits a certain correlation structure in delay. To address this shortcoming we introduce a novel channel model that is based on a Gaussian Markov random field (MRF) for the complex channel gains. This graphical model is used to capture the local nature of the statistical dependencies (in time and space) of the channel taps. In order for the MRF model to fit the actual physical channel well, its parameters must be adapted appropriately. Our approach is to find the maximum likelihood (ML) estimate of theses parameters based on given observations. Once these parameters are known the MRF model can then either be used for channel estimation directly or it can be embedded into an iterative (turbo) receiver, where it is expected to improve the data estimation performance significantly as the parameterized MRF carries prior knowledge on the channel.
Issue Date:2010-01-06
URI:http://hdl.handle.net/2142/14667
Rights Information:Copyright 2009 Thomas J. Riedl
Date Available in IDEALS:2010-01-06
Date Deposited:December 2


This item appears in the following Collection(s)

Item Statistics