Please use this identifier to cite or link to this item:
http://hdl.handle.net/11375/28430
Title: | ON THE RATE-DISTORTION FUNCTION OF SYMMETRIC REMOTE GAUSSIAN MULTITERMINAL SOURCE CODING: BOUNDS AND ASYMPTOTICS |
Authors: | Zhou, Siyao |
Advisor: | Jun Chen |
Department: | Electrical and Computer Engineering |
Publication Date: | 2023 |
Abstract: | Due to the development of the Internet of Things (IoT), one frequently encounters the scenarios where data collected at different sites need to be compressed and then forwarded to a fusion center for joint processing. As such data are typically correlated from one site to another, it is desirable to capitalize on the statistical dependencies to improve the compression efficiency. The multiterminal source coding problem and its variants aim to characterize the performance limits of this type of distributed compression systems. This thesis is divided into two major parts. The first part deals with so-called remote multiterminal source coding, where L encoders compress their respective ob- servations and send the compressed data to a central decoder for the joint recon- struction of target signals. The fundamental limit of remote multiterminal source coding is characterized by the rate-distortion function, which delineates the optimal tradeoff between the compression rate and the reconstruction distortion. For simplic- ity, it is assumed that the observed sources can be expressed as the sum of target signals and corruptive noises which are independently generated from two symmetric multivariate Gaussian distributions. For this special case, an explicit lower bound on the rate-distortion function is established and is shown to match the well-known Berger-Tung upper bound in some distortion regimes. The asymptotic gap between the upper and lower bounds is computed in the large L limit. The second part considers the centralized encoding setting where the L sources are jointly observed and compressed by a single encoder. The rate-distortion function for this setting is completely characterized and is leveraged as a rate-distortion lower bound for the symmetric remote Gaussian multiterminal source coding problem in view of the fact that centralized encoding is more powerful than distributed encoding. It is shown that this centralized-encoding lower bound is not as tight as the lower bound established in the first part. The asymptotic analysis of this centralized- encoding lower bound is also provided. |
URI: | http://hdl.handle.net/11375/28430 |
Appears in Collections: | Open Access Dissertations and Theses |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
Siyao_Zhou_Ph_D__Thesis.pdf | 799.22 kB | Adobe PDF | View/Open |
Items in MacSphere are protected by copyright, with all rights reserved, unless otherwise indicated.