Consider the problem of designing capacity-achieving codes for multi-terminal communication scenarios. For point-to-point communication problems, one can optimize a single code to approach capacity, but for multi-terminal problems this translates to optimizing a single code to perform well over the entire region of channel parameters. A coding scheme is called universal if it allows reliable communication over the entire achievable region promised by information theory. It was recently shown that terminated low-density parity-check convolutional codes (also known as spatially-coupled low-density parity-check ensembles) have belief-propagation thresholds that approach their maximum a-posteriori thresholds. This phenomenon, called "threshold saturation via spatial-coupling," was proven for binary erasure channels and then for binary memoryless symmetric channels. This approach provides us with a new paradigm for constructing capacity approaching codes. It was also conjectured that the principle of spatial coupling is very general and that the phenomenon of threshold saturation applies to a very broad class of graphical models. In this work, we consider a noisy Slepian-Wolf problem (with erasure and binary symmetric channel correlation models) and the binary-input Gaussian multiple access channel, which deal with correlation between sources and interference at the receiver respectively. We derive an area theorem for the joint decoder and empirically show that threshold saturation occurs for these multi-user scenarios. We also show that the outer bound derived using the area theorem is tight for the erasure Slepian-Wolf problem and that this bound is universal for regular LDPC codes with large left degrees. As a result, we demonstrate near-universal performance for these problems using spatially-coupled coding systems.

The spread of infectious agents has been observed as long as their hosts have existed. The spread of infectious diseases in human populations, however, is more than an academic concern, causing millions of deaths every year, and prompting collective surveillance and intervention efforts worldwide. These surveillance data, used in conjunction with statistical methods and mathematical models, present both challenges and opportunities for advancements in scientific understanding and public health. Early mathematical modeling of infectious diseases in humans began by assuming homogeneous contact among individuals, but has since been extended to account for many sources of non-homogeneity in human contact. Detecting the degree of epidemic mixing between geographically separated populations, in particular, remains a difficult problem. The difficulty occurs because although disease case reports have been collected by many governments for decades, case reporting is imperfect, and transmission events themselves are nearly impossible to observe. The degree to which epidemic coupling can be detected from case reports is the central theme of this thesis. We present a careful, biologically motivated and consistent derivation of the transmission coupling (fully derived in Chapter 4). In Chapter 2 we consider the simple scenario of an epidemic spreading from one population to another, and present both numerical and analytic methodology for estimating epidemic coupling. Chapter 3 considers the problem of estimating epidemic coupling among populations undergoing recurrent epidemics, such as those of childhood diseases which have been widely observed. In Chapter 4 we present a method for estimating coupling among an arbitrary number of populations undergoing an epidemic, and apply it to estimate coupling among the parishes of London, England, during the Great Plague of 1665. / Thesis / Doctor of Philosophy (PhD)

This dissertation studies methods to achieve reliable communication over unreliable channels. Iterative decoding algorithms for low-density parity-check (LDPC) codes and generalized LDPC (GLDPC) codes are analyzed. A new class of error-correcting codes to enhance the reliability of the communication for high-speed systems, such as optical communication systems, is proposed. The class of spatially-coupled GLDPC codes is studied, and a new iterative hard- decision decoding (HDD) algorithm for GLDPC codes is introduced. The main result is that the minimal redundancy allowed by Shannon’s Channel Coding Theorem can be achieved by using the new iterative HDD algorithm with spatially-coupled GLDPC codes. A variety of low-density parity-check (LDPC) ensembles have now been observed to approach capacity with iterative decoding. However, all of them use soft (i.e., non-binary) messages and a posteriori probability (APP) decoding of their component codes. To the best of our knowledge, this is the first system that can approach the channel capacity using iterative HDD. The optimality of a codeword returned by the weighted min-sum (WMS) algorithm, an iterative decoding algorithm which is widely used in practice, is studied as well. The attenuated max-product (AttMP) decoding and weighted min-sum (WMS) decoding for LDPC codes are analyzed. Applying the max-product (and belief- propagation) algorithms to loopy graphs are now quite popular for best assignment problems. This is largely due to their low computational complexity and impressive performance in practice. Still, there is no general understanding of the conditions required for convergence and/or the optimality of converged solutions. This work presents an analysis of both AttMP decoding and WMS decoding for LDPC codes which guarantees convergence to a fixed point when a weight factor, β, is sufficiently small. It also shows that, if the fixed point satisfies some consistency conditions, then it must be both a linear-programming (LP) and maximum-likelihood (ML) decoding solution.

Ce travail est dédié à la conception, l’analyse et l’évaluation des performances de nouveaux schémas de codage appropriés aux systèmes de stockage distribué. La première partie de ce travail est consacrée à l’étude des performances des codes spatialement couplés pour les canaux à effacements. Une nouvelle méthode de couplage spatial des ensembles classiques de contrôle de parité à faible densité (LDPC) est proposée. La méthode est inspirée du codage en couches. Les arêtes des ensembles locaux et celles définissant le couplage spatial sont construites séparément. Nous proposons également de saturer le seuil d’un ensemble Root-LDPC par couplage spatial de ses bits de parité dans le but de faire face aux évanouissements quasi-statiques. Le couplage spatial est dans un deuxième temps appliqué à un ensemble Root-LDPC, ayant une double diversité, conçu pour un canal à effacements par blocs à 4 états. Dans la deuxième partie de ce travail, nous considérons les codes produits non-binaires avec des composantes MDS et leur décodage algébrique itératif ligne-colonne sur un canal à effacements. Les effacements indépendants et par blocs sont considérés. Une représentation graphique compacte du code est introduite avec laquelle nous définissions la notion de coloriage à double diversité. Les ensembles d’arrêt sont définis et une caractérisation complète est donnée. La performance des codes produits à composantes MDS, avec et sans coloration, à double diversité est analysée en présence d’effacements indépendants et par blocs. Les résultats numériques montrent aussi une excellente performance en présence d’effacements à probabilité inégale due au coloriage ayant une double diversité. / This work is dedicated to the design, analysis, and the performance evaluation of new coding schemes suitable for distributed storage systems. The first part is devoted to spatially coupled codes for erasure channels. A new method of spatial coupling for low-density parity-check ensembles is proposed. The method is inspired from overlapped layered coding. Edges of local ensembles and those defining the spatial coupling are separately built. We also propose to saturate the whole Root-LDPC boundary via spatial coupling of its parity bits to cope with quasi-static fading. Then, spatial coupling is applied on a Root-LDPC ensemble with double diversity designed for a channel with 4 block-erasure states. In the second part of this work, we consider non-binary product codes with MDS components and their iterative row-column algebraic decoding on the erasure channel. Both independent and block erasures are considered. A compact graph representation is introduced on which we define double-diversity edge colorings via the rootcheck concept. Stopping sets are defined and a full characterization is given in the context of MDS components. A differential evolution edge coloring algorithm that produces colorings with a large population of minimal rootcheck order symbols is presented. The performance of MDS-based product codes with and without double-diversity coloring is analyzed in presence of both block and independent erasures. Furthermore, numerical results show excellent performance in presence of unequal erasure probability due to double-diversity colorings.