[pt] Esta dissertação aborda o estudo de códigos fontanais (códigos LT e códigos Raptor) que são uma classe de códigos criados para a transmissão de dados de maneira confiável e eficiente através de canais os quais podem ser modelados como canais com apagamento. Os códigos LT e códigos Raptor são denominados códigos fontanais, devido a que eles são uma boa aproximação para o conceito de fontanas digitais. Além disso, eles são classificados como códigos de taxa versátil, no sentido que o número de símbolos codificados que podem ser gerados a partir dos dados de entrada é potencialmente ilimitado. Códigos LT são capazes de recuperar, com probabilidade maior do que (1 − delta), um conjunto de k símbolos de entrada a partir de quaisquer k + O((raiz quadrada de k)(ln(2))(k/delta)) símbolos codificados recebidos, com uma média de O(k ln(k/delta)) operações XOR. Os códigos Raptor são uma extensão de códigos LT, na qual o processo de codificação é composto de duas etapas: um código de bloco de comprimento fixo (denominado pré- código) e um código LT com uma distribuição de graus apropriada. Investigou-se o desempenho dos códigos LT usando duas novas distribuições de graus (Sóliton Robusta Melhorada e Sóliton Robusta Truncada) e foi proposto um modelo de códigos LT Bidimensionais, na qual os símbolos de entrada são agrupados em forma de matriz. Neste esquema os blocos correspondentes às linhas da matriz são codificados usando um código LT e, em seguida, a matriz resultante tem suas colunas também codificadas usando um código LT. Ainda que a complexidade do esquema tenha sido dobrada o desempenho alcançado pelos códigos LT Bidimensionais superou o desempenho dos códigos LT convencionais para situações em que a qualidade do canal BEC é elevada. / [en] Fountain Codes (LT Codes and Raptor Codes) are a class of codes proposed to efficient and reliably transmit data through Erasure Channels. LT Codes and Raptor Codes are a good approximation to the concept of digital fountain and as such are named as fountain codes. They are said to be rateless codes in the sense that the number of symbols produced by the encoder could grow, potentially, to infinite. With probability of success larger than (1−delta), a decoder of an LT code based scheme can recover the k transmitted symbols from any received block of k + O((square root k)(ln(2))(k/delta)) correct symbols with an average of O(k ln(k/delta)) XOR operations. Raptor codes are an extension of the LT codes idea, with a tandem scheme where a fixed length block code (namely a pre- code) is followed by an LT code that uses a properly chosen degree distribution. In this dissertation the performance of LT codes with two recently proposed degree distributions, the Improved Robust Soliton and the Truncated Soliton Robust Distribution were investigated. A new scheme called Bidimensional LT Codes, has been proposed. In this scheme the input symbols are structured in a matrix form and afterwards the blocks corresponding to the lines of the matrix are encoded with an LT code. The columns of the new matrix so obtained are next encoded with a similar LT code. The complexity of the new scheme is doubled and yet its performance only just surpasses that of the conventional LT scheme for high quality BEC.

An ever increasing demand for instant and reliable information on modern communication networks forces codewords to operate in a non-asymptotic regime. To achieve reliability for imperfect channels in this regime, codewords need to be retransmitted from receiver to the transmit buffer, aided by a fast feedback mechanism. Large occupancy of this buffer results in longer communication delays. Therefore, codewords need to be designed carefully to reduce transmit queue-length and thus the delay experienced in this buffer. We first study the consequences of physical layer decisions on the transmit buffer occupancy. We develop an analytical framework to relate physical layer channel to the transmit buffer occupancy. We compute the optimal code-rate for finite-length codewords operating over a correlated channel, under certain communication service guarantees. We show that channel memory has a significant impact on this optimal code-rate. Next, we study the delay in small ad-hoc networks. In particular, we find out what rates can be supported on a small network, when each flow has a certain end-to-end service guarantee. To this end, service guarantee at each intermediate link is characterized. These results are applied to study the potential benefits of setting up a network suitable for network coding in multicast. In particular, we quantify the gains of network coding over classic routing for service provisioned multicast communication over butterfly networks. In the wireless setting, we study the trade-off between communications gains achieved by network coding and the cost to set-up a network enabling network coding. In particular, we show existence of scenarios where one should not attempt to create a network suitable for coding. Insights obtained from these studies are applied to design a distributed rate control algorithm in a large network. This algorithm maximizes sum-utility of all flows, while satisfying per-flow end-to-end service guarantees. We introduce a notion of effective-capacity per communication link that captures the service requirements of flows sharing this link. Each link maintains a price and effective-capacity, and each flow maintains rate and dissatisfaction. Flows and links update their respective variables locally, and we show that their decisions drive the system to an optimal point. We implemented our algorithm on a network simulator and studied its convergence behavior on few networks of practical interest.

<p dir="ltr"> The general design goal of low latency communication systems is to minimize the end-to-end delay while attaining the predefined reliability and throughput requirements. The burgeoning demand for low latency communications motivates a renewed research interest of the tradeoff between delay, throughput, and reliability. In this dissertation research, we consider slotted-based systems and explore the potential advantages of the so-called sequential codes in low latency network communications.</p><p dir="ltr"> The first part of this dissertation analyzes the exact error probability of random linear streaming codes (RLSCs) in the large field size regime over the stochastic independently and identically distributed (i.i.d.) symbol erasure channels (SECs). A closed-form expression of the error probability <i>p</i>_<em>e</em> of large-field-size RLSCs is derived under, simultaneously, the finite memory length α and decoding deadline Δ constraints. The result is then used to examine the intricate tradeoff between memory length (complexity), decoding deadline (delay), code rate (throughput), and error probability (reliability). Numerical evaluation shows that under the same code rate and error probability requirements, the end-to-end delay of RLSCs is 40-48% of that of the optimal block codes (i.e., MDS codes). This implies that switching from block codes to streaming codes not only eliminates the queueing delay completely (which accounts for the initial 50% of the delay reduction) but also improves the reliability (which accounts for the additional 2-10% delay reduction).</p><p dir="ltr"> The second part of this dissertation focuses on the asymptotics of the error probability of RLSCs in the same system model of the first part. Two important scenarios are analyzed: (i) tradeoff between Δ and <i>p</i>_<em>e</em> under infinite α; and (ii) tradeoff between α and <i>p</i>_<em>e</em> under infinite Δ. In the first scenario, the asymptote of <i>p</i>_<em>e</em>(Δ) is shown to be <i>ρ</i>Δ^-1.5<i>e</i>^-^<em>η</em>^Δ. The asymptotic power term Δ^-1.5 of RLSCs is a strict improvement over the Δ^-0.5 term of random linear block codes. A pair of upper and lower bound on the asymptotic constant <i>ρ</i> is also derived, which are tight (i.e., identical) for one specific class of SECs. In the second scenario, a refine approximation is proposed by computing the parameters in a multiterm asymptotic form, which closely matches the exact error probability even for small memory length (≈ 20). The results of the asymptotics can be further exploited to find the <i>c</i>-optimal memory length <i>α</i>_<em>c</em>^*(Δ), which is defined as the minimal memory length α needed for the resulting <i>p</i>_<em>e</em> to be within a factor of <i>c</i>>1 of the best possible <i>p</i>_<em>e</em>^<em>*</em>_{<em> </em>}for any Δ, an important piece of information for practical implementation.</p><p dir="ltr"> Finally, we characterize the channel dispersions of RLSCs and MDS block codes, respectively. New techniques are developed to quantify the channel dispersion of sequential (non-block-based) coding, the first in the literature. The channel dispersion expressions are then used to compare the levels of error protection between RLSCs and MDS block codes. The results show that if and only if the target error probability <i>p</i>_<em>e</em> is smaller than a threshold (≈ 0.1774), RLSCs offer strictly stronger error protection than MDS block codes, which is on top of the already significant 50% latency savings of RLSCs that eliminate the queueing delay completely.</p>