ITC/USA 2008 Conference Proceedings / The Forty-Fourth Annual International Telemetering Conference and Technical Exhibition / October 27-30, 2008 / Town and Country Resort & Convention Center, San Diego, California / This paper considers the joint selection of the source code rate and channel code rate in an image telemetry system. Specifically considered is the JPEG2000 image coder and an LDPC code family. The goal is to determine the optimum apportioning of bits between the source and channel codes for a given channel signal-to-noise ratio and total bit rate, R(total). Optimality is in the sense of maximum peak image SNR and the tradeoff is between the JPEG2000 bit rate R(source) and the LDPC code rate R(channel). For comparison, results are included for the industry standard rate-1/2, memory-6 convolutional code.

Dans cette thèse, nous nous intéressons aux schémas de codage et de décodage source-canal conjoint des contenus multimédia. Nous montrons comment la redondance laissée par le codeur vidéo peut être exploitée pour réaliser un décodage robuste des séquences transmises sur un lien radio-mobile bruité. grâce au schéma de décodage conjoint proposé, le nombre de paquets corrompus est significativement réduit au prix d'une très légère augmentation du débit. Nous appliquons ensuite ce schéma de décodage robuste à latransmission par descriptions multiples sur une architecture mixte Internet et radio-mobile. Le décodage source-canl conjoint des paquets reçus permet de corriger les erreurs de transmission et d'augmenter ainsi le nombre de paquets utilisés par le décodeur pour compenser les paquets perdus. L'efficacité de ce schéma est étudiée par rapport à un schéma classique basé sur les décisions dures du canal et sur un code correcteur d'erreurs introduisant le même niveau de redondance. Une deuxième partie de la thèse est consacrée à l'étude de schémas de codage source-canal conjoint basés sur une transformation redondante. Deux schémas d'estimation ont été proposés. Dans le premier, nous exploitons la redondance structurée introduite et le caractère borné du bruit de quantification pour construire un estimateur cohérent corrigeant les erreurs de transmission. Dans le deuxième schéma, nous appliquons l'algorithme de propagation de croyances pour évaluer les distributions a posteriori des composantes du signal d'entrée, à partir de sorties bruitées du canal. Nous appliquons alors ces deux schémas pour estimer l'entrée d'un banc de filtres suréchantillonnés. / This thesis aims at proposing and implementing efficient joint source-channel coding and decoding schemes in order to enhance the robustness of multimedia contents transmitted over unreliable networks. In a first time, we propose to identify and exploit the residual redundancy left by wavelet video coders in the compressed bit streams. an efficient joint-source channel decoding scheme is proposed to detect and correct some of the transmission errors occurring during a noisy transmission. This technique is further applied to multiple description video streams transmitted over a mixed architecture consisting of a wired lossy part and a wireless noisy part. In a second time, we propose to use the structured redundancy deliberately introduced by multirate coding systems, such as oversampled filter banks, in order to perform a robust estimation of the input signals transmitted over noisy channels. Two efficient estimation approaches are proposed and compared. The first one exploits the linear dependencies between the output variables, jointly to the bounded quantization noise, in order to perform a consistent estmiation of the source outcome. The second approach uses the belief propagation algorithm to estimate the input signal via a message passing procedure along the graph representing the linear dependencies between the variables. These schemes ares then applied to estimate the input of an oversampled filter bank and their performance are compared.

<p> This thesis launches some new inquires and makes significant progress in the active research areas of joint source-channel coding and multiple description coding. Two interesting but previously untreated problems are investigated and partially settled: 1) can index assignment of source codewords be optimized with respect to a given joint source-channel decoding scheme, and if so, how? 2) can joint source-channel coding be optimized with respect to a given multiple description code, and if so, how?</p> <p> The first problem is formulated as one of quadratic assignment. Although quadratic assignment is NP-hard in general, we are able to develop a near-optimum index assignment algorithm for joint source-channel (JSC) maximum a posteriori (MAP) decoding, if the input is a Gaussian Markov sequence of high correlation. For general cases, good heuristic solutions are proposed. Convincing empirical evidence is presented to demonstrate the performance improvement of the index assignments optimized for MAP decoding over those designed for hard-decision decoding.</p> <p> The second problem is motivated by applications of signal communication and estimation in resource-constrained lossy networks. To keep the encoder complexity at a minimum, a signal is coded by a multiple description quantizer (MDQ) without channel coding. The code diversity of MDQ and the path diversity of the network are exploited by decoders to combat transmission errors. A key design objective is resource scalability: powerful nodes in the network can perform JSC-MD estimation under the criteria of maximum a posteriori probability or minimum mean-square error (MMSE), while primitive nodes resort to simpler multiple description (MD) decoding, all working with the same MDQ code. The application of JSC-MD to distributed estimation of hidden Markov models in a sensor network is demonstrated. The proposed JSC-MD MAP estimator is an algorithm of the longest path in a weighted directed acyclic graph, while the JSC-MD MMSE decoder is an extension of the well-known forward-backward algorithm to multiple descriptions. They outperform the existing hard-decision MDQ decoders by large margins.</p> / Thesis / Doctor of Philosophy (PhD)

ITC/USA 2005 Conference Proceedings / The Forty-First Annual International Telemetering Conference and Technical Exhibition / October 24-27, 2005 / Riviera Hotel & Convention Center, Las Vegas, Nevada / A practical joint source/channel coding algorithm is proposed for the transmission of multiple images and videos to reduce the overall reconstructed source distortion at the receiver within a given total bit rate. It is demonstrated that by joint coding of multiple sources with such an objective, both improved distortion performance as well as reduced quality variation can be achieved at the same time. Experimental results based on multiple images and video sequences justify our conclusion.

International Telemetering Conference Proceedings / October 18-21, 2004 / Town & Country Resort, San Diego, California / This paper presents a joint source-channel decoding scheme based on a JPEG2000 source coder and an LDPC channel coder. At the encoder, JPEG2000 is used to perform source coding with certain error resilience (ER) modes, and LDPC codes are used to perform channel coding. At the decoder, after one iteration of LDPC decoding, the output codestream is then decoded by JPEG2000. With the error resilience mode switches on, the source decoder detects the position of the first error within each codeblock of the JPEG2000 codestream. This information is fed back to the channel decoder, and incorporated into the calculation of likelihood values of variable nodes for the next iteration of LDPC decoding. Our results indicate that the proposed method has significant gains over conventional separate channel and source decoding.

This dissertation is divided into two main topics. In the first topic, we study the joint source-channel coding problem of transmitting an analog source over a Gaussian channel in two cases - (i) the presence of interference known only to the transmitter and (ii) in the presence of side information about the source known only to the receiver. We introduce hybrid digital analog forms of the Costa and Wyner-Ziv coding schemes. We present random coding based schemes in contrast to lattice based schemes proposed by Kochman and Zamir. We also discuss superimposed digital and analog schemes for the above problems which show that there are infinitely many schemes for achieving the optimal distortion for these problems. This provides an extension of the schemes proposed by Bross and others to the interference/source side information case. The result of this study shows that the proposed hybrid digital analog schemes are more robust to a mismatch in channel signal-to-noise ratio (SNR), than pure separate source coding followed by channel coding solutions. We then discuss applications of the hybrid digital analog schemes for transmitting under a channel SNR mismatch and for broadcasting a Gaussian source with bandwidth compression. We also study applications of joint source-channel coding schemes for a cognitive setup and also for the setup of transmitting an analog Gaussian source over a Gaussian channel, in the presence of an eavesdropper. In the next topic, we consider joint physical layer coding and network coding solutions for bi-directional relaying. We consider a communication system where two transmitters wish to exchange information through a central relay. The transmitter and relay nodes exchange data over synchronized, average power constrained additive white Gaussian noise channels. We propose structured coding schemes using lattices for this problem. We study two decoding approaches, namely lattice decoding and minimum angle decoding. Both the decoding schemes can be shown to achieve the upper bound at high SNRs. The proposed scheme can be thought of as a joint physical layer, network layer code which outperforms other recently proposed analog network coding schemes. We also study extensions of the bi-directional relay for the case with asymmetric channel links and also for the multi-hop case. The result of this study shows that structured coding schemes using lattices perform close to the upper bound for the above communication system models.

This thesis proposes a framework for combined source-channel coding under power and bandwidth constrained noisy channel. The framework is then applied to progressive image coding transmission using constant envelope M-ary Phase Shift Key (MPSK) signaling over an Additive White Gaussian Channel (AWGN) channel. First the framework for uncoded MPSK signaling is developed. Then, it’s extended to include coded modulation using Trellis Coded Modulation (TCM) for MPSK signaling. Simulation results show that coded MPSK signaling performs 3.1 to 5.2 dB better than uncoded MPSK signaling depending on the constellation size. Finally, an adaptive TCM system is presented for practical implementation of the proposed scheme, which outperforms uncoded MPSK system over all signal to noise ratio (Es/No) ranges for various MPSK modulation formats. In the second part of this thesis, the performance of the scheme is investigated from the channel capacity point of view. Using powerful channel codes like Turbo and Low Density Parity Check (LDPC) codes, the combined source-channel coding scheme is shown to be within 1 dB of the performance limit with MPSK channel signaling.

We study the problem of communicating a discrete time analog source over a channel such that the resulting distortion is minimized. For ergodic channels, Shannon showed that separate source and channel coding is optimal. In this work we study this problem for non-ergodic channels. Although not much can be said about the general problem of transmitting any analog sources over any non-ergodic channels with any distortion metric, for many practical problems like video broadcast and voice transmission, we can gain insights by studying the transmission of a Gaussian source over a wireless channel with mean square error as the distortion measure. Motivated by different applications, we consider three different non-ergodic channel models - (1) Additive white Gaussian noise (AWGN) channel whose signal-to-noise ratio (SNR) is unknown at the transmitter; (2) Rayleigh fading multiple-input multiple-output MIMO channel whose SNR is known at the transmitter; and (3) Rayleigh fading MIMO channel whose SNR is unknown at the transmitter. The traditional approach to study these problems has been to fix certain SNRs of interest and study the corresponding achievable distortion regions. However, the problems formulated this way have not been solved even for simple setups like 2 SNRs for the AWGN channel. We are interested in performance over a wide range of SNR and hence we use the distortion SNR exponent metric to study this problem. Distortion SNR exponent is defined as the rate of decay of distortion with SNR in the high SNR limit. We study several layered transmissions schemes where the source is first compressed in layers and then the layers are transmitted using channel codes that provide variable error protection. Results show that in several cases such layered transmission schemes are optimal in terms of the distortion SNR exponent. Specifically, if the band- width expansion (number of channel uses per source sample) is b, we show that the optimal distortion SNR exponent for the AWGN channel is b and it is achievable using a superposition based layered scheme. For the L-block Rayleigh fading M x N MIMO channel the optimal exponent is characterized for b < (|N - M|+1)= min(M;N) and b > MNL2. This corresponds to the entire range of b when min(M;N) = 1 and L = 1. The results also show that the exponents obtained using layered schemes which are a small subclass of joint source channel coding (JSCC) schemes are, surprisingly, as good as and better in some cases than achievable exponent of all other JSCC schemes reported so far.

Multiple-symbol detection of space-time differential codes (MS-STDC) decodes N consecutive space-time symbols using maximum likelihood (ML) sequence detection to gain in performance over the conventional differential detection scheme. However its computational complexity is exponential in N . A fast algorithm for implementing the MD-STDC in block-fading channels with complexity O(N 4) is developed. Its performance in both block-fading and symbol-by-symbol fading channels is demonstrated through simulations. Set partitioning in hierarchical trees (SPIHT) coupled with rate compatible punctured convolution code (RCPC) and cyclic redundancy check (CRC) is employed as a generalized multiple description source coder with robustness to channel errors. We propose a serial concatenation of the above with a differential space-time code (STDC) and invoke an iterative joint source channel decoding procedure for decoding differentially space-time coded multiple descriptions. Experiments show a gain of up to 5 dB in PSNR with four iterations for image transmission in the absence of channel state information (CSI) at the receiver. A serial concatenation of SPIHT + RCPC/CRC is also considered with space-time codes (STC) instead of STDC. Experiments show a gain of up to 7 dB with four iterations in the absence of CSI

Joint source-channel coding (JSCC) has been acknowledged to have superior performance over separate source-channel coding in terms of coding efficiency, delay and complexity. In the first part of this thesis, we study a hybrid digital-analog (HDA) JSCC system to transmit a memoryless Gaussian source over a memoryless Gaussian channel under bandwidth compression. Information-theoretic upper bounds on the asymptotically optimal mean squared error distortion of the system are obtained. An allocation scheme for distributing the channel input power between the analog and the digital signals is derived for the HDA system with mismatched channel conditions. A low-complexity and low-delay version of the system is next designed and implemented. We then propose an image communication application demonstrating the effectiveness of HDA coding. In the second part of this thesis, we consider problems in information hiding. We begin by considering a single-user joint compression and private watermarking (JCPW) problem. For memoryless Gaussian sources and memoryless Gaussian attacks, an exponential upper bound on the probability of error in decoding the watermark is derived. Numerical examples show that the error exponent is positive over a (large) subset of the entire achievable region derived by Karakos and Papamarcou (2003). We then extend the JCPW problem to a multi-user setting. Two encoders independently embed two secret information messages into two correlated host sources subject to a pair of tolerable distortion levels. The (compressed) outputs are subject to multiple access attacks. The tradeoff between the achievable watermarking rates and the compression rates is studied for discrete memoryless host sources and discrete memoryless multiple access channels. We derive an inner bound and an outer bound with single-letter characterization for the achievable compression and watermarking rate region. We next consider a problem where two correlated sources are separately embedded into a common host source. A single-letter sufficient condition is established under which the sources can be successfully embedded into the host source under multiple access attacks. Finally, we investigate a public two-user information hiding problem under multiple access attacks. Inner and outer bounds for the embedding capacity region are obtained with single-letter characterization. / Thesis (Ph.D, Mathematics & Statistics) -- Queen's University, 2007-09-28 23:11:21.398