Joint source-channel coding via turbo codes

In this thesis, we investigate three joint source-channel coding issues in the context of Turbo codes.; In the first part of the thesis, a robust soft-decision channel-optimized vector quantization (COVQ) scheme for Turbo-coded additive white Gaussian noise (AWGN) and Rayleigh fading channels used in conjunction with binary phase shift keying (BPSK) modulation is proposed. The log-likelihood ratio (LLR) generated by the Turbo decoder is exploited via the use of a q-bit scalar soft-decision demodulator. The concatenation of the Turbo encoder, modulator, AWGN channel or Rayleigh fading channel, Turbo decoder, and q-bit soft-decision demodulator is modeled as an expanded discrete memoryless channel (DMC), or a discrete block-memoryless channel (DBMC). A COVQ scheme for these expanded discrete channel models is designed. Numerical results indicate substantial performance improvements over traditional tandem coding systems, and COVQ schemes designed for hard-decision demodulated Turbo-coded channels (q = 1).; In the second part of the thesis, we first design systematic Turbo codes for the reliable communication of non-uniform i.i.d. sources over BPSK modulated AWGN and Rayleigh fading channels. One main objective is to achieve the best possible performance vis-a-vis the Shannon limit. The redundancy in the form of non-uniformity is exploited in the decoder. The encoder structure, on the other hand, is optimized to further enhance the performance. We prove that for recursive convolutional encoders, under certain (necessary and sufficient) conditions, the state and output marginal distributions are asymptotically uniform, irrespective of the degree of non-uniformity in the source distribution.; The third part of the thesis investigates the design of Turbo codes for Markov sources to exploit the source redundancy in the form of memory. In the first constituent decoder, the decoding algorithm is modified by incorporating the Markovian property of the source. Due to interleaving, the second constituent decoder is designed for memoryless sources. However, when the source is asymmetric, it can also take advantage of the source redundancy in the form of non-uniformity. Furthermore, the extrinsic information from the second decoder is modified via a weighted correction term. Simulation results demonstrate substantial gains (from 1.29 up to 3.57 dB) over the original Turbo codes' performance. As a result, the OPTA gaps are in the range of 0.94--1.45 dB. (Abstract shortened by UMI.)...