Convolutional Code And Turbo Code Source And Channel Decoding

Shannon's classic separation theorem states that, under ideal conditions, source coding and channel coding can be treated separately without sacrificing any performance for the whole system. However, this theorem holds true only under ideal conditions i.e., infinite source code dimension, infinite long channel code and unlimited delay. Unfortunately, practical communication systems do not meet such requirements. Therefore joint source and channel coding has become the hot issue in coding research areas, since information redundancy can be utilized sufficiently to improve the decoding performance.With the integration of wireless technology and multimedia service, the reliable transmission of high compression data has become the key technology of Next-Generation Mobile Communications. In this thesis, different schemes of joint source-channel decoding for variable length codes transmission over noisy channels are proposed.Multimedia transmission over time-varying channels such as wireless channels has recently motivated the research on the joint source-channel technique. In this thesis, we present a method for joint source-channel soft decision decoding of variable-length encoded multiple sources. By exploiting the a priori redundancy in multiple sources, the decoding performance is greatly improved. Compared with the single source decoding scheme, the proposed technique is more practical in wideband wireless communications. Our method generalizes the conventional Viterbi decoding algorithm, not only the optimal path in channel decoding trellis but also the a priori probability of branches in multiple VLC source trees are considered. The Simulation results show that our new method obtains the gain in SNR of around 1.5dB, compared with the conventional Viterbi algorithm that not exploits the a priori information.Since Variable-Length Codes (VLCs) are inherently vulnerable to channel errors, traditional error-correcting codes are unable to solve the propagation-error problem efficiently. A new error-correcting scheme based on symbol-constrained joint source-channel decoding is presented for VLCs with double-tree construction. By exploiting the a priori redundancy in source at the channel decoder, the decoding performance is greatly improved. Meanwhile, with the construction of double-tree VLCs, propagation-error phenomenon is alleviated. Moreover, by transforming the optimal MAP algorithm to symbol-constrained joint source-channel decoding (JSCD), the symbol error rate (SER) is reduced. Simulation results demonstrate that for VLCs JPEG source over AWGN channels, compared with bit-constrained MAP algorithm and symbol-constrained MAP algorithm, the gain of our method in SNR is around 0.5dB and 0.7dB respectively when SER reaches 0.01.Turbo code has gained outstanding decoding performance either in AWGN channels or in Rayleigh fading channel because it integrated the advantage of block code and convolutional code. Joint source and channel iterative decoding of VLCs for Turbo codes is proposed in this thesis. By utilizing the source redundancy at the iterative decoder, the decoding performance is improved greatly. JPEG source is used in this thesis to demonstrate achievable performance improvements.