For wireless communication under bandwidth constraints, data compression is necessary to get high transmission efficiency. However, the codestreams after compression become more sensitive to channel errors. Thus, strong error protection is required for transmission over wireless channels with high bit error rate and time-varying characteristics, and this lowers the transmission efficiency. In conventional communication systems, source coders and channel coders are designed separately, which fails to get a balance between the efficiency and reliability. Joint source and channel coding/ decoding (JSCC/ JSCD) method has been proposed to solve the problem. Taking both the source characteristics and channel state information into account, it makes full use of resources such as bandwidth and power to obtain a good and stable end-to-end transmission performance. In this thesis, joint source and channel coding/decoding techniques and their applications in wireless image transmission and the detection of signals with low signal-to-noise-ratio are studied. The main work can be summarized as follows:1. Joint source and channel rate allocation. A unified formulation is presented for optimal rate allocation of embedded source coding and rate compatible channel codewords under bandwidth constraints. Then a fast algorithm using channel operation region division and two-way rate selection is proposed, which is fit for both the case of fixed-length information blocks and the case of fixed-length channel codewords. Simulation results show that the proposed algorithm can get a similar transmission performance as other rate allocation algorithms, while its complexity is lower. For adaptive joint source and channel coding (JSCC) systems, the sliding window rate allocation algorithm is improved so that the division of channel operation regions and selection of channel code rates are more accurate, and the transmission performance is better.Further, based on embedded coding and Turbo Trellis-Coded Modulation (T-TCM), an adaptive joint source and channel coding/ modulation system is proposed for the progressive image transmission over bandwidth and power constrained noisy channels. By using the upper bound distortion method, and according to the monotonously nonincreasing characteristics of distortion-rate function of the embedded codestreams, the original goal function of minimizing the average end-to-end distortion is converted to maximizing the average throughput, simplifying the solving process of optimal channel signal-to-noise ratio switching thresholds and adaptive transmission parameters.2. Joint source and channel decoding. Two joint source and channel decoding (JSCD) methods for variable length codes, i.e. source controlled channel decoding (SCCD) and iterative source and channel decoding (ISCD), are studied and compared. And the effects of source residual redundancy and free distance of variable length codes on the decoding performance are deeply studied. Then, for the systematic channel coding, the structure of ISCD method is improved. The information bits are extracted and soft decoded, the likelihood ratio information of which is sent to the channel decoder as the a priori information at the first iteration. The structure combines SCCD and ISCD method in a flexible way, and the decoding convergence is accelerated. Finally, simulation results of the transmission and decoding of images prove the effectiveness and feasibility of the joint source and channel decoding methods. |