Research On Joint Source-Channel Coding Using LDPC Code

With the development in the fields of computer, telecommunication, networking and multimedia, etc, image communication has been studied extensively as an essential component of multi-media communications. However, separation approach of the source and channel coding based on Shannon's separation principle often leads to low efficiency especially for image and video transmission. In real communication systems, whose channel characteristics are known, it will benefit a lot to adopt joint design of source and channel coders.This paper mainly composes of research on joint source-channel coding (JSCC). Firstly, the background and significance of joint source channel coding over wireless image transmission are elaborated, and the present situation and application foreground of this research task are analyzed. Then, source coding and channel coding, are briefly introduced in terms of their basic concepts, primary principles and common methods. Joint source and channel coding theory is also presented, which includes its necessity and advantage, its framework and objective function optimization, its application and classification. With the development of communications networks, the joint source channel coding will be applied to extensively various fields.After that, basic theory of LDPC code is systematically expounded, and the sum-product decoding and its simplified form, the min-sum algorithm, are justified, and brief introduction of density evolution theory is made. It should be noticed that the theoretical results is consistent with the actual case only when code length is infinite, which makes LDPC code finite analysis a research focus.In this paper, a joint coding scheme for image transmission is proposed, which realizes joint source and channel coding using LDPC code. This scheme utilizes the wavelet-based SPIHT codes for source coding and LDPC code for channel coding, which provides unequal protection for SPIHT bit stream based on its importance to improve its performance in the case of unexpected errors. Simulation results show that: this method can conquer the limitation that the corresponding embedded compressed bit streams are subject to severe error sensitivity. Experimental results also illustrate that LDPC code can greatly improve the quality of the reconstructed image under conditions of different code length and bit rate, and when bit-error-rate is the same, unequal error protection scheme can reach greater coding gain than equal error protection scheme for static images.