Study Of Joint Source Channel Coding Methods

Joint source channel coding (JSCC) method considers both source coding and channel coding which are designed and run independently. It optimizes different parts of communication system jointly to get the integrated optimization. Jointing source channel coding scheme based on unequal error protected and intra MB update (β) are investigated in this paper.Firstly, the paper gives a brief overview of the necessary background knowledge involved in designing JSCC scheme. Then the JSCC scheme based on RCPC unequal error protected is investigated. It shows that the compressed bit-streams can be classified according to their importance property. Each part of bit-streams can be protected by using different channel coding because they are influenced differently in channel. Unequal error method is done. Rate compatible punctured convolutional code—RCPC is adopted as the channel coding to carry out the UEP. The experimental results show that this scheme can get higher picture reconstructed performance than that of using equal error protected method (EEP).After analyzing disadvantages of the jointing scheme based on UEP, another joint scheme based on the update of intra MBs (β) is investigated. In this scheme, influence coming from different parts of communication system is considered, then the source coding rate-distortion estimated model (D_s) and channel coding statistical model (D_c) are got. Finally the end to end distortion model D = D_s + D_c is got. Based on this estimated model, new algorithms in the selection of optimal intra MB update rate β and the selection of intra MBs' position are proposed in this paper. In order to minimize the end-to-end distortion, optimal β is obtained by using the mathematic method in this paper. Then according to the MB's activity complexity, the coding mode of current MB is fixed on. Experimental results show that the improved algorithms can increase the PSNR more than the original scheme. Moreover, the reconstructed picture qualities are also better than before.