In this paper, a novel joint source channel (JSC) decoding technique is presented. The proposed approach enables iterative decoding for serially concatenated arithmetic codes and convolutional codes. Iterations are performed between Soft Input Soft Output (SISO) component decoders. For arithmetic decoding, we proposed to employ a low complex trellis search technique to estimate the best transmitted codeword and generate soft outputs. Performances of the presented system are evaluated in terms of PER, in the case of transmission across the AWGN channel. Simulation results show that the proposed JSC iterative scheme leads to significant gain in comparison with a traditional separated decoding. Finally, the practical relevance of the proposed technique is validated under an image transmission system using the SPIHT codec.