ON THE APPLICATION OF APPROXIMATE APP DECODING TO DIGITAL VIDEO TRANSMISSION

For transmitting compressed digital video, we propose using threshold decodable block codes with large block length, and a posteriori probab ility (APP) soft decision decoding. A new approximation to the weight function associated with APP soft decision decoding of threshold decod able codes is presented. When the number of components in the parity e quations is large, it is demonstrated that the new method gives excell ent error performance, whereas there is a substantial degradation in t he performance of the least reliable symbol approximation presented by Tanaka et al. and others [3]-[5]. The effect of feedback on the perfo rmance of the APP decoder is also analyzed. It is shown that if the pe rformance criterion is word error rate rather than bit error rate, fee dback of previously decoded bits is essential to obtain all possible c oding gain from the soft decision decoder. Finally, the performance of the proposed coding scheme is compared to the performance of a concat enated coding system with the same rate.