Joint Iteration Belief Propagation Decoding for Generalized Polar Coded BICM System

Bit-interleaved coded modulation with soft iterative decoding (BICM-ID) system has desirable performance with many channel codes. But for polar code, there are at least two problems to overcome: the high decoding complexity and the undesirable correlation among the decoded bits. The former prohibits its application in practical engineering, while the latter engenders performance degradation for no iterative gains. To solve these problems, a joint iteration BP decoding algorithm based on generalized BICM-ID scheme is presented. In this algorithm, we combine the demodulation and decoding into one factor graph so that the noise corrupted symbols can be simply seen as the extended information nodes of BP decoder. Further, asymptotic performance bound of the proposed system is calculated to analyze its advantages and drawbacks. Simulation results show that the proposed scheme can significantly improve the performance of the unmodified BIPCM-ID scheme with lower complexity.