A Self-Adaptive Decoding Scheme for BICM-ID Embedded Turbo Codes

This paper proposes a self-adaptive decoding scheme based on cross-entropy (CE). Good BER performance can be achieved with lower complexity. The fixed minimum iterative number scheme has been proposed to modify the shortage of the traditional stopping criterion scheme, whose calculations of CE expend hardware resource too much. However, iterative numbers which correspond to different SNRs are varied. It is improper to set a unique minimum iterative number (Imin) for all SNRs. In order to further improve the performances, this self-adaptive decoding scheme is proposed. For various SNRs, different minimum iterative numbers, which are determined by the statistics of the iterative numbers, should be set instead of being unique. The SNR criterion is also adopted in this scheme to further reduce the complexity. Simulation results confirm that the iterative receiver with this self-adaptive scheme has achieved a similar performance with the fixed iterative number scheme but it is of a considerably reduced complexity due to the decreased iterative number of decoding and fewer calculations of CE.