An Adaptive Variable Iterative Decoding Scheme for BICM-ID

Iterative decoding is an effective method to overcome the drawback of Bit-interleaved coded modulation (BICM) over AWGN channels. Since the performance improvement will be very little after sufficient iterations, several stopping criteria have been devised to stop the unnecessary iterations. However, all the existing stopping criteria are designed for turbo-decoder, not for BICM-ID. In this paper, a modified cross-entropy (CE) criterion and sign change ratio (SCR) criterion are considered to stop the unnecessary iterations. An adaptive iteration scheme for BICM-ID based on CE and SCR is proposed. This adaptive iteration scheme uses SCR criterion at low-SNR situations to reduce the computation complexity and CE criterion at high-SNR situations to keep the system performance. Simulation results show that the proposed scheme reduces remarkable iterations compared to CE scheme without noticeable performance loss and achieves better performance than SCR scheme with few iterations increase. Furthermore, this proposed scheme performs negligible difference from the fixed iteration scheme with much less iterations.