Joint iterative estimation and decoding for 16-QAM BICM over correlated fading channels

Joint iterative channel estimation and decoding for 16-QAM bit-interleaved coded modulation (BICM) is considered over frequency-flat fading channels. Estimates of the complex channel gain are derived first from known pilot symbols by optimum Wiener filtering and passed to the BICM iterative decoder as initial channel state information. After each decoding iteration, the channel estimates are refined using the extrinsic information fed back from the decoder. An advanced scheme called "restart" is developed to effectively reduce the bit error rate floor, by restarting the decoder after several iterations with the refined channel gain estimates. Motivated by the observation of the burst errors in the transmitted sequence, the "restart" algorithm is extended by concatenating Reed-Solomon codes with BICM. When RS (255, 223) code is used for BICM with 16-QAM modified set partitioning labeling and a 16-state code, the BER of 10/sup -6/ can be achieved at E/sub b//N/sub 0/ of 10.5 dB for normalized fade rate f/sub d/T/sub s/ = 0.02.