Impact of soft channel construction on iterative channel estimation and data decoding for multicarrier systems

To apply the promising idea of iterative information processing to the problem of joint channel estimation and data decoding, one may utilize the soft information of the coded bits to help refine the channel estimate, e.g., with Wiener filtering. However, the performance of such an iterative approach depends on not only the construction of the iteratively refined soft channel, but also the structure of the adopted error-correction code. Through extrinsic information transfer chart analysis, we investigate the impact of different soft channel construction schemes on multicarrier systems possessing turbo coding/modulation structures: (1) one with bit-interleaved coded modulation signals, and (2) the other with low-density parity- check code encoded modulation signals. It is shown that the former benefits significantly from iterative information processing regardless of channel construction scheme and the performance of the systems with different schemes converges. On the contrary, the same observation cannot be applied to the latter, a capacity- approaching system.