Convergence analysis of iterative soft decoding in partial response channels

In this paper, we apply density evolution to investigate the convergence of iterative soft decoding in partial response (PR) channels and to predict the bit-error-rate (BER) performance of concatenated PR systems. The convergence analysis and computer simulations reveal that column weight-2 low-density parity check (LDPC) codes are similar to recursive systematic convolutional (RSC) codes in terms of extrinsic information transfer characteristics, BER performance, and block-error statistics, while the decoding complexity of column weight-2 LDPC codes is significantly smaller than that of RSC codes.