A new forced convergence decoding scheme for LDPC codes
Jianxiao Fan, Hongwen Yang · 2009
In this paper, we present a new forced convergence decoding scheme for LDPC codes. We remove the early detected variable nodes from the Tanner graph and thus the parity matrix shrinks iteration by iteration and the decoding complexity can be reduced. When the parity check matrix shrinks to zero before the maximum iteration, we get a chance to detect the wrong deletion and, for the decodable codewords, most of such failure can be recovered by restoring the original Tanner graph and resuming the decoding in the remaining iterations. Simulations results indicate that the proposed method can reduce the decoding complexity significantly while keep the error rate performance unchanged or even better.