A Belief Propagation Algorithm with Set-Breaking to Lower Error-Floors of Low-Density Parity-Check Codes

In order to more exactly describe the errorfloor phenomenon in the iterative decoding of Low-density parity-check (LDPC) codes, a modified concept of the stable trapping set is introduced. Based on this new concept, an improved belief propagation algorithm with setbreaking mechanism is proposed to lower the error-floors of LDPC codes. Message ranking of the bit nodes in the stable trapping sets will be greatly lowered than that of other bit nodes in the iterative decoding process. By using this characteristic to label the bit nodes in the set, the corresponding initial log likelihood ratios will be flipped to break the stable trapping set and restart to decode. Simulation results verify the validity of the proposed algorithm.