電子情報通信学会ソサイエティ大会講演要旨
B-8-32
Shuffled Belief Propagation Parametric LDLC Decoding
Brian M. Kurkoski・○△Warangrat Wiriya(JAIST)
This paper proposes the shuffled belief propagation (SBP) parametric LDLC decoding. We define the reliability of the check-to-variable messages for the sequence to update of the variable nodes. The parametric SBP is proposed for increasing the convergence
speed and improving the performance of BP decoder . The updating sequence of SBP follows the reliability of the check-to-variable messages from high to low. The numerical results shows that the proposed algorithm gives the superior performance in terms of VNR and also provides lower number of iteration for convergence compared to BP decoding. At a probability of symbol error equal 10^−4 and n = 100 and 1000, the
proposed algorithm gains 0.2 dB.