Abstract: A new dynamic threshold truncation strategy based on the distributions of the message-vector reliability is designed for solving the high complexity problem of non-binary LDPC decoding algorithms. The presented strategy dynamically selects the truncation threshold based on the difference between the maximum and the sub-maximum reliability values, which can reduce the finite filed elements involved in message computing. In the iterative process, the decoding messages are determined by the truncation threshold, which can effectively reduce the number of the states and branches in the trellis, resulting in lower decoding complexity in average. A dynamic threshold extended min-sum (DT-EMS) algorithm is further presented based on the new truncation strategy. Simulation results show that the proposed DT-EMS algorithm performs almost as well as the well-known Q-ary sum-product algorithm (QSPA) and the T-EMS algorithm. Moreover, the proposed algorithm achieves lower decoding complexity than the T-EMS algorithm and has much lower complexity than the QSPA .

Key words: non-binary low-density parity-check codes, extended min-sum algorithm, dynamic threshold, truncation strategy