| The channel error correction code, especial the method to decode the FEC is the main source of complexity in the wireless communication system so that studying the algorithm of decoding the FEC is necessary and meaningful.The LDPC code was firstly introduced by Gallager and reinvented by Mackay in 1999.Now it has become one of two most advanced FEC technology. The BP or SPA algorithm is well used to decode a LDPC code. However, with lineal decoding complexity, the LDPC also suffer from the performance issue due to the cycles on the Tanner graph. The sign of variable node requires more iterations computations to be converged during the decoding process compared with the Turbo code. In addition, the LDPC decoder costs more memory. These problems undermine the comprehensive performance of LDPC. Therefore the decoding algorithm of Belief Propagation with low complexity for the LDPC code is deeply learned in this paper.The classical process of decoding the LDPC code is studied and a new algorithm named ACMS is proposed in this paper by simplifying the SPA algorithm twice. In this new algorithm, only one variable-node with the exact minimum propagation is used in the check-node iteration. Several alteration and optimization is taken in the algorithm to improve the BER performance especially in the low SNR situation.The quantization method is also applied to the new algorithm. The node messages contributed to the new algorithm are all quantized to reduce the hardware expense. The simulation result shows the new algorithm is of low complexity but keeping BER performance.A new algorithm named Back-Tracking Variable-based Scheduling Belief Propagation is proposed in this paper at last. The new algorithm alters the sequence of edge message update in the Tanner graph. The simulation result also proves it is not only better than the traditional serial scheduling on BER performance but also cost less hardware expense compared with dynamic scheduling. It is a good mean method. |
PDF Full Text Request