The Research Of NB-QC-LDPC In Nonlinear Fiber Channel

Compared with binary quasi-cyclic low-density parity check (B-QC-LDPC) code, non-binary quasi-cyclic low-density parity check (NB-QC-LDPC) code can achieve better error correction ability at the cost of high decoding complexity. This is due to the complexity of updating checking nodes and the requirement of large memory space. In order to reduce the computational complexity, we introduce modified log Fourier sum-product algorithm (LFSPA) which is based on the linear approximation with a scaling factor. We also study the performance of the proposed algorithm in nonlinear fiber channel.In this paper, we first study the B-QC-LDPC code. Based on B-QC-LDPC code, we describe several algorithms to construct NB-QC-LDPC code, which includes Progressive Edge Growth (PEG) algorithm and Extended PEG algorithm. Then we discuss how to optimize NB-QC-LDPC code.Further, we study the performance of LFSPA in AWGN channel. For non-binary LDPC code, there are many multiplications in the decoding algorithm. The log domain sum-product algorithm (LSPA) transforms the multiplication into addition and reduces the complexity. The LFSPA further reduces the complexity of convolution in LSPA. Based on the LFSPA, we propose linear approximation with a scaling factor which could reduce the complexity of addition in the log domain. In order to analyze the performance of the proposed approximation, we calculate the threshold of proposed approximation and find that the performance degradation caused by the proposed approximation is negligible.At last, we study the performance of the proposed algorithm based on LFSPA in nonlinear fiber channel. We firstly design the simulation model of nonlinear optical fiber channel through the simulation software. After signal is transmitted through this model, we use the decoding algorithm based on training sequences to decode the signal. Then we analyze the performance of proposed algorithm when channel parameters are different.