Modified BP decoding algorithm based on oscillation and check update
基于消息振荡及校验更新的改进BP译码算法

To reduce the average number of decoding iterations in the region of low SNR (Signal-to-Noise Ratio) and weaken the oscillation of the LLR (Logarithm-Likelihood-Ratio), the variation law of the check nodes' LLR and the checksum was discovered, in the error frame of the short and middle length LDPC codes. Based on the law, an improved BP algorithm of LDPC decoding was proposed. In the algorithm, an early stopping criterion was developed to reduce the average number of decoding iterations in the region of low SNR (Signal-to-Noise Ratio). And to improve the decoding performance by weakening the oscillation of the LLR, the update algorithm of check node was modified. Simulation results show that, the algorithm reduces the number of iteration while the performance is not reduced in the region of low SNR, the decoding performance is improved while the number of iteration is not changed in the region of middle and high SNR.