A Multivariate LDPC Decoding Simplification Method and Performance Simulation Analysis
Xiaojun Wang, Yang Cao · 2023
Aiming at the problem of large computation and high decoding complexity when updating the check node in the LLR-BP algorithm of multicode LDPC code, a simplified method of dimensionality reduction in accordance with the weights is proposed, in which the q-dimensional likelihood ratio information of a variable node is descending-ordered and a weight is set for each dimension of the node, and the dimensional information of two variable nodes will be summed up according to the weight value size set in the process of updating check node. The size of the weight value will be added according to the set rule, which effectively reduces the wasted addition operation resources on the low likelihood ratio information. The simplified algorithm reduces about 90% of computation compared to the LLR-BP algorithm during one iteration of the algorithm. The BER performance simulation of the algorithm shows that the simplified decoding algorithm has only a very small difference in performance compared to the LLR-BP algorithm, and the overall decoding iteration speed is greatly improved.