Double decision residual confidence propagation decoding algorithm for LDPC code

Abstract:

In order to solve the problems of greediness and silent variable nodes in multiple dynamic scheduling residual belief propagation (RBP) decoding algorithms for low-density parity-check (LDPC) codes, a double decision residual belief propagation decoding algorithm (DD-RBP) is proposed by introducing the method of joint decision based on the check equation and probability residual value. Firstly, the variable nodes that need to be updated are selected according to the calculated probability residual value, which can reduce the number of silent variable nodes. Then, according to the results of the check equation, the residual values of the corresponding edges of the relevant check nodes are updated to further reduce the greediness. Finally, combined with the updated residual value, in the set of edges connected by the variable nodes that need to be updated, the edge with the largest residual value is selected locally or globally and updated, and the above process is repeated until the maximum number of settings is reached. Theoretical and simulation analyses demonstrate that, for low-density parity-check codes under IEEE802.16e standard and 5G NR standard, the proposed double decision residual belief propagation decoding algorithm performs better than other algorithms in decoding under additive Gaussian white noise channels and Rayleigh fading channels with increased complexity.