Improved decoding of LDPC codes by variable-to-check residual belief propagation

This paper proposes an improved variable-to-check residual belief propagation algorithm, called Lazy-RBP, in which a new informed dynamic schedule (IDS) combined with the lazy schedule (LS) is adopted. Firstly, Lazy-RBP exploits the probabilistic girth-based approach, one of the LS strategies, to decide the set of updated variable nodes in each iteration. Then among this set, a new IDS scheme is introduced, where both the unreliable variable nodes and the residual of the variable-to-check messages are utilized to locate the message to be propagated first. Simulation results show that the Lazy-RBP algorithm can not only improve the error performance but accelerate the convergence compared with the VC-RBP algorithm.