Decoding of non-binary low-density parity-check codes based on the genetic algorithm and applications over mobile fading channels

In this study, an efficient decoding algorithm is proposed to decode non-binary low-density parity-check (LDPC) codes. The algorithm is derived from the belief-propagation (BP) decoding with genetic algorithm of binary LDPC codes. For the proposed algorithm, two decoding constraints are introduced to determine variable nodes which are considered highly reliable. The messages from these highly reliable variable nodes are then magnified with an appropriate parameter γ. This process can make the messages propagate in the Tanner graph more efficiently. Simulation results show that, compared with the fast Fourier transform-based BP algorithm, the proposed algorithm can have an equal or better performance in low bit-error-rate region over both the additive white Gaussian noise channels and the mobile fading channels.