Optimization of the iterative decoding algorithms for irregular block codes

In this paper, we analyze the performances of iterative decoders for linear block codes. In particular, we consider two modifications of the gradient-descent bit flipping (GDBF) algorithm with momentum, where multiple component decoders with different momentum values are concatenated to improve the decoder performance. The learning parameters of the component decoders are obtained by using a Genetic algorithm based on the database of the uncorrectable error patterns of the previous decoder. We present three optimization strategies and provide a comparison with the state-of-the-art decoders. The numerical results are presented on short Bose-Chaudhuri-Hocquenghem (BCH) codes and the channel with additive white Gaussian noise (AWGN).