Serial Decoding Algorithm with Continuous Backtracking for LDPC Convolutional Codes

Two main decoding algorithms are usually used for low-density parity-check convolutional (LDPC-C) codes, belief propagation algorithm and on-demand variable node activation algorithm. However like the decoding for block codes, the iterations of these algorithms cannot utilize the messages of newly joined variable nodes within the next iteration windows continuously, limiting the decoding performance. Thus a serial decoding algorithm with continuous backtracking for LDPC-C codes is proposed. Based on the iteration windows, the proposed algorithm performs advance iteration with two adjacent iteration windows and then backtracks to the preceding iteration window. During each iteration the latest updated and newly joined messages are utilized. The whole iteration uses the backtracking continuously until all the results output. Different LDPC-C codes are used to test the performance of proposed algorithm. Simulation results show that the proposed algorithm outperforms other algorithms and requires less number of iterations, improving the decoding convergence rate.