FFT-based parallel encoding algorithm for structured LDPC codes

Structured LDPC, such as RA, IRA, ARA, QC-LDPC etc., are some important LDPC codes, these codes usually have very good performance and facilitate the implementation of encoding and decoding for the structural features they have. A new method for the encoding of the structured LDPC codes has been presented, namely FFT-based parallel encoding algorithm, which uses FFT to realize the parallel encoding of LDPC codes by achieving the circular convolution in the frequency domain. The study shows that the proposed algorithm has lower computational complexity than the previously proposed encoding algorithm. Although transforming to the frequency domain requires complex operations in real domain which increasing the number of the quantization points, the throughput of the encoder has greatly improved because of the increase of the degree of parallelism. The simulation results show that the computational complexity of the FFT-based encoding algorithm has an approximately linear relationship with the size of sub-circulate matrix, and when this size goes larger, the newly proposed algorithm will possess a lower computational complexity and a higher throughput compared with the previously proposed algorithms.