On the Non-concentrated Check-node Degree Distributions in QC-IRA Raptor-like LDPC Codes

The quasi-cyclic irregular-repeat-accumulate (QC-IRA) raptor-like LDPC codes in ATSC 3.0 have concentrated check-node degree distributions in the core matrix and extended matrix as well. Actually, the low-rate raptor-like LDPC codes with non-concentrated check-node degree distributions show a promising performance improvement. In this paper, the benefits of non-concentrated check-node degree distributions are first analyzed. After that, a genetic-like algorithm is proposed to search good pairs of non-concentrated check and variable-node degree distributions. Simulation results show that the proposed low-rate LDPC codes with QC-IRA raptor-like structure have at most 0.22 dB gain over those in ATSC 3.0 when bit-error rate (BER) is at 1$0^{-4}$ level over additive white Gaussian noise (AWGN) channels using QPSK modulation.