Improving the Thresholds of Generalized LDPC Codes With Convolutional Code Constraints

CC-GLPDC codes are a class of generalized low-density parity-check (GLDPC) where the constraint nodes (CNs) represent convolutional codes. This allows for efficient decoding in trellis with forward-backward algorithm, and strength component easily can be controlled by encoder memory without changing graph structure. In this letter, we extend CC-GLDPC introducing different types irregularity at CNs investigating their effect on BP MAP thresholds binary erasure channel (BEC). For considered codes, an exhaustive grid search is performed to find BP-optimized MAP-optimized ensembles compare regular ensemble same design rate. The results show that significantly improve thresholds, whereas only slightly from ensembles. Simulation AWGN presented as well compared corresponding thresholds.