An improved PEG construction of irregular LDPC codes by conditioning low-degree nodes

Progressive Edge-Growth (PEG) algorithm is an important approach to construct good LDPC codes at short block lengths. PEG construction with optimized symbol-node degree distributions yield irregular codes with very good performance, especially in the water-fall region. In this paper, we present a modification to the PEG algorithm by conditioning low-degree symbol nodes in the Tanner graph. Our modified algorithm yields irregular codes with improved girth histogram at low-degree symbol nodes. Simulation results show that the modified algorithm can lower the error floor of irregular PEG codes considerably, while having slightly better performance in the water-fall region.