Effective lazy schedule for the layered belief propagation algorithm

Abstract

This study presents an effective lazy schedule (LS) for the layered belief propagation (LS-LBP) algorithm; it significantly reduces the number of iterations and the error floors of decoding. This novel LS forces a check node into a sleep status when the reliabilities of its neighbour variable nodes exceed the threshold, and the check node will turn to awake as all the nodes are in sleep status. This LS reduces latency by ignoring useless check-node processes, and it also achieves a low error rate by overcoming some trapping sets. The simulation experiment shows that when compared to the classical layered belief propagation for world interoperability for microwave access (WiMAX), random codes constructed using the progressive edge-growth (PEG codes) algorithm, and digital video broadcasting-satellite-second generation (DVB-S2) codes, this LS increases the decoding convergence speed by up to 36% to achieve the bit error rate 1 × 10–6 and reduces the error floors to 20–45% for the PEG and DVB-S2 codes. Compared with other schedules for the layered belief propagation, the LS-LBP algorithm is more suitable for the codes with a long block length and has lower complexity.