Trapping sets in stochastic LDPC decoders
Kuo-Lun Huang, Vincent Gaudet, Masoud Salehi · 2015
Stochastic decoding is a hardware and energy-efficient approach to implement iterative Low-Density Parity-Check (LDPC) decoders. However, stochastic decoding often has a performance loss compared to the sum-product algorithm and suffers from additional trapping sets, which limit the performance of LDPC codes in the error floor region. In this paper, we investigate the error characteristics of stochastic LDPC decoding. We also report stochastic-decoding-specific trapping sets for the (1056,528) LDPC code from the WiMAX standard. This observation has a potential to guide code structure designs that lower the error floor in stochastic decoding.