Abstract
In recent years, low-density parity-check (LDPC) codes catch high attentions because of its near Shannon limit performance with sum-product decoding algorithm. LDPC codes are regarded as the main competitor of turbo codes. However, sum-product algorithm is a high hardware complexity algorithm. This restriction makes LDPC codes hard to realize. In this thesis, we propose a low hardware complexity algorithm to realize the decoding of LDPC code by modelling the AWGN channel as a noisy erasure channel. In the proposed decoding algorithm, we basically use all-but-one rule to decode erasures while carefully managing the possibly noisy hard decisions. Since this algorithm takes only very simple operation units and does not need a huge number of memories, it can be applies to decode very long codes, which may be attractive in practical situations.