Improved trellis-based decoder for non-binary LDPC codes

In this paper, we use a column-wise, row-wise approach for efficiently choosing reliable messages in decoding non-binary Low-Density Parity-Check (LDPC) codes, and build configurations with those reliable messages for updating check nodes. This method decreases the number of configurations compared to conventional trellis-based decoding algorithms. In addition, a novel algorithm with precomputed messages is proposed which improves decoding performance. Using the reduced configuration set with the improved message updating scheme it is shown that we can achieve an improved decoding performance while reducing the hardware complexity. Compared to trellis-based decoders, the proposed decoder reduces the number of required operations in computing check-to-variable messages and achieves better error-correcting performance.