A method for decoding error-correcting codes such as LDPC codes and repeat accumulate codes normalizes messages generated by a bit node processor, and normalizes messages generated by the check node processor. This decoding has significantly improved performance when compared with conventional min-sum and normalization min-sum decoding methods. At the same time, the 2D normalization min-sum decoding has a similar and better performance than BP decoding method, in waterfall and error floor regions. Furthermore, the 2D normalization min-sum decoding method requires much less computational complexity than conventional BP decoding. The 2D normalization min-sum decoding can also be extended to 2D offset min-sum decoding.
展开▼