A New Finite Word-length Optimization Method Design for LDPC Decoder

摘要

A new word-length optimization method based on Monte Carlo simulation is proposed. The word-length of the check node extrinsic message is also further optimized in this paper. In the proposed optimization method, and in the process of optimizing the word-length of the channel data, the statistical distribution results of variable node's posterior probability data and check node's extrinsic message are also obtained. The optimized word-length of variable node's posterior probability data and check node's extrinsic message is concluded by the statistical distribution result and the BER (Bit Error Rate) curves. Compared to the pure Monte Carlo simulation, the proposed method could reduce the amount of simulation work by more than 50%, and have the same word-length optimization results.