EBDN: Entropy-Based Double Nonuniform Sensing Algorithm for LDPC Decoding in TLC nand Flash Memory

Low-density parity-check (LDPC) codes are widely employed in NAND flash memory to improve the reliability of data. However, LDPC has serious latency problems when the raw bit error rate (RBER) is high. The reason is that not only a sufficient sensing level is required to obtain accurate soft information but also a high number of iterations are needed. To reduce the latency of LDPC, an entropy-based double nonuniform (EBDN) sensing algorithm is proposed in this article. The basic idea of this algorithm is to exploit the entropy to quantify the nonuniformity of intrastate and interstate. And the sensing levels are placed in a targeted manner based on the entropy, thereby significantly reducing the number of sensing levels without reducing the LDPC error correction performance. The experimental results show that the proposed algorithm can decrease the number of iterations of LDPC by approximately 70% and reduce the read latency by 34.52% compared with the traditional nonuniform sensing algorithm.