LAEPS: LDPC LLR Adaptive Estimation Based on Pattern Set Distribution Variation in 3D Charge Trap NAND Flash Memories
Qianhui Li, Yiyang Jiang, Qi Wang, Yang Liu, Zexia Wang, Xiaolei Yu, Jing He, Zongliang Huo · 2021 IEEE International Conference on Integrated Circuits, Technologies and Applications (ICTA) · 2021
Low-density parity-check (LDPC) code with the soft decoding algorithm has been widely used in 3D NAND flash memories to improve the system reliability, because of its strong error correction ability. However, the accuracy of the soft information represented by the Log-Likelihood Ratios (LLRs) is an important factor which affects the error correction ability of the LDPC decoder. To improve the accuracy of LLRs, a novel LLR Adaptive Estimation scheme based on Pattern Set distribution variation (LAEPS) in 3D charge trap NAND flash memories is presented in this paper. LAEPS estimates the LLR based on not only the information proved by multi-read but also the information proved by adjacent cells in the same channel. Experimental results by silicon test illustrate that LAEPS increases the correction capability of LDPC soft decoding to 1.4x, and decreases the decoding iteration and RBER to about 0.7x and 0.8x.