| Solid-state drives(SSDs),which consist of NAND flash memory chips,are widely used today as a primary medium of data storage.The use of advanced manufacturing technology and multi-level cell(MLC)storage technology has effectively increased the storage capacity of NAND flash memory,while seriously affecting storage reliability,and making flash channel noise increasingly intensified.In order to cope w ith the severe and severe flash memory channel environment,the combination of error correction coding technology and signal processing technology is an effective solution to improve the reliability of NAND flash memory.Cell-to-Cell interference(CCI)is one of the major noise sources in flash memory channel,which results in the shift of flash memory threshold voltage.CCI is mainly caused by the parasitic coupling-capacitance effect caused by the special physical structure of floating gate transistors.In order to reduce the impact of these channel noises on the storage reliability,the flash memory control chip adopts the low-density parity-check code(LDPC)with strong error correction performance and the soft input soft output decoding algorithm to improve the storage reliability.However,the performance of the decoding algorithm depends directly on the accuracy of the threshold voltage detection methods.In this dissertation,the threshold voltage detection algorithm and the voltage signal post-compensation algorithm are studied deeply by analyzing the channel characteristics of high-density MLC NAND flash memory.The specific research contents are as follows:(1)Analyze the basic principle,construction and decoding algorithm of LDPC code,and achieve the simulation of several soft decision decoding algorithms in additive white Gaussian noise(AWGN)channel,analyze and compare their error correction performance.(2)Deeply study the physical structure and basic operation principle of NAND flash memory,as well as the major noise sources of flash memory channel,i.e.cell-to-cell interference and retention noise interference,verify and analyze the impact of these two kinds of channel noise on the threshold voltage of flash memory.(3)With an aim to overcome the issue of high complexity of existing voltage detection algorithms,this dissertation proposes a non-uniform voltage detection algorithm based on the threshold voltage states by utilizing the channel noise characteristics and threshold voltage distribution characteristics.By optimally setting the sensing reference voltage for each storage state after being disturbed by channel noise,we can improve the utilization of sensing accuracy and greatly increase the accuracy of the read-back voltage.(4)To mitigate the overcompensation phenomenon which is in the existing post-compensation signal processing algorithms,this dissertation proposes a post-compensation algorithm using the posterior information of adjacent cells.By exploiting the characteristics of cell-to-cell interference noise and using the state mean voltage of the successfully decoded flash memory cells,in order to achieve the purpose of accurately calculating the shift of threshold voltage of adjacent cells required by the post-compensation algorithm,so as to mitigate the overcompensation and greatly improve the performance of the post-compensation algorithm. |