Font Size: a A A

Research On Detection And Decoding Technology Based On Error Characteristics Of 3D NAND Flash Memory

Posted on:2022-12-16Degree:MasterType:Thesis
Country:ChinaCandidate:X Y ZhangFull Text:PDF
GTID:2518306779994719Subject:Computer Hardware Technology
Abstract/Summary:PDF Full Text Request
NAND flash memory has become the mainstream storage medium for data storage due to its advantages of fast read and write speed,small size and large capacity.With the development of process technology and multi-bit technology,the physical size of planar NAND flash memory is close to the physical limit,and it is difficult to further increase the storage density.To further increase storage capacity,3D NAND flash memory continues to increase capacity while maintaining the high performance of storage devices by being vertically stacked.The structural innovation of 3D NAND flash memory brings new noise to the NAND flash memory channel,and the reliability of data storage faces new threats.With the widespread popularity of 3D NAND flash memory,how to ensure the data storage reliability of 3D NAND is one of the current research hotspots in the storage field.Although the previous threshold voltage detection technology and soft-decision-based Low-Density Parity-Check(LDPC)codes can effectively improve the reliability of planar NAND flash memory,the structure of 3D NAND flash memory has led to its parallel with the planar NAND flash memory.The error characteristics of flash memory are different,and directly adopting the detection and decoding scheme in planar NAND flash memory will result in low reliability and low performance.Therefore,it is very important to design a new detection and decoding scheme based on the error characteristics of 3D NAND flash memory.In view of this,this thesis studies an efficient threshold voltage detection scheme and a neural network-assisted soft-decision decoding scheme based on the error characteristics of3 D NAND flash memory,and analyzes the performance of the scheme.The specific research contents of this thesis are summarized as follows:(1)The basic mechanism,structural characteristics,etc.Through the actual test of 3D NAND flash memory chip based on the actual platform of FPGA,the characteristics of threshold voltage shift under different programming erasure noise and data retention time persistence noise are analyzed in detail,and the error characteristics in 3D NAND flash memory are analyzed in detail.Differences were analyzed.(2)An efficient threshold voltage detection scheme is proposed based on the threshold voltage distribution shift rule,error characteristics and data writing characteristics of 3D NAND flash memory.As the data storage time increases,the effect of reducing the original code rate is more obvious,which is about 20% lower on average than the default reading reference voltage method.Compared with the re-read mechanism in the flash controller to achieve the optimal read reference voltage,the adjustment times of the reference voltage at the same time is reduced by about 30% on average.The simulation results show that the scheme can effectively reduce the original bit error rate of data without losing too much performance.(3)According to the error characteristics in 3D NAND flash memory and the structural similarity of 3D NAND flash memory,an inter-layer similarity-assisted soft-decision decoding scheme is proposed.Simulation results show that this scheme can improve reliability and read performance to a certain extent.
Keywords/Search Tags:3D, NAND flash, Reliability, Threshold voltage detection
PDF Full Text Request
Related items