Research On Decoding Optimization Of Polar Codes For NAND Flash Systems

In the information transmission process,the performance of the error correction coding code affects the information storage reliabilily.In 2009,Turkish scholar Arikan first proposed polar codes.As a linear block code,it proves to be an error correction code that can be infinitely close to the Shannon limit,and has a lower coding complexity.With the advent of the digital age,NAND Flash error correction codes originally used for high storage capacity will gradually fail to meet data storage requirements,and the emergence of polar codes has opened up new directions for new error correction research for NAND Flash.This paper focuses on the reliability of polar decoding combined with the structural characteristics of NAND Flash memory cells.The research content and innovations are as follows:(1)Belief propagation(BP)decoding is a parallel decoding algorithm with high parallel processing information advantages.The BP decoding information transmission adopts an iterative method.When the number of iterations is large,the decoding complexity is high.Therefore,reducing the number of iterations as much as possible is beneficial to reduce decoding complexity.Based on BP decoding,a BP decoding algorithm based on segmented cyclic redundancy check(CRC)assisted early termination iteration is proposed.It replaces a long check code inserted at the end of the information sequence by inserting a plurality of short check codes in the information sequence by segmentation CRC auxiliary decoding.Segmented CRC-assisted decoding reduces the time required for CRC check,and can terminate iterations early,reducing decodingcomplexity and delay,and obtaining more performance gains at high SNR;(2)The self-characteristics of the polar codes make it possible to complete a codeword structure with a power length of 2,and it is impossible to obtain an arbitrary length codeword.In order to construct a length-compatible,code rate adaptive polar codes,a reliability puncturing decoding scheme based on segmented CRC is proposed.Using the puncturing polar codes and channel polarization property,the less reliable channel is used as the punctured bit,which improves the reliable channel utilization and obtains better decoding performance;(3)The channel polarization of the polar codes makes the channel's error probability have a large difference.After the channel is polarized,the channel is divided into a channel with a higher error probability and a channel with a lower error probability.Moreover,when the error probability is lower,the channel reliability is higher,and the protection of information is stronger.The Multi-Level Cell(MLC)NAND Flash can obtain a threshold voltage model according to the threshold voltage distribution characteristics.Combined with the unequal error protection characteristics caused by polar codes channel polarization,the segmented CRC-assisted puncturing decoding can be applied to MLC NAND Flash to obtain better decoding performance,which has certain research value.In this paper,the optimization and improvement of the polar codes decoding algorithm effectively reduces the number of decoding iterations and improves the decoding performance.Applying it to MLC NAND Flash has also achieved good results and has certain practical value.