Research On Efficient Decode Algorithm For Multilevel Flash Memory Channel Fusing Priori Information Distribution

With the development and application of the Internet of things,cloud computing and big data,the global data volume is exploding at an unprecedented rate.In order to meet the rapid development of the data business,the storage system adopt multi-level(MLC)storage technology to continuously increase the storage capacity.However,with the large increase of flash storage capacity and the further reduction of semiconductor manufacturing process,the storage unit is more complicated with noise interference,which greatly reduces the reliability of flash storage.To provide reliable data storage services,multilevel flash memory improves storage reliability by introducing advanced error control coding techniques.LDPC code,as a kind of error-correcting code with excellent error correction performance,has been gradually applied to multi-level flash storage system.However,there are still some problems in decoding complexity,convergence speed and comprehensive performance.In this paper,based on the research of error control coding technology in multi-level flash channel,this paper mainly studies the channel initial prior information quantization scheme,and combines the serial messaging mechanism to optimize the performance of low complexity decoding algorithm.The detailed work and innovation points in the paper are summarized as follows:i)Firstly,the basic theoretical knowledge of LDPC code,the principle of coding algorithm and its algorithm process are studied in depth,which lays a foundation for the research of error correction coding technology under flash channel.ii)This paper analyzes the structural features,programming and erasure mechanism and noise statistics characteristics of multi-level NAND flash memory,and establishes a multi-level NAND flash channel simulation model.iii)In view of the complexity of error correcting decoding algorithm in flash channel,the serial messaging mechanism is introduced into the RBI-MSD decoding algorithm,and an improved RBI-MSD decoding algorithm is proposed,which can effectively utilize the latest information of the nodes in the process of message updating,so as to greatly accelerate the iterative convergence speed.IV)Based on the prior information distribution characteristics of flash channel,a non-uniform quantization scheme integrating the distribution characteristics of the prior information is proposed,and the non-uniform quantization function is constructed.By combining the improved RBI-MSD algorithm,the simulation analysis is carried out under the multi-level flash channel.The simulation results show that the quantization scheme integrating the distribution characteristics of prior information can effectively improve the detection decoding performance of multi-level flash memory channel compared with the traditional quantitative scheme.