A Hybrid Scheme of Reducing Read Latency in NAND Flash by Integrating Duplications and Soft Decoding
Yaofang Zhang, Peixuan Li, Ping Xie · 2022
The continuous development of the technology scale has made the storage density of NAND flash memory larger and larger, and its reliability has become lower and lower. Therefore, reliability has become an issue of concern to everyone. In order to alleviate reliability problems caused by technological expansion, LDPC codes are widely used in flash memory to provide high error correction capabilities. However, the decoding methods of LDPC codes are divided into hard decoding and soft decoding. The LDPC code is directly used and as the quantization level of soft decoding increases, the delay caused to the Solid-State Drive is also increasing. Taking into account the characteristics of the quantization level of LDPC soft decoding, this paper proposes a new method IDSD, which combines duplications and LDPC soft decoding. This scheme is mainly for hot data. While reading the data, it needs to be based on the quantization level of the record. Use the duplications or soft decoding method to correct the erroneous data. When the quantization level of LDPC soft decoding reaches one, it needs to read the data in the duplications at this time. This method reduces the reading of the Solid-State Drive. Take the delay and improve the average response time. Experimental results show that compared with the original scheme, the average response delay time of the IDSD scheme is increased by 5%-30%, and compared with the traditional refresh scheme, it is increased by 1%-30%.