Machine-Learning-Based Read Reference Voltage Estimation for NAND Flash Memory Systems Without Knowledge of Retention Time

Abstract

To achieve a low error rate of NAND flash memory, reliable reference voltages should be updated based on the accurate knowledge of program/erase (P/E) cycles and retention time, because those severely distort the threshold voltage distribution of memory cell. Due to the sensitivity to the temperature, however, a flash memory controller is unable to acquire the exact knowledge of retention time, meaning that it is challenging to estimate accurate read reference voltages in practice. In this paper, we propose a novel machine-learning-based read reference voltage estimation framework for the NAND flash memory systems without the knowledge of retention time. To establish an unknown input-output relation of the estimation model, we derive input features by sensing and decoding memory cells in the minimum read unit. In order to define the relation between unlabeled input features and a pre-assigned class label, namely label read reference voltages, we propose three mapping functions: 1) k-nearest neighborsbased, 2) nearest-centroid-based, and 3) polynomial regression-based read reference voltage estimators. For the proposed estimation schemes, we analyze that the storage overhead and computational complexity are increasing function of the exploited feature dimension. Accordingly, we propose a feature selection (or dimension reduction) algorithm to select the minimum dimension and corresponding features to reduce the overhead and complexity while maintaining high estimation accuracy. Based on extensive numerical analysis, we validate that the derived features successfully replace unknown knowledge of retention time, and the proposed feature selection algorithm precisely adjusts the trade-off between overhead/complexity and estimation accuracy. Furthermore, the simulation and analysis results show that the proposed framework not only outperforms the conventional estimation schemes but also achieves the near-optimal frame error rate while sustaining low latency performance. INDEX TERMS NAND flash memory, read reference voltage estimation, machine learning, dimension reduction.

Modeling of 3D NAND Characteristics for Cross‐Temperature by Using Graph Neural Network and Its Application

“…In addition, some studies introduced methods to enhance detection during the read process. [ 19,20 ] They used neural networks to distinguish the distributions more clearly and figure out the characteristics without specific knowledge.…”

Dynamic Error Recovery Flow Prediction Based on Reusable Machine Learning for Low Latency NAND Flash Memory Under Process Variation

“…The read reference voltage is ideally set as the intersections between two adjacent symbol states' threshold voltage distributions. Since it is hard to obtain information on the intersection point, read reference voltages are estimated utilizing the methods introduced in [15]- [18]. In this paper, we assume that the optimal read reference voltages are obtained using one of the pre-mentioned methods.…”

“…However, it is impossible to find such intersections since the knowledge of the threshold voltage distribution affected by process variations and additive noise cannot be perfectly obtained. Instead, the optimal read reference voltages are estimated by updating from default value using sentinel-cells-enabled method [15], read-retry method [16], valley tracking method [17], or machine learning (ML)-based method [18]. Although the read reference voltage estimation can reduce raw bit error rate (RBER), the inevitable bit error may occur.…”

Read Reference Voltage Adaptation for NAND Flash Memories With Neural Networks Based on Sparse Histograms

“…Methods to estimate the flash channel using neural network (NN) were proposed in [26] and [27]. In [26], a NN approach is proposed based on meta-information and ECC decoding, i.e., the knowledge of the number of P/E cycles, the frame error rate (FER), the average number of iterations of the LDPC decoder, and the bit flip ratio of the decoder are used as features. This method shows good results, but requires metadata and therefore needs additional readouts.…”