Error Characterization and Correction in NAND Flash Memory Systems

Error Characterization and Correction in NAND Flash Memory Systems encompasses the study of error mechanisms arising from the physical and electrical characteristics of memory cells, and the development of strategies to detect, model and rectify these errors to ensure data integrity. As cell geometries shrink and multi-level storage densities increase, flash memory endures disturbances from program–erase cycling, charge leakage, cell-to-cell interference and temperature fluctuations. These lead to shifts in threshold-voltage distributions, elevating raw bit error rates and challenging the efficacy of error correcting codes. Research combines statistical modelling, machine-learning and advanced coding techniques to characterise error distributions under varying conditions, predict device endurance and adapt read reference voltages in real time. Low-density parity-check and Bose–Chaudhuri–Hocquenghem codes are deployed alongside soft-decision decoding and machine-learning-assisted parameter estimation to enhance correction capability without unduly increasing latency or energy consumption. Integrated modelling frameworks enable designers to balance reliability, performance and energy efficiency, facilitating the deployment of high-density solid-state drives in consumer, enterprise and cloud environments around the globe.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Error Characterization and Correction in NAND Flash Memory Systems publication trend

The graph below shows the total number of articles in error characterization and correction in nand flash memory systems across all publications each year (not limited to Nature Index journals).

Technical terms

NAND flash memory: A non-volatile storage technology that organises floating-gate transistors into arrays of cells capable of retaining charge to represent data even when power is removed.

Program–erase cycle: A sequence in which a memory cell is programmed with charge and then erased, with repetition leading to wear-induced degradation of cell characteristics.

Raw bit error rate (RBER): The proportion of bits incorrectly read from memory before any error correction is applied, typically increasing with cell wear and environmental stress.

Error correcting code (ECC): An algorithmic scheme that adds redundancy to stored data, enabling detection and correction of bit errors during read operations.

Read reference voltage: A threshold voltage used by the memory controller during read operations to distinguish between different cell states in multi-level cells.

Graph neural network (GNN): A machine-learning architecture that operates on graph-structured data, capturing relationships between interconnected entities such as adjacent flash memory cells.

References

1. Modeling of 3D NAND Characteristics for Cross‐Temperature by Using Graph Neural Network and Its Application. Advanced Intelligent Systems (2023).
2. Modeling 3D NAND Flash with Nonparametric Inference on Regression Coefficients for Reliable Solid-State Storage. Future Internet (2023).
3. Machine-Learning-Based Read Reference Voltage Estimation for NAND Flash Memory Systems Without Knowledge of Retention Time. IEEE Access (2020).
4. Low-energy error correction of NAND Flash memory through soft-decision decoding. EURASIP Journal on Advances in Signal Processing (2012).
5. Reducing latency overhead caused by using LDPC codes in NAND flash memory. EURASIP Journal on Advances in Signal Processing (2012).
6. An Early-Life NAND Flash Endurance Prediction System. IEEE Access (2021).