Article Details

Title · Authors · Abstract · Keywords · DOI · References

A Reinforcement Learning-based Garbage Collection Algorithm for NAND Flash Memory

RAN Yong, YAN Hua

(College of Electronics and Information Engineering, Sichuan University, Chengdu 610065, China)

Section：Artificial Intelligence · Vol. 2026, Issue 5 · Pages：56-59 · 2096-4706（2026）05-0056-04 · CLC number：TP333；TP301.6 · Document code：A · DOI：10.19850/j.cnki.2096-4706.2026.05.011

Abstract

NAND flash memory requires a garbage collection strategy due to its physical constraints, which significantly degrades system efficiency. Precise hot and cold data separation is critical. In addition, NAND flash memory can only endure a limited number of erase operations. Therefore, improving wear leveling is crucial for extending its lifespan. Existing methods fail to learn data access patterns online and cannot accurately identify hot and cold data under dynamic workloads. To address these issues, a Reinforcement Learning-based Garbage Collection algorithm (RLGC) is proposed. It realizes accurate classification of hot and cold data through online Adaptive Learning and integrating wear leveling mechanism. Experimental results show that compared with the existing best performance of the baseline algorithm, the proposed algorithm reduces the block erase times by 16.4%, the effective page copy times by 50.0%, and improves the wear leveling degree by 33.3%, which significantly improves the garbage collection efficiency and prolongs the lifetime of flash memory.

Keywords

NAND flash memory; garbage collection; Reinforcement Learning; wear leveling

References

  1. [1] ABDULHADI A，CHUNG T S. Crash Recovery Techniques for Flash Storage Devices Leveraging Flash Translation Layer: A Review [J/OL].Electronics，2023，12（6）：1422（2023-03-16）[2025-07-11].https://doi.org/10.3390/electronics12061422.
  2. [2] LEE D，JIN Y，JANG H，et al. CATDOG: Cost-age-time Data Organized Garbage Collection [C]//2022 IEEE International Conference on Networking, Architecture and Storage (NAS).Philadelphia：IEEE，2022：1-6.
  3. [3] LUO L F，LI S C，LV Y N，et al. Performance and Reliability Optimization for High-density Flash-based Hybrid SSDs [J/OL].Journal of Systems Architecture，2023，136：102830[2025-07-11].https://doi.org/10.1016/j.sysarc.2023.102830.
  4. [4] XU G X，LIU Y-B，ZHANG X Q，et al. Garbage Collection Policy to Improve Durability for Flash Memory [J].IEEE Transactions on Consumer Electronics，2012，58（4）：1232-1236.
  5. [5] SADMAN S，ALEKSANDAR M，BISWAJIT R. Flash-DNA: Identifying NAND Flash Memory Origins Using Intrinsic Array Properties [J].IEEE Transactions on Electron Devices，2021，68（8）：3794-3800.
  6. [6]LEE J，KIM J-S. An Empirical Study of Hot/cold Data Separation Policies in Solid State Drives (SSDs) [C]//SYSTOR'13: Proceedings of the 6th International Systems and Storage Conference.Haifa：ACM，2013：1-6.
  7. [7] CHAE S-J，MATIVENGA R，PAIK J-Y，et al. DSFTL: An Efficient FTL for Flash Memory Based Storage Systems [J/OL].Electronics，2020，9（1）：145（2020-01-12）[2025-07-11].https://doi.org/10.3390/electronics9010145.
  8. [8] XU L S Y. Flash Storage Technology Status and Development Research [J/OL].Journal of Physics: Conference Series，2023，2649（1）：012046（2023-06-24）[2025-07-11].https://iopscience.iop.org/article/10.1088/1742-6596/2649/1/
  9. 012046.
  10. [9] WU M，ZWAENEPOEL W. eNVy: A Non-volatile, Main Memory Storage System [J].ACM SIGPLAN Notices，1994，29（11）：86-97.
  11. [10] KAWAGUCHI A，NISHIOKA S，MOTODA H. A Flash Memory Based File System [C]//TCON'95: Proceedings of the USENIX 1995 Technical Conference Proceedings.New Orleans：ACM，1995：155-164.
  12. [11] CHIANG M L，CHANG R C. Cleaning Policies in Mobile Computers Using Flash Memory [J].Journal of Systems and Software，1999，48（3）：213-231.
  13. [12] KWON O，KOH K，LEE J，et al. FeGC: An Efficient Garbage Collection Scheme for Flash Memory Based Storage Systems [J].Journal of Systems and Software，2011，84（9）：1507-1523.
  14. [13] YAN H，YAO Q. An Efficient File-aware Garbage Collection Algorithm for NAND Flash-based Consumer Electronics [J].IEEE Transactions on Consumer Electronics，2015，60（4）：623-627.
  15. [14] YAN H，HUANG Y，ZHOU X Z，et al. An Efficient and Non-time-sensitive File-aware Garbage Collection Algorithm for NAND Flash-based Consumer Electronics [J].IEEE Transactions on Consumer Electronics，2019，65（1）：73-79.
  16. [15] VIDYASAGAR M. A Tutorial Introduction to Reinforcement Learning [J].SICE Journal of Control， Measurement, and System Integration，2023，16（1）：172-191.
  17. [16] HUYNH T N，NGUYEN K D. Reinforcement Learning-driven Small Droplet Detection for Precision Agricultural Spraying [J].Smart Agricultural Technology，2025，12：101450-101450.
  18. [17] DABNEY W，OSTROVSKI G，BARRETO A. Temporally-extended ϵ-greedy exploration [J/OL].arXiv:2006.01782 [cs.LG].（2020-06-02）.https://arxiv.org/abs/2006.01782.

About the authors

Ran Yong (2000-), male, Han ethnicity, from Zunyi, Guizhou, currently pursuing a master's degree in NAND flash memory garbage collection algorithm; Corresponding author: Yan Hua (1971-), male, Han ethnicity, from Qu County, Sichuan Province, Professor, PhD, research direction: Intelligent Information Systems.