9.1x Error acceptable adaptive artificial neural network coupled LDPC ECC for charge-trap and floating-gate 3D-NAND flash memories

Toshiki Nakamura, Yoshiaki Deguchi, Ken Takeuchi

Abstract: Adaptive Artificial Neural Network coupled (ANN) LDPC ECC (ANN-LDPC ECC) is proposed to increase acceptable errors by 9.1-times and to extend the data-retention lifetime by 76-times for charge-trap and floating-gate 3D-NAND flash memories. Adaptive ANN automatically compensates for complex memory cell errors such as lateral charge migration, vertical charge de-trap, inter floating-gate capacitive coupling noise and inter word-line variations. In addition, proposed ANN-LDPC can reproduce the dynamic endurance and data-retention time dependence of errors. Proposed ANN-LDPC is implemented in the storage controller and can precisely and adaptively estimate BER. As a result, memory cell errors are corrected without read time penalty or storage controller size increase.