Endurance Prediction Based on Hidden Markov Model and Programming Optimization for 28nm 1Mbit Resistive Random Access Memory Chip

Abstract

We proposed a state transition probability model based on Hidden Markov Model (HMM), which can predict the lifetime for different endurance failure modes. The prediction span of this model is 500 cycles, and the accuracy rate can reach 83.6 %. Based on the prediction results, we proposed an optimized programming algorithm to rescue the failing cells during endurance. The failure rate of memory chip is reduced by 38 % after 5000 cycles. The forecasting model is effective and of practical value since it is based on the data from 28nm 1Mbit Resistive Random Access Memory (RRAM) chips, including the peripheral circuit noise, machine noise and device randomness. This prediction model is significant for promoting RRAM application and improving memory chip utilization.