Variability and reliability aware surrogate model for sensing delay analysis of SRAM sense amplifier

Variability and reliability have become major threats in nano scale era. Both leads to variation in transistor parameters that eventually affect the performance parameters. One of the failure mechanism is Bias Temperature Instability (BTI) which impacts mobility and threshold voltage (V
_th
) of the transistor. This paper investigates the impact of BTI on SRAM sense amplifier at different process corners along with variability effect in 45nm technology. The results depict that, impact of Positive BTI (PBTI) is less than Negative BTI (NBTI) on sensing delay of sense amplifier. Moreover across all process corners sensing delay increases as supply voltage (V
_DD
) decreases and temperature increases. A surrogate model has been developed by using support vector machine (SVM) for variability and reliability analysis of sensing delay. Adaptive learning is used in order to develop the model with less number of samples which attributes to less run time. Evaluation of single sample of sensing delay requires 0.077ms and 0.081ms for read 0 and read 1 respectively. Correlation coefficient has been obtained between HSPICE and SVM in order to validate the model. The values of correlation coefficients are 0.9996 and 0.9997 for read 0 and read 1 respectively.