SRAM Memory Testing Methods and Analysis

Abstract

In the scenario of growing technologies towards single digit nanometer range, the existing algorithmic contemporary test methods have become inadequate in detecting all the faults within the static random access memory. To address the issues related to contemporary test methods, machine learning-based test analysis is proposed, which elevates the method of dataset preparation using various process parameters that are drawn from functional fault models (FFMs). The outcome of this proposed work is modeling of FFMs using ML regression, classification, and further prediction with accuracy analysis. The experiments resulted that logistic regression is best suited model that resulting with high accuracy in the range of 95% to 97%, compared to the linear regression model that results in accuracy levels in the range of 26.58% to 63%.