Correlated Rare Failure Analysis via Asymptotic Probability Evaluation

Abstract

In this paper, a novel Asymptotic Probability Estimation (APE) method is proposed to estimate the probability of correlated rare failure events for complex integrated systems containing a large number of replicated cells. The key idea is to approximate the failure rate of the entire system by solving a set of nonlinear equations derived from a general analytical model. An error refinement method based on Look-up Table (LUT) is further developed to improve numerical stability and, hence, reduce estimation error. Our numerical experiments demonstrate that compared to the state-of-the-art method, APE can reduce the estimation error by up to 45× without increasing the computational cost.

Automatic Design of a Broadband Directional Coupler via Bayesian Optimization

“…In [155], [156], the authors relate the overall failure rate of an SRAM circuit (i.e., the target P f, in their context) with the failure rate of each SRAM cell or several cells, referred to as partial failure rate in [156]. APA is proposed in [156] to treat this relation as a linear function, while APE [155] does a further derivation and shows that the relationship is nonlinear. Both APA and APE extend the subset simulation to calculate partial failure rates of different orders.…”

“…In addition to the aforementioned works, a few other notable works also exist in the literature. In [155], [156], the authors relate the overall failure rate of an SRAM circuit (i.e., the target P f, in their context) with the failure rate of each SRAM cell or several cells, referred to as partial failure rate in [156]. APA is proposed in [156] to treat this relation as a linear function, while APE [155] does a further derivation and shows that the relationship is nonlinear.…”