作者: Michael S. Hsiao , Sarmad Tanwir
关键词: Algorithm 、 Fault grading 、 Sequential logic 、 Linear regression 、 Logic simulation 、 Mean squared error 、 Speedup 、 Entropy (information theory) 、 Fault coverage 、 Computer science
摘要: In this paper, we explore the relationship between fault coverage for sequential circuits and observed entropy during test application. We find that measured at a subset of primary pseudo-primary outputs is highly significantly correlated to regions design are exercised by vectors. Consequently, they can be used estimating coverages newly generated sets. As computing requires only logic simulation, it greatly speed up grading process. our experiments on ISCAS'89 benchmarks, able achieve an average root mean squared error 2.1% in estimation sets varying lengths, together with order magnitude speedup over simulation.