Fault simulation and test generation by fault sampling techniques
Sami A. Al-Arian, M.A. Al-Kharji · 2003
Novel techniques and procedures for choosing a sample from the entire fault population are presented. The fault coverage evaluation of the sample and the fault population, given a random test set, are evaluated and proved to be equal, or within a sampling error (+or-e), which represents the difference between the fault coverages of the sample and the total fault population. The sampling procedures presented are applied in such a way that e is small. Moreover, the sample is not purely random but in fact mirrors the total fault distribution. The expected test length and fault coverage evaluation of a given random test set, applied or generated for a given circuit, are calculated. For the example used in this study, exact detection probabilities for every fault in the circuit are calculated.>