DYTEST: a self-learning algorithm using dynamic testability measures to accelerate test generation

Abstract

A self-learning algorithm called DYTEST, which uses a dynamic testability measure (DTM) to accelerate test generation, is presented. DYTEST's DTM guides the choice of a potential main sensitizing path during test generation. A full-logic value label backward implication creates estimates of logic values for the primary inputs. Dependent and k-limited backtracks are used to further reduce the number of backtracks. >