Using dependency analysis to predict fault dictionary effectiveness

A case study was performed to determine how fault isolation ambiguity groups for electronic systems can be predicted. The baseline method for fault isolation was the fault dictionary based on fault simulation. Several predictors were investigated: a simplified fault dictionary with a detect limit of only 1, dependency analysis, and intersections of cones of logic that feed erroneous outputs. Dependency analysis (based on a hierarchical logic model) and the simplified fault dictionary both produced fault isolation statistics that closely approximated those of the full fault dictionary. Cones of logic did not generate useful fault isolation results.