Looking for Functional Fault Equivalence

Recognition of test equivalent faults is usually applied to reduce the number of target faults for test generation and fault simulation. Also fault diagnosis benefits from this knowledge as it allows fast dropping of undistinguishable faults. Equivalent faults are generally identified by mean of a structural analysis of the circuit. Functionally equivalent faults are not considered as their identification is computationally too expensive for real circuits. This paper presents new theorems about functional fault equivalence and dominance. They provide a constructive basis upon which a functional fault collapsing algorithm is built. Application to a set of benchmark circuits establish that identification of functionally equivalent faults is feasible, and that their number may be a not negligible fraction of the faults in a circuit. Results apply both to combinational and synchronous sequential circuits.