Multiple Fault Models Using Concurrent Simulation

This paper presents the versatility of our fault simulator in handling different fault models by adding new activity functions such as n-terminal bridge faults. We use our TUFTsim simulator, which is based on concurrent simulation algorithms to efficiently fault simulate large networks, and the Multiple List Traversal mechanism handles the propagation of concurrent elements through the topology. Results on some of the ITC ’99 Benchmarks are shown based on the stuck-at and logical bridge fault models. Overview Concurrent algorithms are the basis of our simulation environment [1]. This method can efficiently handle timing models, asynchronous circuits, and functional models. The goal of concurrent simulation (CS) is to produce simulation results equivalent to simulating each faulty circuit separately. Scenarios of experiments are created for the reference experiment and for each fault. The main idea is to evaluate these scenarios and propagate only those experiments that create state differences from the reference output. In CS, a primary experiment, called the reference experiment, exists during the entire simulation and exhibits the defect free or good behavior of a circuit. The behavior of these experiments can be observed and contrasted, since each experiment propagated 1 This work is supported by NSF MIP 9528194 and NASA NAG8-1311