Spatial fault simulation and the saturation effect

Abstract

Stochastic fault simulation processes have been used successfully to generate fault distributions for evaluating fault tolerant VLSI designs. In one of these processes, faults in subareas of integrated circuits are simulated as a function of time. This leads to an exponential increase of the average number of faults in the area segments of the integrated circuits. It was discovered analytically that by forcing a correlation between the number of faults in adjacent area segments, the increase in the number of faults with time exceeds exponential growth and exhibits a singularity. At the singularity point the fault population becomes infinite. The time associated with this singularity has been denoted as 'saturation time'.<>