Parameterized Random Testing

Abstract

Random testing uses random inputs to test digital circuits. A major problem in random testing is the cost to compute the test length which is required for achieving an acceptable fault coverage. Different input distributions on the random inputs produce different fault detection probabilities. Therefore parameterized input distributions are analyzed and analytical methods are given for computing the fault coverage as a function of the parameters. The parameters are chosen so that the fault detection probability is maximized and the test pattern length is minimized. This analytical method of analyzing random test patterns tends to be faster than fault simulation.