On the distribution of fault coverage and test length in random testing of combinational circuits

Abstract

Models for the process of testing faults in combinational circuits under both random and pseudorandom tests are proposed. Based on these models, the probability distribution function (PDF) of fault coverage is derived. By using an important relationship between the PDF of fault coverage and that of test length, an expression is derived for the PDF of test length. These analytical results allow expressions to be obtained for expected values of fault coverage and test length, and other important statistics not obtained by existing techniques. Empirical results validating these models for fault coverage analysis are presented from experiments with several circuits. Techniques for estimation of necessary parameters, based on both statistical and probabilistic models, are proposed. These techniques, combined with the theoretical results, define a comprehensive methodology for random testability prediction of combinational circuits.<>