Efficient Random Testing with Global Weights

This paper describes a new and highly efficient approach for weighted random pattern generation. In contrast to the state-of-the-art approaches, where input specific weights are computed, the proposed method is based on the computation of global weights. This set of a very few weights (e.g., 4 or 8) is pattern oriented and therefore, with each weight the generation of the related random patterns is uniquely specified. Starting with a deterministic test pattern set and the inherent pattern specific weights, columns or rows can be inverted such that the initial weights are maximized in order to minimize the number of random patterns. Our experiments with the prototype system POWER-TEST (Pattern Oriented WEighted Random TESTing) show that very high fault coverage can be achieved with low computation and implementation effort at low self-test hardware costs.