ON GENERATING OPTIMAL SIGNAL PROBABILITIES FOR RANDOM TESTS - A GENETIC APPROACH

Genetic Algorithms are robust search and optimization techniques. A Ge netic Algorithm based approach for determining the optimal input distr ibutions for generating random test vectors is proposed in the paper. A cost function based on the COP testability measure for determining t he efficacy of the input distributions is discussed, A brief overview of Genetic Algorithms (GAs) and the specific details of our implementa tion are described. Experimental results based on ISCAS-85 benchmark c ircuits are presented. The performance pf our GA-based approach is com pared with previous results. While the GA generates more efficient inp ut distributions than the previous methods which are based on gradient descent search, the overheads of the GA in computing the input distri butions are larger. To account for the relatively quick convergence of the gradient descent methods, we analyze the landscape of the COP-bas ed cost function. We prove that the cost function is unimodal in the s earch space. This feature makes the cost function amenable to optimiza tion by gradient-descent techniques as compared to random search metho ds such as Genetic Algorithms.