Combinational test generation using satisfiability

We present a robust and efficient algorithm for combinational test generation using a reduction to satisfiability (SAT). The algorithm, TEGUS, has the following features. We choose a form for the test set characteristic equation which minimizes its size. The simplified equation is solved by an algorithm for SAT using simple, but powerful, greedy heuristics, ordering the variables using depth-first search and selecting a variable from the next unsatisfied clause at eachbranching point. For difficult faults the computation of global implications is iterated, which finds more implications than previous approaches and subsumes structural heuristics such as unique sensitization. Without random tests or fault simulation, TEGUS completes on every fault in the ISCAS networks, demonstrating its robustness, and is 11 times faster for those networks which have been completed by previous algorithms. Our publicly available implementation of TEGUS can be used as a base line for comparing test generation algorithms; we present comparisons with 45 recently published algorithms. TEGUS combines the advantages of the elegant organization of SAT-based algorithms, such as that by Svanæs, with the efficiency of structural algorithms, such