Structural heuristics for SAT-based ATPG

Due to ever increasing design sizes more efficient tools for Automatics Test Pattern Generation (ATPG) are needed. The application of the Boolean satisfiability problem (SAT) to ATPG has been shown to be a robust alternative to traditional ATPG techniques. A major challenge of research in the field of SAT-based ATPG is to obtain a robust algorithm which can solve hard SAT instances reliably without slowing down easy-to-solve SAT instances. This is particular important, since easy-to-solve SAT instances form the majority of an ATPG run. This paper proposes two structural heuristics. The first one uses testability measurements to obtain an improved initial variable order while the second heuristic prunes many easy-to-test faults by finding easy-to-control paths. Experimental results on large industrial designs confirm that the proposed methodologies result in a significant overall speed-up.