Untestable fault identification through enhanced necessary value assignments
Vishnu C. Vimjam, Manan Syal, Michael S. Hsiao · 2005
In this paper, we propose novel low-cost methods that combine static logic implications and binary resolution to significantly increase the number of non-trivial signal relations learned from the circuit. The proposed method first applies resolution techniques to learn new static single-node implications and then uses them to learn powerful multi-node implications. All the newly learned relations help in extracting more necessary assignments for a given fault, potentially increasing the chance for a conflict to occur among the necessary assignments. Experimental results on ISCAS89 and ITC99 benchmarks show that our method can identify significantly more untestable faults compared to existing non branch-and-bound based techniques.