On the Sensitization Probability of a Critical Path Considering Process Variations and Path Correlations

In this paper, the problem of determining the sensitization probability of a path by a test vector is investigated which is important when testing for delay defects due to process variations. An algorithm and its extension are presented with a tradeoff in accuracy and execution time. Gate delays are modeled as probability mass functions, and novel operations are introduced that take into consideration circuit reconvergences and path correlations. The proposed approach estimates accurately the sensitization probability of a path for a test vector. It relies on novel operations at each gate on the target path. Experimental results and comparisons with Monte Carlo are presented on the ISCAS'85, ISCAS'89, and ITC'99 benchmarks.