Test pattern generation in presence of unknown values based on restricted symbolic logic

Abstract

Test generation algorithms based on standard n-valued logic algebras are pessimistic in presence of unknown (X) values, overestimate the number of signals with X-values and underestimate fault coverage. Recently, an ATPG algorithm based on quantified Boolean formula (QBF) has been presented, which is accurate in presence of X-values but has limits with respect to runtime, scalability and robustness. In this paper, we consider ATPG based on restricted symbolic logic (RSL) and demonstrate its potential. We introduce a complete RSL ATPG exploiting the full potential of RSL in ATPG. Experimental results demonstrate that RSL ATPG significantly increases fault coverage over classical algorithms and provides results very close to the accurate QBF-based algorithm. An optimized version of RSL ATPG (together with accurate fault simulation) is up to 618x faster than the QBF-based solution, more scalable and more robust.

Accurate QBF-Based Test Pattern Generation in Presence of Unknown Values

“…On average over all circuits and X-values in Table III, the fault coverage would then be less than 0.01% smaller while the runtime decreases by 68.05%. Also, a heuristic approach such as [41] can be used if the runtime is too high for large circuit instances. Figure 10 depicts the range of the additionally achieved fault coverage of the proposed accurate ATPG compared to the commercial tool for the five iterations per circuit and Xratio (cf.…”