Multiple controlled antirandom testing (MCAT) for high fault coverage in a black box environment

Among the black-box approaches to digital circuit testing, Random testing is popular due to its simplicity and cost effectiveness. Unfortunately, available evidences suggest that Random testing is equipped with a number of redundant patterns that increase test length without signi cantly raising the fault coverage. An extension to Random testing is Antirandom that removes redundancy by introducing a divergent pattern with every subsequent test pattern selection. A divergent pattern is induced by maximizing the Hamming distance and Cartesian distance of every subsequent test pattern from the set of previously applied test patterns. However, an enumeration of input combinations is required for the selection of a divergent pattern. Therefore, selection of a divergent pattern from all input combinations restricts the scalability of an Antirandom test pattern generation. One of the recently considered approaches is the stacking of locally optimized short sequences to generate a complete test sequence. Locally optimized short sequences originate from randomly chosen patterns instead of divergent patterns to avoid enumeration of input space. Seeding of random patterns for short sequences affects global diversity of the generated test sequence and hence, fault coverage is compromised. Therefore, this paper rstly proposes a tree traversal search based selection of divergent patterns that eliminates the search space. Ease in divergent pattern selection is used to generate optimal short sequences for divergent patterns instead of random patterns. Consequently, Multiple Controlled Antirandom Tests (MCATs) are generated that maximize distance between locally optimal short sequences to elevate the fault coverage. Fault simulation results on both ISCAS'85 and ISCAS'89 benchmark circuits prove the scalability and effectiveness of the proposed approach. Moreover, the comparison shows that up to 12% of fault coverage is improved as a result of proposed MCAT test pattern generation.