Test vector chains for increasing the fault coverage and numbers of detections

The authors introduce the concept of test vector chains which allows one to obtain new test vectors from existing ones through single-bit changes. A test vector chain is defined based on a pair of test vectors t ₁ and t ₂. It consists of a sequence of single-bit changes, which gradually modifies t ₁ into t ₂. The authors demonstrate that, based on a test set T that does not detect all the detectable target faults, it is possible to define a significant number of test vector chains, which are effective in detecting yet-undetected target faults. It is also possible to find test vector chains that are effective in increasing the numbers of detections of target faults that are detected by T. Increasing the number of detections increases the coverage of untargeted faults, that is, faults that were not targeted during the generation of T. The authors study criteria for identifying the most effective test vector chains of a test set in order to avoid considering m(m-1) test vector chains for a test set of size m, and describe test generation procedures based on test vector chains.