Abstract
The demand for fault diagnosis has increased with the increasing complexity of VLSI devices. Defects that result from process variations may cluster in certain areas. When a large number of defects cluster in an area, diagnosing these defects is a challenging problem because defects frequently exist that are partially or completely dominated by other adjacent defects. The most common approach for modeling delay defects is the transition fault model. We propose a diagnostic method that can handle clusters of transition faults. The experimental results for the full-scan version of the ISCAS’89 benchmark circuits demonstrate the accuracy of the proposed method.
