Publication

Abstract : Fault diagnosis is the process of identifying a potential faulty site. It is crucial to consider bridging faults and Byzantine faults during diagnosis, as their fault effect manifests at a completely different location than the actual physical faulty site. Ignoring them may misguide the diagnosis analysis tool, thereby leading to a failed diagnosis. This paper aims at improving the precision with which a fault is identified by removing any ambiguity between a bridging fault and a stuck-at fault. It also aims at identifying Byzantine faults with improved accuracy. Experimental results confirm 100% accuracy in diagnosis for ISCAS′85, ISCAS′89, ITC′99 and IWLS′05 benchmark circuits. There is more than 10% reduction in the number of patterns needed for diagnosis as compared to previous works. It was also observed that the average value of index of the actual injected faults in the list of defect candidates is 1.6, which is 46% lesser than related previous work. For all the circuits considered in the proposed work, the defect candidate list contains only one type of fault. This indicates that the proposed method is efficient in distinguishing bridging faults from stuck-at faults even in the presence of Byzantine faults.