Programmable cellular automata based testbed for fault diagnosis in VLSI circuits

This paper analyses the fault diagnosis capability of Programmable CA based signature analyzer. Polynomial algebraic tools have been developed for selection of a group CA of appropriate size for generation of faulty signatures due to faults in different partitions of a circuit. Further, a special class of CA termed as multiple attractor CA has been utilized for classification of faulty signatures that leads to unambiguous diagnosis of faulty partition(s). This classification has resulted in phenomenal reduction of fault dictionary size to the order of number of partitions in the circuit. Experiments conducted on some real-life and benchmark circuits establishes the simple, modular, and cascadable structure of CA as an efficient tool for fault diagnosis in VLSI circuits