On the adaptation of Viterbi algorithm for diagnosis of multiple bridging faults

This paper proposes a very efficient method to diagnosis multiple bridging faults. This method is based on differential or Delta I/sub DDQ/ probabilistic signatures, as well as on the Viterbi algorithm, mainly used in telecommunications systems for error correction. The proposed method can be seen as a significant improvement over an existing one based on maximum likelihood estimation. The use of the (adapted) Viterbi algorithm allows us to take into account additional information not considered previously. The existing and the proposed method are first described. Then, simulation and experimental results are presented to validate the concept in the context of double faults. Bounds on false diagnosis probability are also provided, estimating the number of test/diagnosis vectors required to reach a given diagnosis reliability for a given number of gates. The bounds allow us to show that this probability exponentially decreases with the number of test vectors and that for, a given value of this probability, the number of vectors required is O(log2(G)), where G is the number of gates.