Skip to Main Content
Methods are presented for designing fault-detection experiments for sequential machines which are realized as parallel connections of simpler component machines. The outputs of these components are assumed to be inaccessible for measurement but it is shown that knowledge of the structure can be utilized to design simpler experiments. The procedure is based upon placing all components but one in a fixed reference state prior to measuring input/output sequences for this one component in order to deduce its state table. This means that the only measurable transitions are those which return the other components to their reference states. Such transitions are made in response to application of restricted input sequences and it is necessary that the state table be deduced from such a set of observations.