224_pdfsam_VLSI TEST PRINCIPLES & ARCHITECTURES

224_pdfsam_VLSI TEST PRINCIPLES & ARCHITECTURES -...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
Test Generation 193 In other words, g i D - frontier A i is the set of necessary assignments for detecting the target fault. Finally, another form of dynamic learning consists of finding a partial circuit decomposition in the form of a frontier called the evaluation frontier (or E-frontier for short) [Giraldi 1990]. The idea behind this is that at any point in the decision process there exists a frontier of evaluated gates, and that the same frontier may be achieved by a different set of decision variables. For instance, three value assign- ments are possible to achieve the output of an AND gate set to logic 0. Each frontier can be associated with an edge in the decision tree. Suppose a set of E-frontiers has been learned for fault f i and the corresponding decision tree for f i is available. Now, for a different fault f j , if a similar E-frontier is obtained, where the E-frontier has at least one fault effect as illustrated in Figure 4.28, the subtree for f j ’s decision
Background image of page 1
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: tree could be directly copied from the subtree in f i ’s decision tree, to which the E-frontier was mapped. Note that the set of current primary input assignments is sufficient to justify the E-frontier, and all nodes to the right of the E-frontier are all “don’t cares.” In this figure, the only primary inputs that could have been used to propagate the fault effect are a , b , and m . If there was an assignment on these three primary inputs that was able to propagate the D for fault f i to a primary output, then the same assignment would be able to propagate the D for f j as it had the same E-frontier. In other words, the decision variables in the subtree corresponding to this point in the decision process consisted of only these three variables outside the f i a = X b = X m = X 1 D 1 E-frontier a = X b = X m = X f j 1 D 1 E-frontier ± FIGURE 4.28 Example of evaluation frontier....
View Full Document

Ask a homework question - tutors are online