Improving accuracy in path delay fault coverage estimation

Abstract

A recently published method computes path delay fault coverage from the count of the number of path faults newly sensitized by a simulated vector pair. Such an estimate is pessimistic since several paths may share a set of lines. In this paper, we present a continuum of approximate methods, approaching exact fault simulation, for a tradeoff between accuracy and complexity. Higher accuracy is obtained at the expense of CPU time. We propose the use of flags corresponding to fixed-length path-segments. A flag indicates whether or not the segment has been included in a previously detected path fault. A path fault detected by a pair of vectors is counted as a new detection only if it includes at least one segment not included in any previously covered path. This gives us a pessimistic estimate of the number of newly detected faults by a simulated vector pair. When the numbers of fan-in and fan-out branches per gate are small, the method adds a modest overhead to good machine simulation provided the flagged path-segments are short. As the length of segments is increased, the accuracy approaches that of exact fault simulation. Results show that the estimates with small segment lengths are very close to actual fault coverages.