Various experimental designs for estimating the number of faults in a syste
m are studied including:
i. removal of each fault as it is detected,
ii. marking of each fault as it is detected,
iii, introduction of a known number of faults into the system followed by #
i or #ii.
A unified framework is developed for comparing these designs; it also produ
ces simplified estimators having high efficiency relative to maximum likeli
hood estimators. The designs are compared in terms of
statistical accuracy,
the number of failures that need to occur to achieve a given accuracy.
On the basis of these comparisons, some general recommendations are made on
the level of seeding as well as the choice of removal or recapture designs
. When the testing effort is sufficient so that roughly two thirds of the f
aults are detected, the removal-design is preferred over the recapture-desi
gn, and there are no gains from seeding. However, this conclusion depends o
n assigning unit cost to all fault detections, which might not always be re
asonable.