Optimal Scheduling for Combinatorial Software Testing and Design of Experiments
Robert V. Binder · 2018
Test case ordering can have significant effects on the cost, duration, or safety of a test suite. As the total number of possible orderings isn!forntest cases, finding a cost-optimal ordering can be a non-trivial problem. Combinatorial algorithms that generatet-wise test suites either explicitly randomize sequence or order them as a side effect of the algorithm. Design of experiments uses similar strategies to select test configurations and requires sequence randomization for statistical validity. Both approaches produce test sequences that are very likely sub-optimal with respect to cost. This paper presents an integer programming model that minimizes the total cost of a test sequence and notes, for experimental design, how statistical validity may be preserved for a non-random order.