Construction of Mixed Covering Arrays of Variable Strength Using a Tabu Search Approach

The development of a new system involves extensive tests on the software functionality in order to identify possible failures. Also, a system already built requires a fine tuning of its configurable options to give the best performance in the environment it is going to work. Both cases require a finite set of tests that avoids testing all the possible combinations (which is time consuming); to this situation Mixed Covering Arrays (MCAs) are a feasible alternative. MCAs are combinatorial structures represented as matrices having a test case per row. MCAs are small, in comparison with brute force, and guarantees a level of interaction among the parameters involved (a difference with random testing). We present a Tabu Search (TS) algorithm to construct MCAs; the novelty in the algorithm is a mixture of three neighborhood functions. We also present a new benchmark for the MCAs problem. The experimental evidence showed that the TS algorithm improves the results obtained by other approaches reported in the literature, finding the optimal solution in some the solved cases.