A Simulated Annealing with Variable Neighborhood Search Approach to Construct Mixed Covering Arrays.

Nowadays, the software use has been extended along almost all human activity areas and along with it, the reliability of software has become a need. Software testing consumes a significant portion of the cost of development of a new piece of software, and it is a stage that it is not feasible to ignore. A basic way to test a software is by doing an exhaustive testing process, but sometimes in the real world it is not feasible due to time and resource constraints. An alternative approach is called software interaction testing which is based on combinatorial structures called Covering Arrays and Mixed Covering Arrays (MCA) that are used to represent test-suites. This paper focuses on constructing MCAs by using a hybrid approach based on Simulated Annealing and a Variable Neighborhood Search function (SA-VNS). The quality of the solutions found was measured by solving a benchmark reported in the literature, and a comparison against other approaches shows that our proposed approach is an attractive and competitive option to construct MCAs.