A Degraded ILP Approach for Test Suite Reduction

As the cost of executing and maintaining a large test suite is always expensive, many heuristic techniques have been brought out for test suite reduction in spite of no guarantee of minimum size. The integer linear programming (ILP) approach can generate minimum test suites but it may cost exponential time. This paper proposes a degraded ILP (DILP) approach to bridge the gap between the ILP method and traditional heuristic methods. The DILP can produce a lower bound of minimum test suite and then search a small test suite close to the lower bound. An empirical evaluation of DILP is designed on Boolean specification-based testing. Four typical heuristic reduction strategies: G, GRE, H and GC are compared with DILP empirically. The experimental results show that DILP always outperforms other heuristic reduction strategies and it sometimes can guarantee the minimum size.