Generating High Strength Test Suites for Combinatorial Interaction Testing Using ZDD-Based Graph Algorithms

Abstract

Combinatorial interaction testing is a well practiced method for detecting faults for various computing systems. This method requires that any t-wise parameter interactions must be exercised by at least one test case. The value of t is usually referred to as strength. A large body of research exists about constructing test suites of small or moderate strength, typically t = 2 or 3, but very few techniques are known for constructing those of very high strength. This paper proposes the use of ZDD-based graph algorithms to construct a very high strength combinatorial test suite. Constructing high strength test suites is challenging because of the large number of interactions that must be handled during test suite construction. A ZDD is a data structure that can be used to enumerate a very large number of paths in an undirected graph and to perform operations on large sets of graphs. In our approach test cases and interactions are represented as subgraphs of the same graph. Using a ZDDbased graph library to operate on the graphs, we succeeded in constructing test suites of high strength up to t = 11 for some problem instances.