|
Abstract:
|
Over the past 20 years, factor covering designs, also known as covering arrays or transversal covers, have been increasingly advocated as a mechanism to derive suites of test cases for testing software products. Typically, the software product is viewed as a k parameter function with the input space of each parameter partitioned into distinct classes (i.e., levels). The challenge is to construct a design, with the fewest number of runs (i.e., test cases), so that the combined input space is effectively exercised. If the design has the property that for any t (t< =k) parameters, all level combinations exist, then the design is referred to as a t-covering design. A t-covering design is "optimal" if the number of runs is minimal for t. We describe how "optimal" t-covering designs may be constructed and illustrate the use of such designs for testing software products.
|
