Stability of Product Sampling under Product-Line Evolution: Master's Thesis

Abstract

Product-line sampling is a common method to cope with the exponential growth of products in product-line testing. Over the years, different sampling algorithms have been developed and validated against each other. Researchers strive to create efficient sampling algorithms to cope with large product lines. Typical criteria to evaluate sampling algorithms are the computation needed to calculate a sample, and the number of configurations a generated sample contains. Until now, no evaluation criteria considers the product-line evolution, as a factor for evaluating sampling algorithms. With this master thesis we present the stability of samples under product-line evolution as new evaluation criteria for sampling algorithms. Therefore, we define the meaning of stability in context of product-line evolution. Furthermore, we develop and implement metrics to measure the stability of sampling algorithms. Moreover, we classify whether established sampling algorithms produce stable samples or not, based on the results of our metrics.