Guiding random test generation with program analysis

L. Ma, C. Artho, C. Zhang, H. Sato, J. Gmeiner, R. Ramler. Guiding random test generation with program analysis. volume 252, pages 15-16, 2, 2016.

  • Lei Ma
  • Cyrille Artho
  • Cheng Zhang
  • Hiroyuki Sato
  • Johannes Gmeiner
  • Rudolf Ramler
BuchSoftware Engineering 2016 Fachtagung des GI-Fachbereichs Softwaretechnik - Proc. SE 2016
TypIn Konferenzband
VerlagGesellschaft für Informatik
SerieLecture Notes in Informatics

Random test generation is effective in creating method sequences for exercising the software under test. However, black-box approaches for random testing are known to suffer from low code coverage and limited defect detection ability. Analyzing the software under test and using the extracted knowledge to guide test generation can help to overcome these limitations. We developed a random test case generator augmented by a combination of six static and dynamic program analysis techniques. Our tool GRT (Guided Random Testing) has been evaluated on real-world software systems as well as Defects4J benchmarks. It outperformed related approaches in terms of code coverage, mutation score and detected faults. The results show a considerable im-provement potential of random test generation when combined with advanced analysis techniques.