Industrial Impact

Faster fault finding at Google using optimisation

Testing at Google poses many challenges: release cycle is extremely short and there exists such a heavy layer of dependency that you may break someone else's product without knowing. Dr. Shin Yoo has been collaborating with Google to release a Google-internal tool, based on multi-objective test suite optimisation, that enables the developers to get earlier fault feedback before submitting their code changes. The project was funded by Google Research Award and the result has been invited to be presented at Google Test Automation Conference (GTAC) 2010 in India as well as in ESEC/FSE 2011.

- Faster Fault Finding at Google Using Multi Objective Regression Test Optimisation, Shin Yoo, Robert Nilsson and Mark Harman, 8th European Software Engineering Conference and the ACM SIGSOFT Symposium on the Foundations of Software Engineering (ESEC/FSE '11) [pdf] [bibtex]

SBST used in Microsoft Pex Automated White-Box Testing Tool

Microsoft Research developed an automated white-box test data generation tool called Pex, which is now part of MS Visual Studio 2010 Power Tools. Pex is based on Dynamic Symbolic Execution (DSE) of source code. While this approach is very powerful, there are certain limitations that prevent Pex to achieve higher code coverage. Dr. Kiran Lakhotia collaborated with Microsoft Research and developed an extension for Pex that uses SBST to overcome those limitations.

- FloPSy - Search-Based Floating Point Constraint Solving for Symbolic Execution, Kiran Lakhotia, Nikolai Tillmann, Mark Harman and Jonathan de Halleux, 22nd IFIP International Conference on Testing Software and Systems, pages 142-157 [pdf] [bibtex]

Speeding up IBM's middleware testing

Smoke-testing IBM's middleware products require the tester to come up with different sets of test that can be executed within a very short time (typically 1 hour). Dr. Shin Yoo's collaboration with IBM Haifa led to the use of graphics card to parallelise multi-objective test suite optimisation algorithm. Sequential algorithm took more than 1 hour to optimise IBM's test data, which defies the aim of smoke testing. The parallelised algorithm using General Purpose computation on Graphics Processing Unit (GPGPU) brought in 25x speedup and reduced the optimisation time down to less than 3 minutes.

- "Highly Scalable Multi-Objective Test Suite Minimisation Using Graphics Cards", Shin Yoo and Shmuel Ur, Proceedings of the 3rd International Symposium on Search Based Software Engineering (SSBSE '11) [pdf] [bibtex]



Mutating Robots for Better Testing at ABB

ABB is not only a world-leading manufacturer of industrial robots but also a supplier of robots used in the latest Terminator movies. Yue Jia has been collaborating with ABB to improve the quality of ABB's robot testing by adopting mutation testing.

-Constructing Subtle Faults Using Higher Order Mutation Testing, Yue Jia and Mark Harman, Proceedings of the 8th International Working Conference on Source Code Analysis and Manipulation (SCAM '08), pages 249-258 (Best Paper Award) [pdf] [bibtex]

Optimising Ericsson's Future Product Line

Search-based approaches have been applied to Requirements Engineering to help organisations to decide which features to include in the next release of their product. However, in today's fast-changing market, it is important to balance the requirements need of today against those of future. Dr. Yuanyuan Zhang has collaborated with Ericsson AB to apply search-based approach to study the trade-off between the requirement values of today and tomorrow.

- Today/Future Importance Analysis, Yuanyuan Zhang, Enrique Alba, Juan Durillo and Mark Harman, Proceedings of the 12th Annual Conference on Genetic and Evolutionary Computation (GECCO '10), pages 1357-1364 [pdf] [bibtex]


This page was last modified on 10 Oct 2011.