Fuzz Testing for Quality Control in Systems with Complex Input Data

Sammanfattning

Fuzz testing is a testing technique used to generate a large amount of random or semi-random input data. This data is then fed to a target system which is then run with said data and monitored for anomalous behaviour. But as systems become increasingly complex, and as such, their input, fuzz testing becomes less efficient as pure randomisation no longer yields many useful results, and the long execution chains that may arise from complex systems create a demand for configurability in order to generate useful test data and make the testing efficient long-term. This thesis applies high-level configurability to a fuzz testing tool and tests this on a proprietary hard real-time operating system. The results show that this approach might not work all that well on the target system used during this thesis, but it is still believed that it is an approach to fuzz testing which may be useful in other regards.