Testing for Software Safety

Herbert Hecht, Testing for Software Safety, NASA Dependable Flight SystemsConference <

Abstract. The concerns of software safety are (1) that functional programs do not generate hazards, and (2) that safety monitoring programs execute flawlessly. The latter poses the bigger challenge, because when the program is called upon to act in an anomaly the scope of which cannot be completely defined in advance. There is also the need to minimize false alarms and there may be additional problems when the monitored function and the monitor share an environment.

The present approach to constructing test cases is empirical, determined by local precedent, with little global sharing of the effectiveness of test case selection criteria. Test reports typically concentrate on the performance of the system under test with very little effort devoted to advancing test technology. In view of the large resources that go into software safety testing it is recommended that a fraction of that amount be allocated to research into specific software safety areas that are not covered by either systems safety guidelines or software testing methodology.

Click to Download



© Copyright 2001 SoHaR Corporation. All rights reserved.