Scientific paper ID 1280 : 2016/1

The paper is dedicated to a study on the problems of safety critical software in various applications. The following logical structure is proposed: there must be no mistakes made in any of the stages and levels of software development. For that purpose, some approaches, formal methods and tools of programming created by science as well as operating standards for critical software are used. However, in complex systems errors continue to exist. To detect and remove errors before putting the system into operation, approaches, methods and tools for verification as well as off-line testing are applied. Nevertheless, a number of mistakes still remain in operation, some of which result in limiting functionality while others can cause dangerous incidents. The unfavourable effect of the former and the dangerous effects of the latter can be limited if errors are found during operation (on-line), the consequences are stopped and they are removed or tolerated not to appear. It is concluded that the problems of safety critical software are connected with the standards of admissible risk, methods for building safe and fault-free software, methods for building structures of fail-safe and fault-tolerance SCS, the assessment of post-failure danger, the possibility to reduce danger as well as models of reliability and safety evaluating the impact of software errors.