Analyzing Software Safety

TLDR: This paper defines software safety and describes a technique called software fault tree analysis which can be used to analyze a design as to its safety and has been applied to a program which controls the flight and telemetry for a University of California spacecraft.

Abstract: With the increased use of software controls in critical realtime applications, a new dimension has been introduced into software reliability–the "cost" of errors. The problems of safety have become critical as these applcations have increasingly included areas where the consequences of failure are serious and may involve grave dangers to human life and property. This paper defines software safety and describes a technique called software fault tree analysis which can be used to analyze a design as to its safety. The technique has been applied to a program which controls the flight and telemetry for a University of California spacecraft. A critical failure scenario was detected by the technique which had not been revealed during substantial testing of the program. Parts of this analysis are presented as an example of the use of the technique and the results are discussed.