Design patterns as components of functional models for analyzing the reliability of software systems

The goal tree-success tree provides a technique for developing models of a system by successively breaking a high-level function into subfunctions until those subfunctions can be realized by physical components. Since software is not perceivable in a physical sense, tests of maintainability and reliability are more abstract than those of physical systems. Software design patterns can support analysis of complex software systems by serving as analogs of physical components. In this manner we have developed a goal tree-pattern tree model which combines the techniques of functional modeling and the goal tree-success trees of physical systems. At one level of abstraction in a GTPT, the components of the patterns themselves are available for traditional fault tree analysis techniques. At a higher level of abstraction the individual components are captured in a "pattern component" which greatly simplifies the reliability analysis of complex software systems.