A model for the practical architecture-based design of computer based systems

University of Technology, Sydney. Faculty of Engineering.NO FULL TEXT AVAILABLE. Access is restricted indefinitely.NO FULL TEXT AVAILABLE. Access is restricted indefinitely. ----- Computer based systems are typically large, complex, distributed, heterogeneous, multi-faceted systems, whose sub-systems may be systems in their own right. CBSs regularly incur long and difficult development lifecycles, and the resulting systems often fall short of the required attributes, especially non-functional attributes, including performance, reliability, openness, and security. Such systems pervade a wide range of industry domains. As time continues, the level to which these systems pervade industry, and society’s reliance upon them, increases. The consequence of poorly designed systems which do not operate as intended can range from mere inconvenience (an automatic teller machine being down) to catastrophic (loss of life). With respect to computer based systems, design can be seen as a process of ensuring that the system's architecture will allow the system’s requirements to be met. Unfortunately, designers receive little practical assistance in achieving this. There is a great need for practical tools and methods for performing architecture-based design. Existing methods typically ignore the engineering practice and employ complex notations. To be used in a practical context, new methods for addressing the lifecycle problems of computer based systems must stem from those currently in use by practitioners. Graph-based methods are cited as being practical and commonly used by practitioners. This thesis addresses these problems by investigating how architecture-based design is currently performed; by studying current standards and practice based approaches. A generic model for architecture-based design is developed and evaluated against practical criteria. Further, architectural refinement is highlighted as a key aspect of performing architecture-based design. Graphs are cited as a practical means by which practitioners currently design systems. Accordingly, a graph-based mathematical model of architectural refinement is developed. The model of refinement is also evaluated with respect to its practicality. The first major contribution of this thesis is the specification for architecture-based design and the resulting abstract architecture-based design process that is developed. The second main contribution is the specific graph-based and mathematically defined formalism which models architectural refinement and abstraction