Managing the evolution of dependability cases for systems of systems

Dependability is a composite property consisting of attributes such as reliability, availability, safety and security. The achievement of these attributes is often essential for the operational success of systems undertaking critical and complex tasks. Assurance that the final system will demonstrate the required dependability qualities, can be crucial to the acceptance of the system into service. Safety cases are a well established concept used to establish assurance about the safety properties of a system. However, safety cases focus only on one attribute of dependability. The principles and processes of creating an integrated dependability case – that assures all aspects of dependable system behaviour – are less well understood. A number of challenges are faced when attempting to support dependability case development. These include the systematic elicitation of dependability goals, the management and justification of trade-offs, and the evolution of multi-attribute arguments in step with the design process. This thesis addresses these challenges by defining a rigorous framework, accompanied by a set of methods, for establishing dependability cases. Firstly, a method for eliciting dependability requirements is defined by extending existing safety deviational analysis techniques. Secondly, a method for systematically identifying and managing justified trade-offs is presented. Thirdly, the thesis describes the co-evolution of dependability case arguments alongside system development – using a dependability case architecture that corresponds to system structures. Finally, the thesis unifies these contributions by defining a metamodel that captures and interrelates the concepts underlying the proposed methods. Evaluation of the work is presented by means of peer review, pilot studies and industrial examples.