Methods for the Domain-Spanning Conceptual Design

The development of self-optimizing systems is a highly interdisciplinary task, as several domains are involved. Existing design methodologies do not adress this issue, as they focus on the respective domain; a holistic domain-spanning consideration of the system occurs - if at all - only rudimentally. The partial solutions developed by the respective domains may be optimal from the point of view of this domain. However, it does not automatically mean, that the sum of the optimal domain-specific solutions forms the best possible overall solution: ”the whole is more than the sum of its parts”. This especially holds true for the early design phase, the conceptual design. Its result is the so-called principle solution, which is further refined in the domain-specific design and development. Thus, a great need for methods arises which support the domain-spanning conceptual design for self-optimizing systems in a holistic manner. In this chapter we will introduce such methods. In particular, we will explain the specification technique for the domain-spanning description of the principle solution of a self-optimizing system. Furthermore, methods are explained which support the creation of the principle solution. This includes a method to ensure the consistency of application scenarios, a method for the design of the system of objectives, which is crucial for a self-optimizing system, as well as a method for the re-use of proven solutions for recurring problems (solution patterns). Finally, some analysis methods are explained that are performed on the specification of the principle solution. These are: the early analysis of the reliability and the analysis of the economic efficiency.