Representing heuristics for architectural optimisation

University of Technology, Sydney. Faculty of Information Technology.NO FULL TEXT AVAILABLE. This thesis contains 3rd party copyright material. The hardcopy may be available for consultation at the UTS Library.NO FULL TEXT AVAILABLE. This thesis contains 3rd party copyright material. ----- The design of computer-based systems is an endeavour that is facing an ever increasing degree of complexity. With the scale and ubiquity of computer-based systems increasing, the demands placed upon designers to not only understand the systems they are creating, but also to create the best possible system, are testing the limits of the techniques and skills available to designers. Nowhere is this more obvious than in the attempts to manage the non-functional qualities of complex computer-based systems. The research presented in this thesis investigates the idea that one method that shows great promise for managing the trade-off between the non-functional qualities of a system is through the optimisation of its design. Specifically, by optimising the architecture of the system, the conflict between the non-functional qualities can be reduced and the best potential trade-off identified. Design optimisation proposals are often severely limited in their usefulness due the manner in which they represent and generate new designs. To work around this issue, a heuristic-based optimisation approach is presented that is capable of using existing expert design knowledge to identify design deficiencies and propose potential solutions. The framework utilises expert knowledge, captured as heuristics, for this purpose. The principal focus of the research presented in this thesis has been the development of a framework, model and formal foundation for heuristics so that they may be used in architectural optimisation. Through the development of these elements, a solid representation for heuristics has been elicited. These provide for the capture and application of heuristics to improve architectural design. Finally, the results of a design study that was performed to assess the ability of the proposed heuristic framework and model to capture the heuristics are presented. This investigation showed that heuristics could be captured from design experts and codified using the proposed model. It is also showed that the formal foundations of the heuristics allowed them to generate new architectures, and new system designs. The designs generated provide strong evidence in support of the use of heuristics for optimisation and the trade-off of non-functional system qualities