An Approach for Dealing with Dynamic Multi-Attribute Decision Problems
Rachel Edith Baker · 2005
There will be a requirement for future systems to increasingly undertake self-management functions, one aspect of which is the dynamic resolution of what are, effectively, multi-attribute decision problems. Such problems involve acts of choice and selection based upon the consideration of several issues. Whilst there are many techniques to guide a decision maker to a satisfactory solution where problems arise once, and in a static context, these may not be readily suitable for dealing with dynamic problems encountered in a system's environment. This thesis investigates the definition of multi-attribute decision problems within a real-time system. It proposes a methodology by which to structure the definition phase of the problem and to help in the identification of an appropriate evaluation methodology. Adopting a reasoned approach towards the definition of the problem also allows for the simultaneous description of adaptive behaviours by which to improve the robustness of the evaluation method without undermining the integrity of the results it produces. ii