Browse by author
Lookup NU author(s): Dr Salah Merad, Dr Rogerio De Lemos, Emeritus Professor Tom Anderson
Software components are characterised by their functional and non-functional properties. It is both difficult and expensive to build a software component that excels in all its non-functional properties and that can cope with a large input range for all environmental variables. We envisage to provide a computing system operating in a changing environment with a pool of software components, each having a specific profile, and a decision maker that selects the best component subject to the time varying environmental requirements. The resulting decision problem is a multi-attribute optimisation problem. In this paper, we propose several solution concepts for this problem and evaluate them qualitatively with respect to many criteria. Two of these concepts are treated in detail. The first is based on the combined utility function and the second is the compromise between the preference patterns of the various properties. The latter is game theoretic, and the solution is found by formulating the problem as a bargaining game. While the combined utility function can be expressed in a simple form when some independence assumptions between the properties hold, and it can incorporate the decision makers value trade-offs, it involves evaluating subjectively a number of coefficients. But because the evaluations also need to be performed during run-time for new requirements, this becomes problematic. The game theoretic solution does not incorporate explicitly the decision makers value trade-offs, but it is free of the subjective element and no evaluations are needed during run-time.
Author(s): Merad S, de Lemos R, Anderson T
Publication type: Report
Publication status: Published
Series Title: Technical Report Series
Year: 1999
Pages: 32
Print publication date: 01/03/1999
Source Publication Date: March 1999
Report Number: 664
Institution: School of Computing Science
Place Published: University of Newcastle upon Tyne
URL: http://www.cs.ncl.ac.uk/publications/trs/papers/664.pdf
