SGER: Effective Theories of System Design

Kestrel Institute, Palo Alto CA

Investigators

Abstract

CCF - 0737840 Effective Theories of System Design Smith, Douglas R. Kestrel Institute Effective Theories of System Design The overarching objective of this research is to develop scientific foundations for the mechanization of software system design. These foundations lead to tools that mechanize a significant fraction of the work in developing correct and efficient software systems from specifications of their requirements. The research focuses on the capture and reuse of design knowledge. Design know-how is the prime intellectual capital of an organization, and capturing it in mathematical form enables both machine application in current design projects and communicating design knowledge to younger workers. The research seeks to integrate requirements engineering, formal specifications, mechanized refinement, and generation of correct-by-construction system code. The least developed aspect of this approach is tool support for the mechanization of refinement steps, which incrementally add implementation detail. Refinement machinery must be informed by codification of best-practice design knowledge about system architectures, algorithms, data structures, GUIs, and so on. The specific technical objective is to codify system design knowledge by developing the upper reaches of a taxonomy of abstract system designs, generalizing previous research on taxonomies of algorithm theories, data structures, and others. A taxonomy of system design theories would not only organize a large body of design knowledge, but would directly support mechanized design by providing incremental access to knowledge that is relevant to a specified design problem.