System Maps: Integrating Knowledge System Models

When we design and develop a system, knowledge about the problem and solution is typically fragmented across a number of tools and models -- and the minds of the engineering team. Unified system modelling tools capture the overall problem and solution at a certain level of abstraction, but detailed knowledge and decisions remain scattered across multiple environments. A great deal of the time and effort expended on large projects such as the SKA radio-telescope goes into the problem of flowing information across development silos and maintaining consistency among them. We are exploring the possibility of creating SystemMaps, a unified repository for system knowledge that can serve as an underlying integrating layer across the various tools and models. Our goal is to explicate and capture the semantics of the relationships among the various models, to enable consistency checking among them. Our starting point is to organize the space of system models into a dimensional framework, such that consistency relationships among models lie along the dimensions. This paper discusses the dimensional framework, and our proposed research directions for working through the problem of creating SystemMaps using a knowledge-centric approach.