Issues in Enhancing Model Reuse

INTRODUCTION Reuse of existing models can significantly reduce development time and costs and improve the quality of newly developed simula tions. However, reusing existing mo dels is difficult in practice and is the focus of much research in the simulation community [1-7]. It is a central objective of HLA (High Level Architecture) developed by the U.S. Department of Defense. HLA is well documented at the Defense Modeling and Simulation Office website [8]. We believe that improving model reuse is a grand challenge in simulation. While the problems of mo del reuse are similar to those of code reuse in general, we suspect model reuse is more complex due to the centrality of abstraction in modeling. For example, because of different objectives, alternative models of the same physical system may use different abstractions that result in executions that differ in accuracy of outputs (high or low fidelity) or execution speeds. Progress in model reuse will require significant development in several areas: 1) understanding what information is needed to support reuse and how it should be represented, 2) developing mechanisms, automated and manual, to collect and record this information, 3) understanding how to design for reuse, 4) developing analy sis and search tools to locate appropriate existing components, 5) developing the ability to determine when model reuse is desirable. We discuss some reasons for these difficulties and address some of what is needed to support reuse of existing components. Our goal is more to discuss several aspects of the reuse problem in the optimistic hope that articulation of a problem can make a contribution to its solution. However, some parts of the solution seem clear. For example, we argue that a key part of the solution must involve capturing the objectives, assumptions and constraints under which the original models were developed in a form that can be searched and analyzed.