Reuse and Scalability in Modeling and Simulation Software Engineering

摘要

Studying a system using discrete-event computer simulations implies several activities: conceptual model speci cation, software model architecture description, software development, simulation scenario, instrumentation, experimentation planning, computational resources con guration, execution, post-processing and analysis, validation and veri cation (V&V). Many software are required to complete all these activities. However, it is common practice to create a simulator from scratch when starting a new a simulation study. It is therefore necessary to redevelop a whole suite of tools to ensure support for all simulation activities. This thesis addresses the challenge of developing new simulators that reuse existing models and simulator parts. Indeed, reusing software increases dependability, is less error prone, makes better use of complementary expertises, improves standards compliance, and accelerates development. Reusing software can be applied to all simulation activities. Several problems have to be solved to derive full bene t of reuse. In this thesis, we address three major issues: Firstly, we investigate practical means of reusing and combining valuable pieces of modeling and simulation software at large, including models, simulation engines and algorithms, and supporting tools for the modeling and simulation methodology; Secondly, we focus on issues related to instrumentation; Thirdly, we focus on problems of integration of existing simulation tools. To achieve these objectives, we investigate advanced software engineering techniques such as component-based software engineering (CBSE) and aspect-oriented programming (AOP), and use them to derive a novel approach for Modeling & Simulation based on reusable layers. We developed a prototype software architecture that proves the feasibility of this layered approach.