Model-integrated development of complex applications

Abstract

Many large distributed applications are tightly integrated with their physical environments and must be adapted when their environment changes. Typically, software development methodologies do not place a large emphasis on modeling the system's environment, and hence environmental changes may lead to significant discrepancies in the software. The authors argue that (1) the modeling of the environment should be an integral part of the process, and (2) to support software evolution, wherever possible, the software should be automatically generated. They present a model-integrated development approach that is capable of supporting cost effective system evolution in accordance with changes in the system's environment. The approach is supported by a "meta-architecture" that provides a framework for building complex software systems using COTS and custom developed software components. This framework has been successfully used in various projects. One of these projects, a site production flow visualization system for a large manufacturing operation, will be analyzed in detail. First, they show how the model-integrated process can be generalized and used to build families of model-integrated tools that support the development of specific systems. Next they describe how the generic architecture was customized for the particular domain and discuss how specific components were implemented and present a detailed experience report (both from developers and end-users).