PREMOS: Tools for model-based programming.

摘要

In the model-based programming approach to building complex embedded systems, such as monitoring and control systems, the application is generated from the model of the system, its environment and their interactions. This greatly increases the system's autonomy, maintainability and reliability. Model-based programming environments support the building, and maintenance of models and facilitate the automatic generation of an application from the models through interpretation.; This thesis centers around the question of how to build model-based programming environments in general. A new level of model-based programming is introduced. The commonalities of model-based programming environments are analyzed and conceptualized in the PREMOS framework. A general system structure is proposed which integrates the interpretation process with the generated embedded system in order to allow reconfiguration at run-time. Furthermore, it is argued that a centralized interpretation process often cannot satisfy the scalability and real-time requirements of the generated system. To overcome this problem a parallel, distributed interpretation mechanism is introduced.; The PREMOS framework is realized as a set of generic reusable tools for model representation and distributed model interpretation. A meta language, called MOPSEL, is used to specify an application domain. The specification is compiled and used to customize the set of reusable tools. As a result, the efforts required for implementing a model-based programming environment are greatly reduced. Furthermore, the distributed character of the interpretation is completely handled by the PREMOS system and does not require any additional implementation by the system developer.; The development of a model-based programming environment is supported by advanced tools, such as a graphical monitor and a general distributed computing platform.; The PREMOS approach is applied to the generation of a model-based programming environment for the signal processing domain. The generated environment is then used to build a complex application. The results are analyzed to demonstrate the feasibility of the approach.