Embedded Data Fusion Toolbox: A Model Driven Architecture

Reducing time to market, design cost and fast developing, prototyping and testing of embedded industrial and automation models are very interesting topics in recent manufacturing and industry methodologies. In recent years, Model driven architecture or engineering methodology has been introduced in the field of software engineering by OMG. This methodology separate platform independent, and platform specific concerns, and could be utilized and extended to other fields of engineering. In many industrial applications, embedded, distributed and fast sensory data fusion operators are necessary, and system complexities could be dominated by low level dedicated fast fusion operators on real time hardware platforms such as CPLDs and FPGAs. In this experience, embedded data fusion toolbox has been developed for fast design, prototyping, test and implementation of simple fusion operators on embedded industrial platforms. This toolbox has been implemented in MATLAB/Simulink environment via behavioral arithmetic to the implementation on either FPGA. Reusability, scalability, chip area, precision and other design criteria are considered in design process, and desired HDL models could be obtained automatically. This fully automated toolbox is extendible for complex data fusion operators, subsequently the platform-independent model can be translated to a platform-specific model (PSM) by mapping the PIM to some implementation language or platform using formal rules in order to preparing targeted models in different platforms.