Model based self adaptive behavior language for large scale real time embedded systems

At Fermi lab, high energy physics experiments require very large number of real time computations. With thousands of processors (around /spl sim/1000 FPGA's, /spl sim/2500 embedded processors, /spl sim/2500 PC's and /spl sim/25,000,000 detector channels) involved in performing event filtering on a trigger farm, there is likely to be a large number of failures within the software and hardware systems. Historically, physicists have developed their own software and hardware for experiments such as BTeV [J.N. Buttler (2002)]. However, their time is best spent working on physics and not software development. The target users of this tool are the physicists. The tool should be user-friendly and the physicists should be able to introduce custom self-adaptive behaviors, since they can best define how the system should behave in fault conditions. The BTeV trigger system is being used as a model for researching tools for defining fault behavior and automatically generating the software. This paper presents a language to define the behaviors and an application scenario for the BTeV system and its expected fault scenarios. These self adaptive system tools are implemented using model integrated computing. The domain specific graphical language (DSL) is implemented within the generic modeling environment (GME) tool, which is a meta-programmable modeling environment developed at Vanderbilt University.