The increasing complexity of real time software has led to a recent trend in the use of high level modeling languages for development of real time software. One representative example is the modeling language ROOM (real time object oriented modeling), which provides features such as object orientation, state machine description of behaviors, formal semantics for executability of models, and possibility of automated code generation. However these modeling languages largely ignore the timeliness aspect of real time systems, and fail to provide any guidance for a designer to a priori predict and analyze temporal behavior. We consider schedulability analysis for automated implementations of ROOM models, based on the ObjecTime toolset. This work builds on results presented by M. Saksena (1997), where we developed some guidelines for the design and implementation of real time object oriented models. Using the guidelines, we have modified the run time system library provided by the ObjecTime toolset to make it amenable to schedulability analysis. Based on the modified toolset, we show how a ROOM model can be analyzed for schedulability, taking into account the implementation overheads and structure. The analysis is validated experimentally, first using simple periodic models, and then using a large case study of a train tilting system.

We present our experiences in applying real time scheduling theory to embedded control systems designed using ROOM (Real time Object Oriented Modeling) methodology. ROOM has originated from the telecommunications community and has been successfully applied to many commercial systems through the supporting case tool ObjecTime. It is particularly suitable for modeling reactive real time behavior. Furthermore, it provides many other advantages through the use of object orientation, and the use of executable models from which code may be generated quickly and efficiently. Since many real time embedded control systems have significant reactive, event driven behavior, it is attractive to use ROOM methodology to develop such systems. However, the ROOM methodology does not provide tools to specify and analyze the temporal behavior as is required for the hard real time components of embedded systems, and for which the real time scheduling theory provides an analytical basis. We show how real time scheduling theory may be applied to ROOM models using a cruise control example to illustrate. The biggest challenge comes from minimizing the adverse effects of priority inversions. Our results are very encouraging, and we show that not only is it possible to apply real time scheduling theory, but that it can be done very efficiently provided certain guidelines are followed in the design and implementation of the ROOM model.