A Meta–Modeling Concept for Embedded RT–Systems Design Benito Liccardi Thomas Maier–Komor Muslim Elkotob Johann A. Oswald Georg F¨ arber Institute for Real–Time Computer Systems Prof. Dr.–Ing. Georg F¨ arber Technische Universit¨ at M¨ unchen, Germany Benito.Liccardi,Thomas.Maier–Komor,Muslim.Elkotob,Hans.Oswald,Georg.Faerber @rcs.ei.tum.de Abstract The work in progress presented in the present paper is focused on the HW/SW Codesign of embedded systems with a special emphasis on the Bottom–Up design process. A Meta–Model concept is introduced that gives the ability to create strict, computable and unambiguous component de- scription, based on the OMG Standard MOF and the W3C recommendation XML Schema. The concept is currently be- ing implemented and verified with a real world embedded system for medical applications. 1. Introduction The domain of medium volume systems is character- ized by a complex heterogeneity of application specific con- straints. Methodologies that are well understood in the rapid prototyping or high volume system domain fail in this area, because of their inflexibility and scarce adaptability. The key issues for developing these medium volume sys- tems with a definable time to market is based upon two main aspects: abstraction in the design process of hardware and software and reuse of already designed entities with the help of architecture platforms [6, 13]. While the former is mostly controlled by a top–down design process, the latter is also driven by a bottom–up process. The problem with the bottom–up part is mainly that it can be characterized as a reverse engineering task. The de- signer has to acquire knowledge from documents in order to be able to know if the composition of the selected system components will fulfill the requirements. This task is not only error prone but also done from the very beginning in case some major component substitution will happen. Fig- The work presented in this paper is supported by the Deutsche Forschungsgemeinschaft as part of a research program on “Embedded Sys- tems” under Grant Fa 109/12-3. ure 1 shows the dilemma of the reverse engineering gap dur- ing the development process. Chip A Manual A Chip B Manual B Designer Embedded System creates Engineering Process Components Requirements Reverse Engineering Process Knowledge Components Engineering Process Systems Knowledge+ Requirements Reverse Engineering Process ? Components Requirements Figure 1. Reverse Engineering Gap The designer has to cope with informal, uncomputable and different manual types, that sometimes do not include all the information that is needed and sometimes even are not up to date. Additionally complex ICs have fairly exten- sive manuals, which often lead to conflicts ( e.g. double as- signment of an I/O Pin or an unavailability of HW functions due to already used timers) and which in turn can cause ex- pensive redesigns within very late phases of the design pro- cess. Thus we developed a Meta–Model called ATLAS (Ar- chitecture Templates for Embedded Systems Design Meta Model) 1 for describing already existing hardware and soft- ware components on a very abstract level. This abstraction level allows us to provide strict, computable and unambigu- ous component descriptions. The result is that the ability of composition is realized by well defined semantics of the components and architectures. 1 the Architecture Template will not be described in this paper, but is built upon the Meta–Model that is introduced in chapter 2 1