||Embedded systems are increasingly complex and have tight constraints in terms
of cost, performance, energy consumption and reliability. Embedded systems
have to be designed such that they correctly implement the required functionality. In the case of real-time embedded systems, the correctness of an application
depends not only on the results of the computation, but also on the physical
instant when the results are produced.
Modern embedded systems design relies on models for the application behavior and hardware platform. Starting from these models, the embedded systems
system-level design tasks are responsible for finding a model of the implementation that satises all the imposed constraints. The Unied Modeling Language
(UML) is a commonly accepted modeling language for modeling complex systems. Recently, an extension called Modeling and Analysis of Real-time and
Embedded systems (MARTE) has been proposed as a standard way to model
embedded real-time systems.
MARTE is currently under development, and supports two real-time operating systems (RTOS): OSEK/VDX, an automotive RTOS, and ARINC, which
is used in avionics. The rst objective of the thesis is to evaluate the currently
proposed MARTE extension, for which no documentation is available beyond
the specication document, and determine how it can be used to model real-
time embedded systems. The main objective of the thesis is then to develop a
generic library that is not tied to a specificc RTOS, thus allowing designers to
use MARTE in the early stages of the design process. Such a generic library is
also useful for education purposes, and for enabling the rapid development of
other, OS-specific, libraries.
We have implemented the generic library in the open-source Papyrus UML
modeling tool. Once a model is developed in Papyrus, we are interested to deter-
mine if the system is schedulable. We have proposed an automatic translation
technique from Papyrus MARTE models into input for MAST (Modeling and
Analysis Suite for Real-Time Applications), which is a state-of-the-art schedulability analysis tool used in the academia. Through the use of examples, we
have shown that, by using the proposed modeling method, real-time embedded
systems can be successfully designed and analyzed.