Ambrosia/MP is the successor of the realtime operating system
Ambrosia is the abbreviation for "AModular Basic Realtime Operating
System In Automotive units".
The Appendix /MP means "for Multi Processors".
Ambrosia was developed at the
University of Hildesheim in coorporation with the Volkswagen AG.
The projectleader was Frank Plagge. The project ended at June 1997.
In comparison with classical computer systems there is great importance in embedded systems
for most efficient use of available resources. Besides to
the hardware cost this is a question of such systems size.
In automotives there usually can be found more than one embedded system today
in order to handle complex tasks like gear and engine control or driver assistance systems.
It it foreseeable that in future the existing controllers in automotives will be
connected through communication networks. So the available resources can be used
in a distributed manner.
In the project Ambrosia/MP an operating system based on the system Ambrosia will be developed which
provide mechanisms for optimal resource utilization within distributed embedded real-time systems.
The work includes the introduction of a model used for schedulability analysis of real-time application
and as a base for process partitioning. Furthermore in the project a framework will be created
in which the whole engineering process can be embedded.
Distribution and scalability
One of the features of the already existing system Ambrosia is to provide
runtime systems which are mostly adopted to the requirements of the respective applications.
For this during the development process the runtime system is assembled based on so
called components. Hence the arising system contains exactly the parts to handle the
underlying hardware and to provide the system services required by the application.
One goal of this project is to extend this feature to distributed systems. Therefore
it is nessesary not only to configure the individual controllers but also the interconnection
networks. With respect to the scalability Ambrosia/MP should support diverse communication
structures. Therefore Ambrosia/MP can be used to develop real-time systems
with a wide range of underlaying hardware.
In addition the system will provide mechanisms to allow the synchronisation across
Models, partitioning and schedulability
An essential problem within the development of real-time applications is the
adherence to given time limits. Starting from a set of processes the
task is to place these processes to the available controllers such that every
process is able to match its deadline under all conditions.
In real systems the determination of the schedulability is difficult if not impossible.
One of the main problems is the integration of the system environment in such
tests. Hence it is advisable to use models in which formal criteria can be obtained to
determine if an application is schedulable or not. In this model each process is descibed
by a set of properties like processing time, interarrival times and deadlines.
Within the model rules can be established which provide a formal predication to the
schedulability. Afterwards tests can be used to verify this predications. In this
project such a model will be developed.
The development of applications based on Ambrosia/MP is rather complex. Therefore
in this project a framework should be provided to simplify this task. The primary
goal is to create a platform which support the complete development of
Ambrosia/MP applications. This platform consists of an interface between
the system creation and the GUI (graphical user interface). With this interface it is easy to
implement development tools within an integrated development environment.
Frank Plagge Ambrosia - Ein Echtzeitbetriebssystem für eingebettete Steuergeräte
Ph.D.Thesis, Universität Hildesheim, 1998 (full report online)
(available by VDI Verlag GmbH, Düsseldorf, 1998, ISBN 3-18-356010-0)
Alan Burns Preemptive priority based scheduling: an appropriate engineering approach
technical report, University of York, UK, 1993 (full report online)
A. K. Mok Fundamental Design Problems of Distributed Systems for the Hard-Real-Time Environment
Ph.D.Thesis, TR-297, Massachusetts Institute of Technology, USA, 1983 (abstract)(order)