Model Driven Architecture of Large Distributed Hard Real Time Systems
Michael A Giddings, Thursday 3rd June, 10:30am - 10:45 am, Student Presentations 1, JLB Meeting Room 10
Cloud created by:
28 May 2010
He is going to start with describing what distributed is. He indicates that distributed when there are two processors (?) working.
Speaks of two types of real-time systems: hard and soft
He investigated performance in real-time systems and indicated that in the literature there seems to be a fix-it-later approach. This is not a problem when the system is simple or when there is a prototype or extending an existing design or where there is number of releases. However, it is difficult when there isn’t a large production run and uses hardware.
He lists the types of system features that have been studies including client server, event driven, telecommunication systems.
He discovered a real-time system call process control system using open/closed loop systems.
Temporal considerations: reaction time, accuracy, reconstruct data (e.g. in telecommunication systems), latency, capacity
He is now showing a time process control system (talking about states and how they change). Ok – think he has lost me now (my poor computing processing knowledge!).
He has designed a graphical representation of functional elements in the real-time system. The advantage of producing a functional model is that you can produce a performance chain which helps you to extract the things that influence the end event and excludes those that don’t.
One problem is discovering problems at the end of development which is difficult to fix/ expensive at the end of the production. Thus the method has the ability to model real world requirements and model functionality and model user interface etc.
How does it help in non-functional requirements? If you can find the non-performance chain (?) then you can annotate all the non-functional requirements and manage where they occur.
09:25 on 3 June 2010