Travel Report from RTSS 2000

Ola Redell
Damek, Inst. för Maskinkonstruktion, KTH
ola@md.kth.se

Dec 11, 2000

In November this year I visited the Real Time Systems Symposium (RTSS'2000) in Orlando, USA. RTSS is one of the best known conferences in the area of real time systems and the papers presented are usually of high quality. The conference was preceded by a workshop on real-time Linux which I did not attend.

A paper by Saksena and Wang, "Scalable Real-Time System Design using Preemption Threshold" was specifically interesting as it relates to the problem of scheduling analysis together with object oriented design methods, e.g. ROOM and OCTOPUS. Generally such methods are based on the concept of active objects that respond to incoming signals using an internal non-preemptable scheduler. The paper presents a method for designing systems with such active objects.

Another interesting paper was the one by Palopoli et al. ("Real-time control system analysis: an integrated approach") who had implemented a tool that allows evaluation of control quality resulting from scheduling. Work that take an overall view of control system design and the following real-time implementation is rare but important as most of the embedded real-time applications are control systems.

A third interesting paper in a different field was the one by Lipari, Carpenter and Baruah ("A framework for achieving inter-application isolation in multi programmed, hard real-time environments"). It addresses how to isolate different applications from each other by using scheduling techniques.  The basic idea is that two applications executing on two separate dedicated processors, should be possible to implement on one single processor with preserved behavior for each application. This can be specifically interesting for safety critical systems in which e.g. a possible failure of one application should not affect the success of another potentially more critical application.

The keynote speaker - Richard Stalllman - gave a very interesting and entertaining talk on the GNU project (that he initiated) and free software in general (free as in free speech, not free beer!). He described how he had come to work full time on free software and how the GNU project, whose goal was a free operating system, became a part of the Linux OS. He urged us to start calling Linux "GNU/Linux" instead. Richard Stallman is obviously one of those rare idealists and I think his talk worked as an eye-opener for many of us.

In all, the presented papers held a high standard and most of the american part of the real-time community was represented. The week came to a perfect end when we got to see the launch of the Endeavour space shuttle at Cape Canaveral.

Date: Mon, 18 Dec 2000
From: Ola Redell ola@md.kth.se