General Impressions

The people attending RTAS are mostly North Americans (US and Canada) (most of the European real-time researchers tend to attend the Euromicro Workshop on Real-Time Systems that takes place at approximately the same time). About half the participants where from industry, and many of the US participants were working with defense-related projects. It is obvious that the Department of Defense (DoD) is an important financier for US real-time research. I estimate the number of participants to be about seventy.

The conference vista was the Sheraton Wall Center hotel, which was a very good hotel (if a bit pricey, but apparently Canadian hotels tend to be expensive). The city of Vancouver is very nice: snow-covered mountains in the backdrop, but still warm enough to support a long stretch of beautiful beach by the Vancouver bay. Downtown is pedestrian-friendly (almost like a European city) and there are plenty of good restaurants.

Technical Presentations

The standard of the accepted papers was quite good, and the program was well-varied, with sessions on everything from scheduling theory through databases and worst-case execution time analysis. There was a large number of interesting case studies presented. In general, US real-time research tends to deal with larger systems than European research: there were more material on COTS hardware, distributed multiprocessor systems, middleware, etc., and less on scheduling theory and verification of hard real-time systems than Euromicro. I will not try to point out any "outstanding" papers, since it what constitutes outstanding depends on your research interests.

The conference contained a very interesting tutorial on real-time Java, one on real-time communications, and one on real-time CORBA.

Obvious trends were that Windows NT was being considered for (soft) real-time applications, and the tendency for real-time systems to become large, distributed, and based on databases. Command and control systems are very important to the US military.

The real-time Java people seem to have narrowed down their target applications to applications where a GUI is used together with some soft real-time applications. The main competitor is considered to be Windows CE.

I presented a paper on the properties of actual real-time programs, and the implication for the construction of worst-case execution tools. Mikael Sjödin (Uppsala) presented a paper on real-time call admission for ATM networks.

For IEEE Computer Society members with an active E-account, the conference proceedings are available online

Recommendations

RTAS is a good conference, both as a target for paper submissions and as a conference to go to to learn about recent advances in the real-time field. The large number of case studies provides inspiration and information, and many interesting discussions can be held during the coffee breaks and lunches.

Submitted: Mon, 09 Aug 1999 15:20
By: Jakob Engblom
At: Uppsala University