Editorial 2Q96
Get connected!
This cry is not only heard in the PC world but also in the Real-Time
world.
There are several reasons for this.
First, the business world needs more and more real-time subsystems. You cannot
think of all the Internet connectivy without installing large communication
equipment which are essentially real-time machines.
Second,on the factory flooryou need more connectivity. More intelligence
goes to the sensors and actuators. But the whole system needs to be co-ordinated
and managed!
Therefore, it was high time to devote an issue to this connectivity problem
in small and large real-time systems.
The field can be subdivided in three major parts. First, there are the military
and aerospace networks like ARINC and Mil-Std-1553. In this field proven
technology and predictability are major issues. The data rates are low.
Second, there are the new factory floor networks like CAN, Fieldbus, InterBus-S,
P-Net, Profibus, WorldFIP..... too many to mention them all. We could say
that there is literally a war going on between these networks due to the
high commercial interests involved. Standardisation seems to be a major issue,
but as they all become standardised....
Third, there are the high-speed network interconnects needed in high-demanding
applications which are being increasingly distributed over longer distances.
Here I think ATM will play a major role. Today a Quality of service feature
is included in ATM. This makes it possible to set up predictable connections
between two points. Just what we need for real-time applications. And 5 years
from now on. ATM will be a commodity with very low silicon prices. What more
do you want?
It is impossible to deal with all these topics in detail in one single issue.
We have selected the best of the articles which were submitted. As you will
observe, not only hardware aspects are dealt with. Protocol issues are equally
important. Some articles even describe how to use the classic network protocols
on parallel busses which may be considered as a specialform of a network.
There is, and should be, more migration between the software on serial and
parallel busses.
One example is the plug and play technology coming up in the PCI world. This
technology is far from mature, butyou may not expect to have everything in
one day andnot even in a life-time. This technique looks very promising.
Similar techniques are needed in the factory floor bus environment. But this
will probably take another decade and a bunch of standards to achieve such
elegant solutions.
One world should take the other as an example and stop reinvesting the wheel.
Isaac Newton said: "If I have seen further than others, it is because I stand
on the shoulders of giants! " Computer scientists should start work every
morning, keeping his words in mind. Of course the problem is to know whether
a similar solution exists in this other world. This is one of the purposes
of our Real-Time Encyclopaedia (http://www.dedicated-systems.com): "creating
the shoulders".
Dr
Martin Timmerman