The following content has been provided by the University of Erlangen-Nürnberg.

Okay, so I'm going to talk a little bit about predictability, real-time oriented topic of course,

not just for real-time but also cyber-physical systems and I'll spend probably more time on the Unicore side,

but a little bit about multi-course as well.

I don't know what your background is but I'm just going to make sure everybody can follow the slides.

So what is a cyber-physical system?

I have a few definition slides up front.

The idea, I mean I didn't coin this, okay, so I don't take credit for this terrible acronym

or even the combination of words, but the idea is simply that you have some control system

that is in the loop, it affects humans and most importantly is networked.

And these systems interact with one another of course over short or long distances due to the networking.

And again I'll go quickly over these introduction slides.

The other part I want to focus on is real-time systems.

Turns out a lot of the cyber-physical systems have real-time aspects, they are control systems,

and then if they have that you typically distinguish between hard and soft real-time systems,

so the soft ones you have deadlines, but if you miss some of them it's okay,

you don't want to miss too many, and the hard real-time systems is really what most of my talk

will focus on are the ones when you miss a deadline potentially,

your system is dead in the water in a sense that some catastrophic effects are on the environment,

loss of life, whatever.

What they have in common these real-time systems is that there is a need for predictability

in terms of the computation, so usually you're just concerned about I-O relationship

for a given input you want to have the correct output.

Here the temporal aspect comes into play and you want to make sure that you also finish a given piece of code in time.

So in real-time systems there are certain terminology, I will touch on that a little bit.

Most importantly you have to understand a computational model

and the more commonly used computational model is that of a periodic system

where you have tasks and jobs of these tasks are repeatedly executed in a given period of time.

They also have deadlines and you want to make sure that their execution finishes by this deadline.

They don't have to necessarily be executing right away as they are released, execution may be slightly deferred,

and all of this is reflected essentially in a set of terms, terminology over here.

We'll talk about response time, that's basically the time that it takes to finish over here

and you have to make sure that that's smaller than the deadline and there can be some blocking effects here as well.

But the important part is real-time systems don't necessarily have to be fast, they just have to be on time.

It means you have to make sure that your execution over there finishes.

In order to guarantee that you actually have to know how much execution you have in here

and of course this varies depending on the input, sensor inputs,

but there is a notion of a worst-case execution time which is the longest possible computation time for a given task

and this is actually something that we'll focus on in the following.

Also because there is a documented need in embedded systems for tool support,

it's something that you don't really want to throw at the developer, the developer wants to have support for that.

I'll put this into a context because I have some nice pictures and maybe also because Airbus is in the news,

but most of all because if you think about the automation that's going on in these systems and especially in the 380 and also the Boeing 787,

it's mind-boggling in terms of flying a plane with a joystick and actually different control systems.

I'm skipping over some of the details, I usually have some more slides on this.

Rises of interest, we actually were involved in a subsystem that was developed for the 380,

one that's not the most critical one, in fact in traditional jets that you're flying in most of the time like say an A320 or a 737,

they wouldn't be considered critical, so it's the air heat system.

It's not critical because if you have some problem in there, typically it can be contained and it doesn't necessarily cause the plane to crash.

But what's different in these new planes is that they are a hybrid of carbon and metal, which is flammable,