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

This talk is actually split into two parts and the reason is essentially because I moved from the operating system group in Dresden,

from the excellence cluster to Luxembourg where we do now intrusion tolerance and therefore a little bit about me.

I was born in Frankfurt and then I had the pleasure to study with Jochen Liedtke in Karlsruhe before he died.

So learning from the inventor of L4.

Then I spent quite some time in Dresden which finally also explained my involvement in the excellence cluster.

In between I had a short trip to Rosse in France where I did a little bit of hardware development

and also I spent half a year with hybrid system verification with Andre Platzer at CMU.

But then finally I moved to S&T where I'm now working with Poloverismo.

Our mission in Luxembourg is intrusion tolerance,

kind of preparing our systems to defend against your famous three letter organization,

trying to penetrate your system.

Today I want to first share a little bit of what we did in CFAED and what we also published at Asplos this year.

In the second part I would then go into how this architecture which we developed can be reused, repurposed,

in order to better protect our infrastructure.

Unfortunately I cannot really talk about the solution to the last problem,

but I can at least tell you the problem, why many cores are not distributed systems,

as we are used to talk about when we talk about intrusion tolerance and by some type fault tolerance protocols,

simply because a few of these patent lawyers are still involved.

But perhaps next time, and if you guess you may get an idea of what we are currently working on.

So after receiving the excellent cluster, while preparing it,

we thought how can we complement a lot of people working in these material layers,

trying to develop new materials such as carbon nanotubes, silicon nanowires, printed electronics,

and even weird things which you see on the right thing,

where you have actually a device which processes chemical information,

and which does computation just by different chemicals put together on top of these laps on the chip.

And how can we make use of whatever they develop in order to benefit from these kind of material advantages at the application level.

So really the question was how can we bridge this gap.

And the first problem of course is when you are starting such a joint project,

is you first have to understand how to talk to these people.

If they are talking about a device that is completely different to your network card or something else, it's a single transistor.

And so that's the first challenge.

And after a while we really understood what they are providing, what they are researching,

and in particular what kind of properties they give us.

So the first class of properties which they give us is what I classify as non-functional properties.

Like you get for example a circuit, and what you see here is a loudspeaker printed on a foil, a simple plastic foil.

And the benefit of this is it's terribly cheap.

So if this technology would be mature enough, I wouldn't print my laptop, I would just print my slides on the foil.

The wall of course would also be printed with the LEDs which you need in order to project this image, so no beamer anymore.

And I just tap on my foil and afterwards I just throw it away simply because it's so cheap.

So this would give us an enormous benefit of just customizing our circuits for the time and for the single use perhaps when we need it.

A similar technology is DNA origami which we are following in Dresden where using the DNA strand patterns we try to connect different DNA strings to form larger structures.

That's a structure which we built in Dresden using this kind of T shape here which we then could metalize in order to create a transistor because gate and essentially the channel.

But other structures have also been built in the US for example little smileys which look pretty nice.

Or what you see here on the right hand side, simply cables which are attached to two parts of a chip and which automatically assemble each other.

So this part of the cable, if you put it in water, heat it up a little bit, make it a nice environment for DNA and then cool it down,

automatically attaches to the opposite cable here at the molecular level.

So this would give us enormous interconnects between chip stacks for example to stack two chips on top of each other.

And if this TPAT idea evolves we can just put in all our logic which we have described in terms of DNA structure.

And then at the end we would have something where we simply cool down the water with all the DNA inside and out of it comes a chip custom made for us.