How IT drives Ferrari's success

Dieter Gundel, head of racetrack electronics at Ferrari, explains the vital role that IT systems play in car development, preparation and race-day action - and why he thinks wireless security is not yet secure enough
Written by Dietmar Muller, Contributor

Driving a race car takes more than good hand-eye coordination and the stomach for G-forces. It also takes a truckload of high-tech gear.

Ferrari's Formula 1 team -- and the results of seven-time world champion driver Michael Schumacher -- are heavily influenced by technology. This includes software designed in-house and hardware with standard components such as the Opteron processor from team sponsor Advanced Micro Devices.

A good deal of the high-performance work takes place off the asphalt in the "box", the garage where cars get their tyres changed during a race. That's where the computers collect the telemetry data on how the car is meeting the demands of the course.

Dieter Gundel, head of racetrack electronics at Ferrari, sat down with ZDNet Germany to explain how the high-profile team gets an edge at the track.

How much standard software do you use? In comparison, how much of the software that you use is developed in-house?
Depends which applications you are referring to. Office products are standard Microsoft. Design and calculation tools (CAD and CAM) are standard workstation products. All car-control and most of the data-analysis applications are written in-house, as are specific database products. It is not easy to give specific numbers because different groups use different tools.

What hardware do you use?
PCs for the majority of work. Workstations for design and component simulations.

What system software do you use?
PCs use Microsoft [Windows] XP companywide.

What development software do you use?
Again, it depends. Control software for the car is C and assembler. Analysis and simulation is mainly done with Matlab. Other projects are written in object-oriented languages (C++, Delphi and Visual Basic, depending on preference).

During a race, how many gigabytes of data flow between the car and the "box"?
Roughly 1GB for an average race that lasts about one and a half hours, and obviously more data for longer races. This is partially redundant data because we use alternative telemetry channels and internal memory to make sure we don't lose data.

How does the communication between the race car and the box work?
As the question implies, there is only communication from the car to the garage allowed. The onboard control unit, in addition to performing a controlling role, samples all relevant signals (sensors, actuators and internal status variables) and hands them over to a dedicated logging controller. This unit both stores data in memory and prepares data for telemetry transmission.

For telemetry, the data is encoded and packaged and then transmitted to the box in the microwave frequency band of around 1.5GHz. In the box, the necessary error correction to the data is applied, then the data is decoded and finally the data is distributed to the real-time workstations that are online. As there is no channel to the car, big gaps in reception have to be compensated by defining areas of the circuit where the telemetry data is buffered rather than immediately transmitted. The data is sent later to the garage for analysis.

What data protocols are employed?
The applied protocols are a combination of standard error-correction protocols and encryption, suitable for the type of error rate on a wireless link. They are specifically optimised for our type of communication.

Can a Ferrari system "crash"?
Yes, temporarily at least. As the function of the car fully depends on software, this software can go wrong. A crash will be only temporarily because as usual in embedded software, there are plenty of watchdogs in the control software that will reboot the controller after a short period of inactivity. Therefore the question should be: can a Ferrari have a short software hiccup? And the answer is yes. But as all software is extremely well tested in the lab, at test benches and during tests before taking it to the race, don't expect to see this at a race weekend.

How big a role does IT play in Schumacher's clear advantage?
This depends on whom you talk to. For someone like me who works with control software and data analysis, I would say that software functionality has a big supporting role for Schumacher. Good software enables a good driver to gain a final edge of performance that is necessary to outperform the competition. However, it is important to note that it is not just Michael Schumacher who plays a role in a Ferrari win. All of our drivers -- Michael's teammate Rubens Barrichello and our test driver, Luca Badoer -- contribute to our present advantage.

What is the ratio of computer simulations compared with testing that still needs to be done in real life, out on the track?
It is difficult to say. As far as functionality is concerned, we increase our simulation efforts day by day because it is a lot cheaper than building parts or writing control software and trying it on the car. Obviously, there are areas like reliability and interaction with other functions where there is no way to bypass the real thing -- which is testing it on the car. But while the contribution of simulation is still increasing, we are not reducing our normal testing at all. We just do much more.

How long does a simulation take?
This also depends. If we take simulation of changes to control strategies (traction control, for example), these simulations are done nearly online between two runs of the cars -- i.e., in a matter of seconds. Other simulations (like aerodynamic simulations) can take hours and are normally prepared before a race weekend, and we get the results as prepared parameter lists. Structural and kinematical simulations can even take days, but are part of another production cycle.

Using more powerful systems...will reduce our simulation times. At a race weekend, this means we can do simulation between runs that [previously] had to be done at night after the race or back in Italy before the next event. The more powerful our computers get, the more simulation will become a real-time tool and the more prepared our cars will be when they leave the garage for the next run.

How do you ensure that McLaren-Mercedes or other racing teams cannot access the data?
For our company network we run the highest degree of protection, both at the factory and the track. We are not using wireless networking at the circuit and at the factory yet because we are not satisfied with the security. All our laptops that leave the factory with critical data have encrypted hard disks. Our telemetry traffic from the car to the garage is encrypted as well.

How much influence does IT have in the development of a new race car?
Well, as mentioned before, all design work is done using CAD and CAM tools. The structural and kinematical analysis is entirely computer-based, so I would assume that there can't be any bigger influence than this.

What influence does computer technology have on the choice of race strategy?
Computer technology has a huge influence. All our strategy decisions are based on computer simulations of the event. We have parameters as a result of pre-event simulations that decide our basic strategy decisions, obviously together with the Friday/Saturday performance. This race strategy is then updated in real time with the progress of the race and new options derived. Clearly not everything can be done by strategy software, but at least this software can give a good base for the spontaneous decisions that have to be made by race engineers.

How do you make sure that IT doesn't pose a problem during the race?
The keyword for us is redundancy and fallback solutions. We practice a fallback of our operation down to the level of data transport between machines using USB sticks, just to be prepared for the worst. The other point is one of the main principles of [Formula 1] operation: never make the same mistake twice. We therefore respond to all problems with a fix, either by improving components or the structure or by stepping back to a less performance-focused but safe scenario.

Have there been mishaps that were caused by IT?
No, not yet. Clearly the level of excitement increases if we have to switch to back-up solutions, but right now we are able to handle all foreseeable situations.

What role does IT play in the car itself?
The car is controlled by software. We have onboard systems control wherever the regulations allow us to intervene and, as important, wherever we find a benefit. Engine control is the obvious example. Gearbox control to allow shift times on the scale of 10 milliseconds is another example. Then there is the area of traction control, to name another important one. Drive-by-wire (throttle control) and clutch-by-wire are other important controls. And there is much more, such as driver information system and driver interface, plus the before-mentioned telemetry and data logging. ASR (acceleration slip regulation) is banned by regulations, as are fully automatic shifts and launch control.

Is more speed possible with more sophisticated electronics?
We would prefer to speak in lap time rather than speed because the highest speed doesn't mean the fastest lap. However, the simple answer is: yes. Take the control software mentioned above and you understand what central role software functionality plays concerning the car performance.

How have the employed IT systems changed within the last few years?
They have changed from auxiliary tools to essential parts of the operation. Just as an example: Some years ago the car would have been sent out without data logging working if there was a problem. Today the car will not leave the garage if data logging is not OK. There is no point in running the car without the IT-based tools operation operating at an optimum level.

Editorial standards