Virtual Centres - Klaus Meier - Skyguide

Picture of Vincent Lambercy
Posted by Vincent Lambercy
Welcome to one more episode of Radar Contact. Today we'll discuss the topic of virtual centers and for that my guest is Klaus Meier, who is the CTO of Skyguide. Klaus, welcome to Radar Contact.

Thank you Vincent. Happy to be here.

Before we go in the virtual center topics, can you please introduce yourself and your career path? Because one thing that I find is interesting is that you were not in ATM your whole career. You had a whole life before joining Skyguide. So can you just tell us a bit more about that and the reasons that brought you to Skyguide?

That's for sure. So I'm an electrical engineer by training at a PhD in Microwave theory. Founded a company and during the PhD and then also helped, basically consulted at that time nascent mobile communication industry in antenna design. However, there was a point where I always had a bug for aviation and during that period having my own company, I decided I go into the airline industry or the aviation industry and that's how I joined Swissair. And I was then at Swissair initially in the technology environment IT and then moved more and more into more process improvement, process optimization and also applying new technologies and particularly and that period the internet technologies in the aviation industry. After Swissair, I then moved to Schindler. Schindler is a global company producing elevators, operating them, maintaining them. And that was a very interesting period because in this company it was all about how data is transforming the business model of a global operating company like Schindler.
And had then the responsibilities for the Americas North and South America. And then in 2014 I got a call from a head hunter and he said Skyguide is looking for someone because they have this idea of a virtual center. And when I heard that I thought that's an opportunity to get back into aviation because my heart is still in aviation. And this idea of a virtual center of being able to change the paradigm in air traffic management sounded really attractive enough. And I thought Switzerland is small, is nimble, is entrepreneurial, has access to good people, good universities and has an efficient bureaucracy. So probably that's where we could try something out like a virtual center. And that's what I did. And I'm here for eight years and I'm happy to be here and has been a heck of a journey in implementing and deploying a virtual center in a country.

Yeah, I can imagine. And it's crazy that it's eight years already. Now, can we start from a very basic and naive perspective? Can you explain briefly what is a virtual center and how different it is from what we have today with sometimes you have the data center in the ground floor, the operational room on top. What will be different or what is different with the virtual center?

The difference of a virtual center is the fact that we believe air traffic management can be done location independent. As you said today, air traffic management is still very much location dependent. Originally, historically you had equipment like a radio with a certain range and with a radio you could talk to a pilot, but very location dependent because a radio had a certain range. The same was true then with electromagnetic means to help an aircraft to land when the site was not so good. Today's ILS, again, you installed as at an airport and an aircraft could use this technology to land at an airport. And the same is also true today. When you talk on area control center, as you said, you have an area control center where the controllers are managing an airspace above them, specific sectors even. And the data center is in the basement.
And the sensors like surveillance navigation or also the combination tools are connected in a star shape to these area control centers. In the virtual center, the systems can be any place, it can be, it doesn't have to be in the basement anymore using modern technology could be even distributed, could be bought as a service or used as a service. And also the radio for example is not connected by a fixed line but by a network, by an IP address. So you could address any radio station in your region or even across Europe. And the same is true also from an operation perspective. You're not tied to a sector which is from a geography perspective above you, but you can access any sector anywhere. And that's the fundamental change that virtual center bring, that you are not location dependent anymore. And this brings the advantages that you can optimize the capacity across various area control centers. You also have resilience because if you have a problem one center, another one can take over. And the same is true on the system side too.

Just to clarify one thing, you speak about location independence. We are not speaking about sending air traffic controllers in home office right?

No. We are not sending them to home office, but I still believe it is a big, big step from an air traffic controller perspective that you won't have a geographic license anymore but you will have a system license. And I think that is a big, big change to what we have done the last 60 or 80 years in air traffic management where we are still operating very much according to geography. And I think in this respect it is a big step but now they won't work from home and they don't work. They still come to a center because we do believe it is important to come together and work together and have a kind of professional setup because at the end we are in the safety business.

For sure. Now, so this is the theory, this is the concept. Where do you stand at Skyguide with your virtual center? Is it a reality already? Do you swap sectors between Genevan Zurich or is it more organization or is it a mix of both?

Well we set up the program which was originally planned to be a 10 year program from 2014 until 2024 that we set it up in three distinct stages and the first stage was called "setting the scene" .There we didn't change anything on the core systems where we built an overlay for Geneva and Zurich. So at least we have for the first time one solution for both area control centers and create benefits for the controllers because it was very important for us to have the controllers in the boat in this journey because at the end they're at the core of what we are doing and we want to make sure that they understand the value on the benefits that come with the virtual center concept. And we introduced stripless with that. We introduced data links into the aircrafts and we were very successful with this first period which we called "building the foundation".
And we also increased efficiency such that during this growth period we could refrain of hiring 30 additional controllers but could use this new technology in managing the airspace. So we had already quite some tailwind because the benefits were very clear. But then came the second phase and the second phase sort of second stage there we said we call it "building the foundation" and that's when we tackled the technology behind the virtual center and we moved from monolithic, from two monolithic systems in Geneva and Zurich to a service-oriented architecture where we basically changed the architecture and introduced services as a way of how to provide functionality to an air traffic controller. This went live in 2018 and since then we are running on this new platform and are now in the third phase which we call "new operations". And that's tackling now the topic as you said, coming up with an operations concept so that a controller in Geneva would be capable of managing an airspace over Zurich and a controller in Zurich would be able to manage an airspace over Geneva for example, so that we become location independent also on the operational side. So to summarize three phases, we finished two phases within the last phase now and since 2018 the new platform is up and running and since then we are moving functionality step by step from the legacy systems onto this new platform.

What happened in the last two or three years obviously was the Covid crisis. Did you in some way could benefit from it because we had less traffic? Was that a chance for you to speed up your program or was that an additional challenge for your program?

Well we always see crisis as an opportunity and definitely we saw Covid as an opportunity to continue on that journey and probably even focus more on the delivery of the virtual center. At the same time it was also challenge because we deployed functionality which then went into operation under not full load conditions and that became afterwards a challenge because certain things were not really tested or were not really operating in an early stage under a full capacity situation. So it had both sides. Covid had a positive impact by being able to focus more to clean-up the portfolio and focus on delivery of the virtual center. Whereas at the same time it was an exceptional situation when it comes to operations, which then caused challenges for us because it was not the normal way of managing an airspace.

I can imagine it was a very special time for all companies in the area. Now touching back on location independence, obviously Skyguide manages the Swiss space, a bit of French one. Will all your service and all your systems still operate on-premises within Switzerland or do you envisage of buying some services from the outside and from other countries possibly?

So at the moment the systems are on-premise and the main reason for that is that we still have the legacy up and running too. As I mentioned earlier, we deployed the new platform in 2018 but we still had the two legacy systems running too and we are moving now since then functionality from the legacy system onto the new platform. And we are doing that because in air traffic management you can't do a big bang, you cannot just have the old systems running, then in parallel develop a new system and at one day when you're ready you push a button and you switch over, the risk would be much too high. So that's why we are having basically the old system and a new system still running until we have moved the whole functionality of both legacy systems onto the new platform. And until then we cannot go into the cloud because the legacy systems, because of latency and legacy challenges, they simply do not run through a network.
So that's why they still on premise maybe the less we are using external partners managing it for us it is managed like a service. Now even if the hardware or the equipment is still locally here in our premises in Zurich. Now this is just during that transition period, once we can switch off the legacy and once there is a service available that I could buy an ATM system service from someone else, we will go that way. And this is the plan that we had from the very beginning and we still have that. We say by the time we need to replace then our FDP in here in Switzerland, we will not go for a monolithic system but we will go for a service and we expect that to be sourced from either a big supplier as a service or even from another ANSP who has the same challenge or the same solution. And this is part of that concept that we say for example an FDP in the future should be available as a service and we are buying it and the time horizon for us will be between 28 and 2030 again depending that someone is out there who provides that service to us as a service.

We are the previous episode this year with Anna von Groote from EUROCAE about standards for virtual centers because if you look at the world of standards, let's say in the last years you have ASTERIX for the data coming in, that's pretty easy but it's one way. But there are not so many standards about how to couple any HMI to any flight plan servers for example. And I think personally the standards will play an important role there and I don't know what role Skyguide plays being one of the pioneers of standardization and virtual centers because when you say you want to buy an FDP service from the outside, you need to have a standard way to integrate that with your systems, don't you?

Yes. And Skyguide as an early move or has been involved in the standardization process from the very beginning in particular the EUROCAE. So as you mentioned before, and I think we have been also instrumental in discussing the standards and driving towards standards because at the same time we are probably the ones who have most experience of what it means to working with a service-oriented architecture to work with services and also to move to open architecture. I think the main drive at the moment is that we need to stop developing locked monolithic systems based on COBRA, an architecture that dates back to the time when I was at university and this was more than 30 years ago and this is what we are currently rolling out and now we are talking about how can we make those system interoperable. And I think it's a complete joke because you create first software that is completely locked and then you say, okay, can I develop a box which this box is connected to another box and there we translate the information of how we want to exchange information.
If you go to an open architecture then intrinsically you use already industry standards in the way of how you share data. And it's the same what we did for the virtual center here in Switzerland. We basically took the SWIM as foundation and we also developed the software based on SWIM such that also the master data, in this case a flight object is following a SWIM standard. Now the question is what profiles and so forth, how deep you want to go there. And I strongly believe if you go to open architecture you will use industry standards which make it already from the onset such that it's built for distribution, distributing information and data and integrating data. And then you have a few standards that still need to be defined because they're connected to aviation or they're connected to safety and they need to be defined and also on a level, let's say on a European level or maybe on a global level, but it's much less of a standard discussion than on a fundamental architecture discussion. And I think on the EUROCAE side, I think they have done a great job and we will continue to support EUROCAE in coming up with reasonable and feasible standards such that an open location independent system is a viable solution

That will be really nice to see and we'll see what the future has for us there. Now looking back, you are, let's say two-thirds through your journey roughly when you look back, what's are the biggest lessons learned for you in term of implementing a virtual center?

Legacy is a drag,
So legacy is a drag and you were laughing and I think that is for me again and again surprising that it takes much longer to switch off old systems. And the main reason is that they have been growing over 20 to 30 years and they're very much also connected to the people who have built the systems and to now come with new concepts and as we call it a paradigm shift also in the architecture. It's a challenge not only from a technology perspective but also from a mindset but also skill perspective. And this is what we must not underestimate in a kind of an effort it takes to really introduce new technologies but also introduce new processes, new roles and new skills and still be safe because again, we are in the safety business here and we have to be very careful in taking risks and where we take risks.
And in this case the legacy has really been for us the biggest challenge. We wanted to switch off the Geneva system in 2020, actually end of life would've been in 2015, but according to the program we wanted to switch it off in 2020 and we will still run it probably into 2024 because it's just full of surprises. More of the functionality you migrate, you find new surprises and you need to implement that in a new system and it just takes time. So I would say probably the biggest challenge, and this is then also connected with mindset question too to say we are air traffic management and we have always done everything on our own and we know best. And that is in many aspects true, but in other areas it is not true anymore. There are better experts out there for technologies with whom we can partner and work together much better, much more professional and much more effectively and efficiently.
The good experience, the thing that went better than we originally had at anticipated is the collaboration with the regulator because any step we take needs to also be safety assessed and then also be checked against existing regulation also with the local regulator. And we have a very good relationship here in Switzerland with our regulator where we understand that in innovation we both are going through a learning curve. So the innovator in this case Skyguide and the oversight organization in this case the regulator, we both need to go through a learning curve and we both have a role to play in order to be safe and to respect each other's role. We are the ones innovate Skyguide and the regulator is the one who needs to make sure that what we do is safe or has we have the means in place to check whether we comply with safety regulations is very, very important. And I think that was possible because it was a relationship of mutual respect and trust and in this respect we could really move forward. And a lot of experts told me in the beginning of that journey, it'll be very difficult and impossible. And I said no, my experience is it is possible as long as you respect each other's role.

If you don't mind, I'd like to digress a bit compared to what we planned because you are speaking about innovation and things like that and one trend that we see in ATM and aviation in general is that it becomes harder and harder to attract new talents. And it's also possibly because of what you said before about all your technologies. If you go to someone out of university and say, Hey look, we do a traffic management, it's an interesting problem, we have challenges and we solve that with technologies for the last 25 years, it it's not so attractive to them. I think because people now want to work with current technologies, but we're still in the safety business. I don't know if it's a challenge that you face or are you still recruiting quite easily?

In the ATM system side, we are recruiting, we can still recruit well because we have a good story to tell. But the problem that you outline is obviously there. If we would tell the new recruits that we develop our virtual center on CORBA could not hire anybody because again, the CORBA people are my age. That's what we learned at university and those people will not join. That's was very important that we went to service oriented architecture with a classic middleware, but even that is already dated. If you go to universities where with data talk about open architecture, they talk about microservices, containerized services, Kubernetes, Kafka, so all the cool stuff out there. So it's a very obvious point that the current model of how our suppliers develop software is not sustainable anymore. And we see also now with the suppliers changes that they have shifted and they move towards open architecture. So path at this Airspace World last month in Geneva where Thalès openly announced that they're going towards open architecture and I think for me this is just a very natural response to the situation that the old technology simply not sustainable anymore. And one of the big challenges is you can't recruit people if you don't offer them tools and solutions that they learn at university and then they want to apply in practical life.

Yeah, that's very true indeed. And it's important I think to offer people a full career, not just something for a few years. Now looking back at something a bit maybe not so easy for you, last summer actually before the last show we had in Madrid, Skyguide that some technical issues and the airspace was closed for a few hours, which is not something very common in ATC. Is it something that was related to the virtual center and is it something that you could have avoided if you were the full virtual center implemented already?

Well, a "clear the sky" could be avoided if virtual centers would be operational because a neighboring ANSP can take over sectors immediately. And I think that is the beauty of the virtual center concept that that's also why Europe adopted that in the European airspace architecture study and then in the current master plan and also in the updated master plan because instead of building resilience for 68 data centers, which we have today, which is prohibitively expensive, another center can take over the service because you are location independent and it doesn't depend anymore whether you are in Gene or in Zurich or in Reims or in Karlsruhe or in Helsinki. So yeah theoretically the virtual center would helped, but the problem is we are the only ones at the moment applying this technology. So until someone else has a virtual center, we need to be able to intercept cases like that here in Switzerland.
And that's what we are doing at the moment. The problem we had in the summer was a network component that malfunctions the resilience that was built in was not triggered. And in the current architecture we had there a gap that we fixed actually that very morning. But it was one of the reasons why we lost. We didn't lose the core systems, but we lost the monitoring capabilities of the systems, which in this respect was a reason for clearing the sky. But that was connected also to the fact that we running at the moment three systems at the same time. So we have a higher degree of complexity and it's now for us an opportunity to speed up the move towards the modern platform and switching off the legacy systems.

Thank you for open answer on that. I'd like to challenge your answer a bit because I see the point with location independence being independent of the airspace you control, which will need a shift on the operation side as well. But could you really imagine a way where Geneva or Zurich become available and if people from France, Germany Italy taking over immediately, would they have enough people on site ready to take over or could they take over within, I don't know, one or two hours the time for them to bring enough staff on site?

Well, if I say immediately, then this definitely means probably an hour or two to be able to adjust and communicate correctly. And again, we had an outage of five hours from three till eight o'clock in the morning and it's usually the response time. First you have a problem, you try to identify the problem then to respond. So there's quickly minutes passing by. But what we want to say here is the fact that just in the case of Switzerland, we have a small airspace and here it should be possible to quickly just extend a flight from, I dunno, from Marseille to Hamburg via Marseille and Karlsruhe or something like that. So it'll not be without any kind of procedures and processes that need to be changed and then put into effect and prepared accordingly. But it would not mean that an airspace gets blank, but today an airspace gets blank. If I have fire in an area control center, this area control center is gone and it's probably gone for weeks until you can run it again. And this is why you have us in the regulation a "clear the sky" concept in there. That's the only way of how you can respond to that. If you have location independent services, then we are not talking about weeks, we talk about probably hours.

Okay. Now to finish, I'd like to ask you our standard question, even if it's what we've been doing for the whole episode. What evolution do you see in ATM in five years, but also to open the door to more fantasies in 50 years from now?

Five years is good, 50 years is really an exciting time horizon. However, in aviation we are moving a little bit slower. So 50 years is maybe still a good one. In five years, I strongly believe that there is now a response on the system side that the current architecture is not sustainable anymore, as I said. And we will open up, we'll develop systems at least from a technology perspective on open architectures. So I think that cannot be prevented anymore. This is just going to happen. But this leads then to a more open virtual setup with more innovation. I don't know yet because we have also regulatory developments which go a little bit in the other direction. So in this respect, I'm here a little bit cautious and that's why I point to another area where we need improvement. And this is in the whole incentive schemes in Europe.
As long as we are incentivized to pump hundreds of millions into equipment and assets and big systems and not having an approach that supports the move towards services, we will not really change a lot. So the performance of European aviation and particular ANSPs need to be incentivized differently. And we do have some ideas that Skyguide, we developed the so-called straw man for the next reference period where we believe will be the right way forwards to set the incentives right, such that we move, that we use and exploit the technology also on the commercial or on the performance side correctly. And I think that's must happen the next five years and we will fight that this will happen in the next five years. Now in 50 years, I think you might have to change the name of your program. I don't think you can use radars anymore because I think radars will die out as a technology.
So I do believe the CNS will change dramatically. CNS will be based on two layers. One layer will be space-based. We have done already on the navigation site with the GPS, we follow on surveillance side with concepts like aireon and we will follow also on the comm side with payloads for communication. So we will have a space-based global service or regional service and not local services anymore. At the same time we will have terrestrial or on the ground services as a fallback, as an independent second layer, which again will most likely not be national, we'll also be regional or even bigger where we provide services as a counterbalance to space-based communication, navigation and surveillance side. I also believe that we will stop managing sectors. We will not have sectors anymore. We will manage flights through an airspace. As I said, this is initiated with this location independence and the flights will, I always distinguish two areas.
One is the end route, which will be a classic network problem, a deterministic network problem where you separate traffic in a network, not in a sector, not in a network node anymore, but in the whole network you optimize that and you can do that by applying systems and intelligence of a system. We are doing that in other areas already successfully. And the challenges will be weather and departure or approach to airports. And that's a flow management topic here. I strongly believe, and we see already indications in that direction that we will use much more real-time simulation to do flow optimization from the pushback until you join a network. And here too, it's going to use much more technology and intelligence in the system and definitely not a controller using an amplitude modulated VHF signal talking to a pilot. I think that will die out.
And I believe that virtual centers will be the normal mode of operations you'll op. We'll have systems, maybe two, three systems that talk to each other. And we will have a lot of innovation and competition on what we call the front end side. That's where the human machine interface, where we have decision support services, where we use more intelligence in separating traffic. And the last point, I think one of the big disruptors and the push will come from players outside of this industry and they come mainly from the unmanned world because the airspace will not be reserved for manned aviation. It'll be a mixed network of unmanned and manned aviation and they will operate in the same airspace and this will be safe and will be highly efficient. And with new players coming in this industry, they will also put some pressure on the current setup. That's my outlook. And for the controllers, no more geography going to be systems and it's going to be still an exciting job, but really using the human being there where the human being can really be better than the machine.

Klaus, I have the feeling we could do a series of episode with you on all these topics, but thank you very much for your time today for sharing your vision and your insights on virtual centers. It was really a pleasure to have you, so thank you for that.

Thank you, Vincent. My pleasure, and always