Ericsson CTO says industry must come together to define 5G

Ericsson's CTO Ulf Ewaldsson tells TMN how the industry must define 5G, how operators can differentiate through the network, the challenges and opportunities of NFV for Ericsson, how vendors can innovate for success in the ICT market, and why the modern network must be structured like an orchestra.

1. On what 5G means, and how to get there.

“Everyone needs to come together and say, ‘What is 5G and why do we need it?'”

We are normally the sort of player that talks about the industry, not so much that we have developed a high frequency radio that can transmit a lot of Bits. We have done that, but it’s not a press release for us.

Most important for us is to come out strongly and say we are in the definition phase for 5G. There’s no use to come out and talk about individual radios and so on, that is not the point. 5G needs everyone to come together and say “What is 5G and why do we need it?” I mean operators have to be convinced that this makes sense, so we say that Ericsson’s view is that 5G is what is needed to build a networked society with all the sort of super high-requirment apps and things we believe the networked society will have.

For a while everyone has been hyping all the things that can be done: everyone exceeding everyone else with visions of remote medical operations, and cars and traffic management and smart cities and whatever.

All these kinds of ideas are really good but we are coming into the implementation phase of how do we do all this: build a network to enable this. Yes the first things are radio coverage and bandwidth but the extra things are the protocols and security of those networks, how safe are these things, can we trust the networks and so on.

This is a licensed network industry; frequency licenses have been given to the industry by nations, and originally they were there for voice and we built a grid for voice. Now, because we are seeing 100% year on year growth of data in the network, regulators are much more in a role to connect industry with users rather than regulate to keep costs down on voice services. And as that is happening we need to see networks as facilitators and enablers of a new society. So if we are going to build sustainable, green cities then the network needs to be extremely safe, very trusted, put the right Bits through at the right moment, and therefore we need 5G to be the evolution of technology to really make this work. And that will require new protocols, for instance, we can see we will need a lot of enhancements to were we are today with LTE technology.

“This is how our industry innovation is different to IT”
Being able to round the industry up around that is very important. This is how our industry innovation is different to IT where it is more proprietary per company. Through these projects we work on research, take patents that we license to each other using FRAND and eventually everybody can build tech that fits together and is inter-operable from day one – in order to build really relevant networks for the world.

(That has been the approach for network interoperability, but it kills services innovation – TMN)

No, for services it doesn’t work. Service innovation has to happen on top of networks. In Ericsson we are really rounding up the role to make networks more relevant; if we don’t many of these services will never happen.

2. How many network providers can make a business out of enabling this new networked society?

We see already that most working markets have 3-4 operators today. There are markets with more, Hong Kong for example. In Europe we are a little bit behind, waiting for consolidation to be stronger in Europe. That’s a general view of the market – not an Ericsson view. Over-competition is partly also preventing in Europe the introduction of new business models in terms of being able to charge more for data, less for voice. We’re stagnating a little bit in some, I stress only in some, of the European markets. Some of them are evolving.

It’s important to understand this is a data centric business and the busines models have to reflect that. Operators need to open up and let innovation come from cloud players, from the cloud. Take the connected car, all the stuff that will benefit from a connection in the car, engine management, navigation, entertainment, eventually remote controlling the entire vehicle, all those things will not come from operators, it will be the car manufacturer that has the cloud.

So each industry is here to take the benefit of these kinds of technologies, and our industry will not have a clue about what they will be doing but they will be doing it over our network.

3. So how will those network providers differentiate themselves, as enablers of cloud-based innovation?

We are seeing operators talking about differentiating by providing better performance. For example Telstra, they are marketing themselves in terms of performance. The reason for that is that this is now about data, and so we have gone back to talking about the network. Even now over 40% of all traffic on networks globally is video, and our prediction is that 60% of all traffic will be video within five years. Operators know that people using video already are comparing between operators and they are differentiating through that.

If you go back a couple of years ago, no-one was differentiating on performance, everyone was talking about their services and so on. For Ericsson that is very good news because what we do is build networks.

4. Talking about this networked society – 5G if you like – world, what levers can the operators look to pull to achieve that network differentiation?

That’s where we are coming to and we can really see the difference between operators. They [operators] are getting more and more different. Some operators are getting to those discussions, what the OSS/BSS systems can do, how they will evolve them to support that, how that links to SDN technology, reprogramming of the network, service chaining, how video services can be provided better, the service chaining discussion talking about caching into the network where it is controlled from a central point of view on service, subscription, user… That’s the dialogue we’re having with a handful of operators and they are reaping the benefits. But it’s not all operator conversations we have, there are many that are not.

We are virtualising several of our nodes in the core network next year. We are already doing customer trials on test software.

5. For those that are having those discussions, how is Ericsson enabling that?

First, it is very important for us to virtualise software, in order to get more out of it. We are the fifth largest software company in the world, we spend around $5 billion on R&D every year, about 14% of our turnover, we need to have absolute enormous scale.

Let’s take caching: if we develop a very efficient caching software (which we have done) we want to be able to run that on different platforms in the market, and to be able to transport different software pieces between the different elements in the network. It could run on a node in the core network or remotely in a base station environment, so we chose two years ago to go OpenStack in our virtualisation engine, and we are virtualising several of our nodes in the core network next year. We are already doing customer trials on test software.

At the same time we want to be able orchestrate that, involving our OSS/BSS systems for the orchestration of that, and eventually that will provide the basis for service provider SDN where we are able to control the whole network more effectively. We will use SDN and NFV to be able to provide a better end user experience.

6. I’m interested in this idea of orchestration. That could be taken to mean you have one element sitting over everything pulling the strings?

If you tried to build a signalling node in the network that will handle everything you are making it impossible to build that. What we are saying is that each node has to have some level of SDN in it, and signalling and management, and eventually you build it (the network) in a layered way.

So the SDN layer will be a evolution of the OSS to a very large degree that will have element management, and have real time analytics in place to get a lot of data, and that’s when you can start to talk about orchestration.

Orchestration is not about having one conductor in an orchestra, in fact in an orchestra each player is actually quite good at playing the violin by themselves. The guy standing up there provides that little bit extra that makes it really sound nice. So it’s a very good analogy because the conductor provides only that little bit that makes it sounds perfect.

(TMN interjects – In fact, if you extend the analogy, you’d note that orchestras actually have section leaders, a layer between the individual player and the conductor…)

…Yes, and the reason why it’s done like that in an orchestra is because the hierarchy has a meaning. So we don’t believe in total mesh – and believe me we’ve done a lot of research in mesh networks over the years, where everything talks to everything and so on. But the moment you introduce any sort of hierarchy into a mesh network you get an enormous efficiency out of it. There will not be a galactic global node that controls everything. That’s possible in a data centre where you can buy in an IBM controller or whatever, but if you put that into a total mobile network it’s not a very efficient way of doing it.

I really like that orchestra analogy, by they way, I think it fits very well.

7. One thing that many wonder about is if the implication of NFV, with its goal of network capex and opex reduction, along with convergence with the IT world, looks like quite a big threat to a company like Ericsson.

When NFV was started they [the NFC Forum] started with a parameter to reduce capex with standardized hardware, and the term COTS, and then you ask who makes COTS and I never met a vendor who makes COTS! So there should be a standardised server, but we should also look at time to market, the possibility to reprogramme the entire network, and these are still there guiding the work.

When it comes to the hardware side, there are both camps. Some are saying the operators could run some parts on not so high performing hardware with some parts requiring more highly performing hardware, and I think we will see more differentiated hardware than just “COTS”. The type of network we talk about, 5G etc, that will require super high performing servers.

The ICT server industry will merge with the IT server industry, and I can’t say more than that. The IT vendors are increasing their performance dramatically to match up with the performance required in cloud, whereas before every enterprise bought their own server. At the same time the telco industry is saying now can we open up our servers. I think all types of co-ordination between the two will create and define future servers.

Over time, our opportunity to co-operate and work closer with the IT industry will make our challenges and opportunities bigger, as long as we make the networks more relevant I see this as a big opportunity,

There might be examples of operators that try to use this as a basis for negotiations on prices of servers and so on: that’s OK, that’s fine, we can adapt to that. No industry will ever be healthy from being protected.

8. As a company, you sit pretty much at the top of the food chain. What advice would you give other companies across the ecosystem as they themselves try to stay relevant to the relevant network?

I would give them the same vision of where things are going I’m giving you. Ericsson’s role is partly to stimulate others to invest. I would like to make sure our industry is attractive to start-ups and VC companies. We are so dependent on venture capital and start up companies looking for innovation to help our industry be more efficient. What stimulates this industry is companies coming in and doing something we haven’t thought about.

Eventually we will incorporate it, or buy them. We can provide 400 customers and market channels all over the world, and Ericsson can then provide tremendous opportunity for these companies to realise their dreams. Do you remember Optimi in Spain? That was a great company, great innovators, absolutely taking a lead in software optimisation of parameter setting in 2G and 3G networks to make sure that cell optimisation could be brought to a new level. We bought them as they were more innovative than we were, and now we’re deploying their software globally.

Another example is BelAir. They were doing great in Wifi, providing cable operators with carrier grade Wifi for example, and today we have globalised their stuff all over the world, and I think almost everyone stayed. We provide a vehicle for globalisation so my message is, we have the muscle to innovate around the big ideas in our industry, we hope for many others to innovate and come up with ideas that lead to all of our success.

9. And what are the key things for other companies to think about to stay successful, whether they are providing signalling nodes, backhaul equipment, SON software – whatever it is they are doing?

To be part of the industry in terms of sharing the idea of what 5G is and to help to support that. These companies should be as much as possible close to operators, then again I will also mention that some of these companies are also innovating things which are not so innovative, some start-ups are looking to attract VC rather than innovating. Like some of these signalling companies have been doing that, and it’s very easy for companies who build the equipment that we do to respond: we just change a few things in our signalling and all of a sudden their whole company goes away.

In that sense I would say that companies should really be innovative, really think about things that make enhancements to the next level, don’t differentiate so much on hype. One problem in our industry has always been over-hype.