Keith Dyer spoke to Yves Bellego, Director of Technical Strategy, Orange Group.
Bellego was certainly interested in topics as small cells, NFV and LTE-A feature sets – but his emphasis was on ensuring any new technology introduction is truly tested for reliability and scalability.
A key phrase he repeated through the interview was “step by step”. His network vision was also one of building a network from large to small, from outside to inside. He wants population coverage first, at best cost, and only then to move to small cells and indoor solutions for extra capacity.
He sees true potential in virtualisation, but he’s not about to rip up anything, or go all-in on a telco in the cloud approach. He’s looking to test a limited set of apps on one hardware platform, then a few more apps from different suppliers on the same hardware. Same principle with SDN – start in a “closed” data centre environment then move into the “WAN”.
Network renewal, LTE-A, SON, TDD
We’re spending roughly the same every year on the radio network globally. We will need to increase capacity on LTE, and enhance it with LTE-A.
We introduce new features every year, and then every 6-8 years we need to replace the equipment.
Network transmission is still quite a huge amount of money to invest and that will also continue.
When it comes to network control we will have to invest but as time goes on it will become closer to the service platforms, to the IT. On one side we invest in the hardware infrastructure and on top of that in different software services – and network control will be one of them.
In Europe, 2011-2013 was about RAN renewal, we replaced all of our equipment with multi-mode multi-standard equipment, a huge programme that is finished now in many countries, across tens of thousands of sites.
That renewal enabled us to launch LTE in 2013, and in 2014 we are continuing our LTE rollout, because what we want now is to have full country coverage.
We have prepared LTE-A. The first feature is multi-band Carrier Aggregation, where the use cases we see are 20 MHz at 2.6 GHz plus 10 Mhz at 800 MHz, and 20 MHz at 2.6 plus 20 MHz at 1800 MHz. Carrier Aggregation is not a big issue for us technically, we can launch it where we have device support.
Then COMP, eICIC, things like that, will come after.
SON – moving from in-vendor to cross-vendor
We have trialled SON for several years using SON on the elements from the equipment providers. Now we still have to work out how to make use of SON at large scale using all SON features. One SON feature that works quite easy is ANR, others are more complex.
What we have done until now is use SON features inside the elements, from the RAN suppliers, basically radio access SON. The idea to have more global SON from 3rd parties on top is something we are testing but we have not come to an answer yet.
First step, let’s do what’s kind of easy and then we will build. For me the difficulty is to go from the concept to what we can really do on large networks with a large number of customers. We can already do today very specific actions for specific customers, but how to do that for millions of customers is the difficult question.
Also SON changes processes for our operations. In the past we had specific tools to work out the design of network, put that to the network and the network itself did not change it. Now the network itself will change some parameters and settings… they (operations teams) have to understand why and consider these are the right settings.
In 3G TDD was far from the FDD technology so there was no scale benefit. LTE TDD technology is closer to FDD so at least we can expect to have networks and devices available.
Having TDD spectrum we would like to use it, so we need to have chips and devices to make use of it, basically that’s it.
First let’s make TDD nice, make TDD work in large scale and then to it aggregate with FDD? We’ll have to see, personally I have not looked at it in detail but it is about integration to the device. There are many things we can do on the network side but the device is very challenging in terms of integration and power consumption.
REFARMING FOR LTE
We discovered that maybe eight years ago when we pushed a lot for 3G at 900, in discussion with Qualcomm and the others they said the chipset is not a big issue but the antennas and so on need to be integrated – and at the end UMTS 900 was not that easy on the devices.
At the end, the idea is to have LTE in all bands, the big question is when and how to reduce 2G and 3G traffic.
I expect small cells not to be a big issue for 2014 or even 2015
One thing we are interested in but still have to work out is small cells. On the equipment side the vendors have done quite a good job designing and manufacturing really nice equipment, now what has not been solved is how to use it for an operator at large scale.
I expect small cells not to be a big issue for 2014 or even 2015 because first we have to deploy LTE global coverage, and I expect our focus on small cells to come back when we have good LTE coverage and capacity. The main issue is not about deploying in shared or unshared spectrum and so on, it is asking what is the need and business case to deploy small cells – including the backhaul – making sure we put them in the right place and we get the traffic [analysis] right.
Site sharing for small cells? We already share quite a lot of our sites, for small cells again it is yet to be addressed but it is just one item among others.
Small cells for indoor? Again the first thing we need to provide is overall coverage and capacity, upgrade all the antennas on rooftops. For indoor on 2G/3G we basically have good coverage today: indoor for 4G? First let’s do good indoor coverage from outside and then move indoor.
The cheapest coverage option at 800-900 bands provide large coverage so let’s do that and then only after do indoor sites. In 2014 we will do some LTE indoor but it will remain quite small in terms of capex.
Cloud-RAN? We already used the base station to separate the baseband and radio head, moving that into indoor that is very nice but let’s first have 99% population coverage.
WIFI AND UNLICENSED SPECTRUM
Some years ago there was a philosophical battle between licensed and WiFi that at least is now over. The way we see it is we need to make best use of all the spectrum resources we have. Licensed spectrum and unlicensed have constraint and benefits. To provide best service to the customer, to keep it easy to the customer, we need features in the network so devices and networks together automatically decide which network to use.
Orange has the Orange SSID so any Orange customer can go onto the Live boxes of the others. So that is one aspect of sharing WiFi. I believe we also need authentication, ANDSF, Passpoint, things like that. The question is not so much in the network, we need that natively on the device. Having that feature native to the device is quite an important step because this really is a booster for that option.
In 2014 we are testing for VoLTE. We have tested some VoLTE pieces already in places, and have IMS in many countries used for fixed VoIP , enterprise and some residential services, also for RCS. What we need now is to prepare for the introduction of VoLTE to the commercial [cellular] network. One thing that is complex to set up and difficult to finalise is SR-VCC to handover to 2G and 3G. This is becoming more and more one project as any other but it is still to be done. There are issues at the handset and on the network to be tested. It is just a software upgrade but there are a lot of elements, radio, EPC, HLR, IMS, to upgrade and test.
NFV AND CLOUD
One good piece of news in terms of this year’s Congress that virtualisation is progressing faster than I had expected. So the way we see it is that we expect to go into virtualisation step by step.
So we do not expect a big bang, but first to test one hardware and several apps from the same supplier as a safe, closed environment. Let’s do that first, put it into the network, run for some time. We will have some good lessons because operating these types of solutions is not that easy.
For years we have a team for IT and one for network equipment and with NFV all that has to be shuffled. For years they were really separate and now they all know they will work together and that’s a big change of mindset. So we need a learning phase. Then after that first phase then we have several applications from several suppliers on the same hardware, and then we adapt.
SDN? Again we go step by step, in our mind one part is still what we call research , typically on the WAN network side. More short term I see SDN inside the data centre, in a closed world. It’s easier, let’s start with that and then after some time it will expand.
On virtualisation the great benefits we see are in fact maybe not that much the capex, but the scalability is one aspect we really expect to benefit from. And then we have to work out the operational aspect, but there should also be some opex synergies. It should be easier to operate because we are separating hardware and software.
One good news is that there is not so much buzz about 5G. I was a little bit afraid, that there might be some…we believe what’s important is that the target of 5G is driven by customer needs and operator needs. Let’s put it the right way and let’s not say we have one piece of new technology and let’s use it and we don’t know why.
Something like mobile broadcast… we have been working on mobile broadcast for years and years and years: finally it might come. The technology worked six years ago but there was no market for it. We’ve had point to point [device to device] and mesh networks for decades -a lot of these technologies, they are quite good but it’s maybe not the right time.
There are a lot technology innovations, we have to define which ones develop the market.