At the forefront of the telecoms industry’s virtualisation revolution are network functions virtualisation (NFV) and software defined networks (SDN), and it is expected that deployment of these technologies over the next five to ten years will enable operators to move to a more sustainable cost base.
‘In five years, the industry will probably be unrecognisable, but I can’t say what it will look like,’ says Philip Bridge, the senior network architect at EE. ‘I liken the situation to the introduction of IP in telecoms and it could possibly be more transformative than that.’
NFV is likely to be deployed first and will enable operators to decouple the network function from specific hardware. The move to more commoditised, server-like hardware that is configured and managed by software to create network functions and capability according to user demands will save costs and enable maximised utilisation of network equipment.
However, virtualisation technologies will not be deployed overnight and there are challenges in making a short-term business case to operators, when the main benefits will only accrue in the longer term.
‘Virtualisation in general is a Virtualisation opportunities are real and transformative for operators really big trend, and in mobile the business case is easier to make because with new network rollout the money has to be spent anyway,’ says Dan Warren, the technical director of the GSMA. ‘When you consider that the capex saving associated with NFV can potentially justify the business case in itself, the opex gain is a bonus.’
The cost efficiency point is the key driver. ‘A lot of trends in networking are towards virtualisation with NFV and SDN but also service virtualisation,’ says Eric Carr, the senior vice president of engineering at Guavus. ‘Operators want to reduce their cost structures and get off traditional architectures, and virtualisation is definitely the number one driver of how to squeeze the vendors on price.’
Bridge at EE says figures of 30% in cost savings are being spoken of in relation to virtualisation. ‘That’s interesting,’ he says, ‘but it won’t get operators out of the hole they’re in where demand doubles every six months or so.’
In the meantime, many virtualisation principles are starting to creep into the mobile operator market. Among these is the concept of moving radio access equipment to more centralised locations. This will see just the antenna plus a power supply and fibre link located on buildings, with network equipment located in a centralised location – potentially a street-level cabinet or in an architecture that supports more antennas such as a specific operator-owned building.
The benefits are significant in terms of maintenance, rollout and real estate cost savings, but there are potential issues around latency in this type of front-haul arrangement.
‘The need for equipment to be closely located to the antenna has lessened in the past two years,’ says Warren. ‘It’s enabled initially by the deployment of fibre to the antenna, not electronics, which means you can deploy cells with very high bandwidth, which means delay is less and it’s easier to do over long distances if you have fibre in place.’
However, there are risks to that, Warren points out. ‘Suddenly you’ve increased delay by whatever the distance is, and the speed of light becomes a limiting factor,’ he adds. ‘Operators are going to have to be very careful watching how that one millisecond delay plays out because it potentially causes a problem if you over-consolidate to centralised networks.’
Others are less convinced that the benefits are truly significant. ‘It’s an interesting concept but I’m a sceptic,’ says John Naylon, the chief technology officer of CBNL. ‘It works well in some scenarios such as stadiums but it’s not clear why an operator with an established network would choose to do it. I see this as a research area; the gains are real but not significant and they can come at quite a substantial cost.’
Naylon backs his argument with recent Ericsson research that has found the cost of hardware is only 20% of the total cost of ownership of a mobile network. ‘There’s a really rich field for research to say the centralised radio access network (C-RAN) is nice, but the challenge is still how an operator would get a performance gain from having a distributed RAN,’ he adds.
‘Antennas are the size and shape they are because that’s what physics dictates, and it’s cheaper to co-locate the electronics with antennas because of Moore’s Law. It might appear more flexible but if the electronics become two-times cheaper, two times as fast and half the size every year, what’s the point?’
At first glance there may seem to be little connection between virtualising access networks in dense urban locations and using microwave technologies in rural areas to address not-spots.
However, both areas present examples of how networks are becoming increasingly virtualised as operators turn to a far wider range of technologies than ever before in order to meet their customers’ needs.
UK operator EE will be making voice services – as well as 3G and 4G mobile data coverage – available in communities that don’t have reliable mobile or high-speed broadband. These areas have remained unconnected by traditional approaches to network deployment that have relied on building large masts.
To cover these communities, EE will build new micro networks that wirelessly connect small mobile antennas to a suitable nearby macro site, without the need for cabling, dramatically improving the economics of connecting hard-to-reach areas.
‘These situations are the same but different,’ says Warren. ‘The whole micro network phenomenon is actually virtualised in the core. When your core is no longer honking great brown boxes that require forced air conditioning, all you need is fairly basic hardware to which you apply software to provide the functionality.’
Warren says that this enables an operator to take a micro network out to a specific location but sees some misunderstandings arising in the marketplace. ‘It’s an odd phenomenon that everyone views virtualisation as a mechanism to go to the cloud model, but this isn’t about putting loads of functions everywhere. What you’re actually doing is using virtualisation of functions to deploy very specific capability in a specific location to enable functions to be served in that micro network.’
Mark Grayson, a member of the steering committee at the Small Cells Forum, sees similar use cases. ‘We’re making use of virtualisation from the core network to address the scaled-down use case,’ he says. ‘Rural and remote use cases don’t necessarily require scale so we see NFV as having the potential to lower the barriers to entry across the boards. These techniques can de-risk the deployment.’
Amid all the excitement about telecoms transformation, it’s important when considering virtualisation to recognise it as a separate thing to network evolution to 5G. ‘Virtualisation on 5G includes a lot of good stuff but it’s dangerous to conflate 5G and NFV,’ says Warren. ‘5G doesn’t have to be virtualised and NFV and heterogeneous networks are separate and need to be delivered more quickly than 5G will arrive.’