So in The Register's ongoing untangling of what 5G might mean, we spoke to Tod Sizer of Alcatel-Lucent.
Sizer – Dr Theodore (Tod) Sizer – is Vice President of the Wireless Research Program in Alcatel-Lucent Bell Labs, and he told El Reg that 5G has to be about more than eye-opening press-dazzling showing off what vendors can do with 5G speed. Yes, higher data rates are part of the picture, but he told El Reg that “the world is bigger than just voice and high speed downloads.”
Sizer put the case that rather than just a new air interface, whatever follows 4G – let's call it 5G for the sake of convenience – is going to be a new network architecture for both wired and wireless networks.
The 5G big picture, he said, has to be “to build a more dynamic network that can provide the connection new applications are going to need.”
It's “connection”, rather than just data rates, that will matter, and for that, some background is in order.
Sizer points out that we're quickly moving to a world where everyone's primary connection is wireless, and wired networks – whether in the home or outside – serve a backhaul function.
“Whether you're at home or mobile, you're connecting to your world by wireless,” he said. “A child born in 2000 has never known a world where you can't connect over wireless.
“The expectation today is that we can have immediate access to information – any place, any time, anywhere.”
That makes end-to-end performance, rather than just the final link, the key. “The network needs to adapt to you”.
Which, Sizer said, means that whatever 5G becomes in the end, in addition to higher end-user data rates, it also needs higher capacity overall, and greater network flexibility. “Focussing on any one of those is missing the point. The point is the experience that the user has,” he said.
In terms of flexibility, Sizer notes that different applications have different behaviours, and the next generation of network has to reflect that – for example, machine-to-machine communications have different requirements to an app that streams video to a handset.
Apps and the network
Even the way applications are built should be part of how people think about the next generation network. For example, Sizer said, consider the relationship between the application and handset life.
“It might seem an odd though to come from an infrastructure provider, but in fact there are many things you can do in designing the network that can impact battery life.
“For example, the best way to save energy is to turn out the lights, turn off the device. So we need to design the network so that if the device has information to send or a question to ask, it can turn on, send it, and go back to sleep.”
How does that impact the network? Sizer pointed out that mobile networks have inherited a model that has elaborate login procedures for handsets, and a requirement for handsets to dwell on the network for a minimum amount of time.
“An awful lot of traffic is little messages – the handset gets woken up to ask 'do I have mail?' and the server answers 'no'.”
These transactions cause congestion for the network, and they drain battery, and they have no true value for the user, “so the kinds of network design we think of to enable Internet of Things apps” would also have value for more familiar apps.
All of which leads to the question that started this discussion: The Register asked Sizer if there's a disconnect between a focus on who has the fastest air interface, and what's really going to matter to end users.
“Yes,” Sizer said. “Do we need higher speeds? Of course. But the user doesn't care about speed, they really care about whether the application works the way they expect it to work”.
A demonstration of a 10 Gbps air interface “is a severe corner case, that's not really relevant”, he added.
“If you could share 10 Gbps among 10,000 users, that's extremely valuable,” he explained. However, the obsession with demonstrating gigabit links to trumpet the 5G future doesn't help that use-case: “those technologies are inherently short-range with very directional antennas.
“If you don't have the range, you don't have the numbers and the amortisation of a 10 Gbps link or a 1 Gbps link is not feasible”.
Rather than raw speed, Sizer said, the 5G world needs a “network that provides the appropriate quality of experience for a particular application”.
Texting has only low bandwidth and modest latency demands; someone playing a multi-user shooter considers latency to be super-critical. A two-way video call's bandwidth demands are higher, but latency is again more important, while a streaming service such as Netflix puts a premium on bandwidth over latency.
Right down to the level of a user in a cell, he said, the network needs to adapt to the individual and the application that they're using.
“There's another metric, something that I call responsivity (and yes, we need a better word for it). If you have a message to send, how quickly can the network accept that message and pass it on?
“That is crucial for Google in searching – they want to take the question you're asking, and get you the answer very quickly.”
There's a strong research stream among vendors to address this question, he said, because “it plays very strongly into battery life. If the network can help the response happen quickly, then the device can go back to sleep.”
Radio access of the future
Another aspect of whatever emerges as 5G is that it will assume that all devices will be built with multiple radio access technologies – WiFi, millimetre-wave, and cellular.
If devices are to be able to use more than one access technology at once, Sizer said, the network has to be able to put that traffic back together “so you can have an aggregation of connection between the different types of technologies available.
“That's another measure of context – which access points are around you,” he said. “What we want to provide is a 'network of me' – the user is the centre of the universe. The network adapts itself to where the user is, doing what, requirements of the application, and adapt accordingly.
“5G is the first generation that will truly embrace the myriad ways to connecting to a user – those traditionally wireline, and traditionally wireless, and marry them into one.”
“It's about recognising that the handset has multiple radios in them – why am I not using all three?”
“4G isn't going to go away – you need to design a solution that can embrace Wifi, 4G and the new capabilities in the 5G technologies,” he said. ®