The term ‘5G’ is bandied about a great deal these days, but even a cursory look into the subject reveals there is very little agreement within the industry as to what it actually is, or even might be.
Despite this, 2020 is being targeted as the date for the first commercially available 5G network. James Goodwin of Anite, the mobile device, infrastructure and network testing firm, says: ‘5G was hijacked as a marketing term very early on, but there is no clear definition of what it is yet.’
Part of the problem, according to Håkan Andersson, Head of Mobile Broadband, Radio Business Unit at Ericsson, is that up until now each mobile phone generation has been easy to predict, but what 5G will be is proving harder to identify.
As a result of this, almost any new network technology innovation not already defined in the 4G LTE or LTE-Advanced standards is being lobbed into a catch-all bucket labelled 5G. This is causing confusion and giving rise to misconceptions as to what should and should not be defined as 5G.
Two industry organisations – the GSMA and the Next Generation Mobile Network Alliance (NGMN) – have attempted to bring some clarity to the situation by publishing White Papers. In December 2014, the NGMN released an Executive Version of its 5G White Paper ahead of full publication at Mobile World Congress in March 2015.
The NGMN has taken the approach of identifying 24 use cases for 5G grouped into eight use case families. These are designed to serve as a way of identifying the requirements needed to deliver the use cases and therefore help to define the cornerstones of the 5G architecture.
The Alliance expects 5G to provide ‘much greater throughput, much lower latency, ultra-high reliability, much higher connectivity density and higher mobility range’. It believes 5G must come with ‘embedded flexibility to optimise the network usage, while accommodating a wide range of use cases, business and partnership models’.
The NGMN notes that if the expected proliferation of sensor networks and the Internet of Things happens then more capacity will be required almost certainly from spectrum in the higher frequency millimetre bands. This in turn may require the development of a new radio interface.
However, the NGMN also emphasises that 5G will not only require technology changes, but the creation of new business models between asset, connectivity and partner service providers.
Also published in December 2014, the new GSMA Intelligence report, Understanding 5G: Perspectives on Future Technological Advancements in Mobile, outlines the technical requirements of future 5G networks and explores potential use cases, as well as the implications for operators and other mobile ecosystem players. It identifies two main industry views, which it notes are frequently mixed together to form the basis of the 5G definition.
View 1 – The hyper-connected vision: In this view, 5G is seen as a blend of existing technologies (2G, 3G, 4G, Wi-Fi and others) that can deliver greater coverage and availability, higher network density in terms of cells and devices, and the ability to provide the connectivity that enables machine-to-machine (M2M) services and the Internet of Things.
View 2 – Next-generation radio access technology: This perspective outlines 5G in ‘generational’ terms, setting specific targets that new radio interfaces must meet in terms of data rates (faster than 1Gbps downlink) and latency (less than 1millisecond delay). The two views identify eight core technical requirements for 5G.
1-10Gbps connections to end points in the field (ie. not theoretical maximum)
1 millisecond end-to-end round trip delay (latency)
1,000x bandwidth per unit area
10-100x number of connected devices
(Perception of) 99.999% availability
(Perception of) 100% coverage
90% reduction in network energy use
Up to 10-year battery life for low-power, machine-type devices.
The GSMA notes that ‘because these requirements are specified from different perspectives, they do not make an entirely coherent list’ and adds that ‘it is difficult to conceive of a new technology that could meet all of these conditions simultaneously’.
However, in the GSMA’s view only two of these: plus-1Gb data rates and sub-1ms latency, relate to a true generational shift, with the remaining six being either economic objectives or aspirations applicable to all network technologies.
Dan Warren, senior director, GSMA Technology, and one of the two main co-authors of the report, says: ‘The real thrust of the report is that there is a lot of good technology innovation taking place already, but people are chucking everything under the 5G banner simply because it is future focused. But then you end up with a confused state of 5G.’
He argues that technologies such as network functions virtualisation (NFV), software-defined networks (SDN), heterogeneous networks (HetNets), Low Power, Low Throughput (LPLT) networks and mobile edge computing are being developed already and should be worked on independently of 5G as fundamental building blocks for networks anyway.
Not really 5G
In the GSMA’s view, 99.999% availability and 100% geographic coverage are not use cases or technical issues, but economic and business case decisions, as both are deliverable using any current technology.
Similarly, higher connection densities and numbers of connected devices can be achieved with existing technologies, although 5G can clearly contribute to and further enhance these goals.
Reducing network energy usage is an economic and ecological goal, and the GSMA says it is not yet clear how adding 5G technology with higher bandwidths as an overlay on top of existing network equipment can result in a reduction in power consumption.
Work to improve the battery life of M2M/IoT devices requiring only occasional connectivity, low throughput and signalling load is already well under way, both in practice and in the standards bodies, and many use cases can and are being delivered without the need for 5G.
Hence, the GSMA report argues that this leaves only sub-1ms latency and plus-1Gbps downlink speeds as unique attributes of 5G, which require a genuine technological generation shift.
Anite’s James Goodwin says: ‘A major factor driving 5G is latency and the vision of a tactile or haptic internet. If you want to provide remote medical services with a surgeon at one end of remote link, you will need to get data to and from the patient being operated on in millisecond-type timescales.’
Autonomous driving and connected cars will also need response times close to zero for safe operation and collision avoidance (not to mention 100% coverage), so 5G’s 1ms latency is clearly called for here.
When it comes to high data rate applications, Goodwin says: ‘If we want to consume video at higher and higher resolutions we will need to push data rates to 1Gb and beyond. But if you combine fast data rate applications with millions of low latency devices then you need a huge increase in capacity and that is not possible with the current technology.’
Options for increasing capacity include: using Wi-Fi and other unlicensed bands; getting more out of existing spectrum by using technologies such as Massive MIMO (multiple input, multiple output); and also finding more spectrum.
‘But the kinds of applications being proposed for 5G means mobile operators will almost certainly need to find more spectrum and with little available spectrum left in the sub-6Ghz bands that means going up the bands to millimetre frequencies,’ says Goodwin.
This means a new air interface for the millimetre frequencies will be required. Goodwin observes: ‘There are lots of different thoughts about how you optimise the air interface and handle the interference mitigation, and it is still very open as to how it might be done. We need to develop different modulation techniques that are better suited to millimetre wave technology, so there’s a lot of work to be done across the board.’
Anite is therefore investigating how radio waves work at those very high frequencies. ‘We are looking at the channel model, trying to understand the propagation properties and how they are affected by the environment, and NLOS (non-line of sight) behaviour. In fact, millimetre wave frequencies have better propagation than people think, so they may make a viable option,’ says Goodwin.
Ericsson’s Håkan Andersson believes 5G is not about increasing mobile broadband subscriptions. ‘It is not about how we need a new technology; it is about how wireless can make an impact on other parts of society that are not using it much yet. We need to focus on developing the set of requirements that the industry needs to put in place to support these new applications.’
So what does the network need to do to support these new requirements? Ericsson’s vision is that the network becomes the physical infrastructure to cater for multi-standard, multi-layered networks designed to support different use cases.
‘Some of these use cases might focus on long battery life and low data rates, while others will be about high data rates and low latency. Sometimes those requirements will be contradictory, so that is what we have to sort out,’ says Andersson.
Cost benefits of virtualisation
However, Andersson points out that if operators were forced to build a completely new network for 5G then many of the applications being talked about wouldn’t take off.
‘But if we can add virtual networks then they can,’ he states. ‘NFV provides a software realisation of the network on common hardware. The mobile network shares underlying resources below the software network. By using these means we can deliver these new requirements in a cost-effective way.’
Andersson points out that technology changes in the radio alone are not going to be enough. ‘Pure modulation can only get us so far, as the gains here are not that big, so we must address network changes too.’
He suggests we can use smart antennas and massive MIMO, especially in the higher frequencies, to do very advanced beamforming. ‘You have a fairly small form factor with antennas, which has the ability to send a very narrow beam to each user to extend range and cut energy use.’
Warren believes it will be challenging to deliver this to enable mobility. ‘You are moving away from a network design where the cell projects pretty much 360 degrees to one that is transmitting lots of very narrow beams to individual devices.
‘The cell beam needs to track the angle of the device, or its antenna, and adjust its beam accordingly; and it needs to handle multiple radio interfaces to many bearers simultaneously. So, if the plane of the antenna is wrong, it will receive the wrong signal. We have got ideas about how this can be done though,’ he adds.
What operators are likely to end up with is a multi-layered network of 2G, 3G, 4G, 5G and Wi-Fi. Andersson says: ‘The network will be made up of multi-layer technologies and as the frequency range expands from 1GH to 100GHz and above, they will have different characteristics. The question is how to combine them so the lower frequencies provide a blanket layer and you boost data rates using the higher frequencies.’
For him, 5G is not about replacing existing networks. Rather, it is about enhancing them by introducing new capabilities that in turn open up new use cases. It is about providing a network that intelligently understands the demands of the user in real time and dynamically marshals its resources to provide the optimum solution for the application being deployed.
Andersson observes: ‘Sometimes we try to look for the killer apps, but we are not really doing that with 5G. Instead, we are trying to build a killer network, because we cannot foresee what all the applications will be in the future. The question then is: how do we cost-effectively develop these services for society?’ he asks.
In Warren’s view the real challenge will be whether sub-1ms latency can be achieved. ‘You are running up against the laws of physics and speed of light,’ he cautions. ‘We have 10-20ms latency at the moment, which comprises about 4ms on the device itself and 4ms on the base station: those two parts need to get down to 0.5ms each, so we are looking at an order of magnitude drop just in the signal processing.’
The GSMA report also notes: ‘Services requiring a delay time of less than 1ms latency must have all their content served from a physical position very close to the user’s device. This is likely to require a substantial lift in Capex spent on infrastructure for content distribution and servers.’
This poses the question of whether sub-1ms latency can really be achieved, or if it can, is the cost of implementing it economically viable? If not, then sub-1ms latency may be quietly dropped from the 5G definition.
Goodwin says: ‘5G will be an evolution of some factors, but we will need a technology step change in other factors to allow us to address certain areas. 5G networks will be a mix of macro cells combined with a collection of small micro cells, probably on these higher frequencies.’
‘5G is really a journey,’ agrees Andersson. ‘There will not be a giant step at the 2020 point; it will be evolution. Yes, we will add higher frequencies, but on top of 4G as a key layer. We want to run these layers in parallel, so we need to see how we do this.’
Warren points out that 4G LTE is still in its infancy in terms of rollout (to date only 6% of global connections are LTE), so to start looking at 5G without first dealing with 4G, which is an amazing enabling technology in itself, is perhaps a bit hasty.
‘What really concerns me,’ he says, ‘is that someone will build a network against some of these requirements and call it 5G. But then you may end up with a situation where one operator’s 5G is different to someone else’s 5G and then the consumer gets confused. It is imperative we develop and define what 5G is and use it consistently so networks are compatible.’