5G networks are moving closer to realisation with the launch of the first silicon designed to support next-generation mobile standards, but this glosses over the fact that those standards are nowhere near defined and, worse still, the telecoms industry has yet to even agree on what the key use case is that 5G networks are meant to target.
At the CES consumer electronics expo in January, chipmaker Intel announced its 5G chipset, which it claimed as the “first global 5G modem” because of its ability to support a wide range of frequency bands within which 5G services around the world are expected to operate. It was preceded by Qualcomm, which unveiled its own 5G modem in October 2016.
Meanwhile, the standards bodies (principally the 3rd Generation Partnership Project, or 3GPP) working towards specifications for a 5G air interface, known informally as the “new radio” or NR, are not expected to approve final standards until 2019. This is expected to provide for data transmission speeds of up to 10Gbps, and aims to keep end-to-end latency down to less than 1ms, much lower than today’s 3G or 4G mobile networks.
However, part of the problem with 5G is the diverse range of applications that mobile networks are anticipated to support. Many expect that 5G will deliver much faster downloads for smartphones and enable workers with laptops to stay always connected with data speeds comparable to the office WiFi network. Others talk of 5G supporting emerging use cases such as remote presence using VR headsets, vehicle-to-vehicle communications, and of course, connecting up swarms of sensors and devices as part of the Internet of Things (IoT).
As a result, many of the requirements that are being tabled for the NR often seem to be pulling in different directions, making it difficult for a single technology to meet all of them at the same time.
A communications smorgasbord
For example, some proposals call for the use of so-called millimetre-wave spectrum (bands above the 24GHz region) as these high frequencies will support the very high data rates being advocated. But this comes at the cost of shorter range as these frequencies are more easily blocked by walls or trees, which would mean that cellular base stations would have to be positioned closer together.
In the UK, however, the telecoms regulator Ofcom is proposing a 3.5GHz band for 5G services, which should allow for cell sizes comparable to those already used for mobile phone services. However, this frequency is likely to imply lower data rates than would be possible at millimetre wavelengths.
Add to this the fact that 4G networks still have plenty of scope for upgrades, with equipment based on the LTE Advanced Pro standard already starting to appear and offering speeds up to 1Gbps, and 5G starts to look like a solution in search of a problem.
“The question is, what will 5G do that 4G can’t?” asks John Delaney, associate vice president for Mobility at analyst IDC.
According to Delaney, there are essentially three areas where 5G could offer differentiation: very high data rates; support for a large number of connected devices per cell (useful for IoT applications); or much lower end-to-end latency to support applications that would be sensitive to any delay, such as connected cars or remote surgery using robotics.
IoT in particular is seen as a key use case for 5G, as this could lead to many more devices all trying to operate simultaneous connections to a cell tower than is currently the case with 3G and 4G networks, which can manage somewhere in the region of a thousand.
“Whichever use case emerges as the key differentiator will influence how the 5G standard evolves,” he says. An outline version of the specifications is expected by the middle of 2018, so “we will have a good idea what will be in the final specs by then”, he adds.
Slicing up the network
One radical solution that is being considered to address the problem of having so many use cases with varying requirements is an approach called network slicing. This calls for the connection to each device to be optimised for the use case of the device it is serving.
It is called slicing, because the technique divides up the network into multiple co-existing logical networks in order to provide the most suitable resources to different types of services. As an analogy, think of the way virtualisation is used to carve up servers into multiple virtual machines.
“It’s about being able to deploy a set of characteristics for a particular connection that are different to another connection, even though both are on the same 5G network,” explains Delaney.
“So for a smartphone user, you would optimise the connection for a high data rate, but if someone within the same network cell was driving a connected car you would want a connection optimised for minimal latency,” he adds.
Network slicing will call for an overhaul of the entire mobile network infrastructure, including extensive use of technologies such as software-defined networking (SDN) and network function virtualisation (NFV). In fact, it will call for more compute power to be deployed at the network edge, effectively making cell base stations more like miniature data centres.
This will be music to the ears of Intel, which in addition to its 5G silicon, has been a driving force behind many SDN and NFV initiatives, and which is always looking for new opportunities to sell more server chips.
“We’re trying to drive a network transformation,” says Rob Topol, general manager for 5G Business and Technology at Intel.
“What we’ve been introducing recently is a concept called mobile edge computing (MEC), as well as network slicing. MEC is about taking more of the services and applications that users are accessing, and putting them closer to the edge of the network, in closer proximity to the subscribers.”
“Network slicing is essentially where you slice the software workload across the network, from the radio access to the cloud, and you allow the network to become more ‘compose-able’, so you can ensure reliability of safety-critical functionality, say for autonomous cars, and keep that separate from the people who are doing 8K media downloads,” he explains.
In other words, 5G networks will require compute power to be distributed more across the network and its infrastructure, rather than being centralised in a handful of large data centres. All of this will call for clever orchestration and use of data analytics to dynamically configure the network to meet changing demands.
Seamless integration with other wireless services
To return to the question of what 5G will be able to do that 4G cannot, Topol says that the new air interface will most certainly be required to support uses cases like IoT devices and streaming high definition video while on the move.
“Low-power IoT requires that sliced network capability to ensure reliability for all those sensors, while the same is going to be needed to support video content streamed to cars because that would essentially stress a 4G network today when you add up the number of people that might want to be doing that at the same time.”
Meanwhile, another facet of the design effort going into 5G services is looking at how to better integrate with current wireless networks. Sometimes referred to as heterogeneous networking, this is intended to enable seamless fall-back to 4G or handoff to an available WiFi network from a 5G device, so that users see no interruption of service.
On the question of frequencies, Topol said he believes that mobile 5G networks are likely to be implemented in the sub-6GHz bands, the portion of the spectrum that already contains 3G and 4G wireless signals. However, the millimetre-wave bands are likely to find business applications in fixed, line-of-sight deployment, such as high-speed campus-wide networks or as an alternative to wires for linking cellular base stations.
Intel’s 5G modem, which is due to start sampling to device makers in the second half of this year, is targeting initial speeds upwards of 5Gbps, while 10Gbps is “not unrealistic” in the future. The ITU has now published a draft report on 5G recommending a speed of 20Gbps.
To summarise, 5G networks hold out the potential to support a wider range of applications and use cases than existing cellular technologies, but unless your organisation is planning to start operating something like a wide-area IoT deployment, 5G services are not going to be something you need to worry about for the near future.
For the present, the technical specifications for 5G networks are still in flux, meaning that we are not likely to see large-scale deployments before 2020. In the meantime, 4G networks will continue to improve and offer higher data rates to support mobile workers.
However, once 5G networks start to roll out, we are likely to see them used in new and unanticipated ways that will drive demand for their capabilities, as happened with 3G and 4G before them.
“When you list out the use cases of the future, you will find there are those that will require that new air interface, even though there will still be a legacy 3G and 4G network in place for many years,” says Topol.
Jon Collins’ in-depth look at tech and society
Phil Muncaster reports on China and beyond