Ski jumper

2018 Winter Olympics will be the proving ground for 5G technology

Image credit: Image source, Getty images, Rex features

The 2018 Winter Olympics will give 5G an early debut as operators promise a transformative mobile network.

By the end of October, and after a delay of more than a year, UK communications regulator Ofcom expects to have sold a large chunk of the spectrum needed to deploy the fifth generation (5G) of cellular networks, a shift that many operators believe will transform not just their businesses but everyone else’s. And they are now in a hurry to get there.

The auction process is already beset with controversy because of the decision to impose a cap on how much of the valuable spectrum each operator can win. BT’s EE subsidiary is already at the limit of what it can use immediately and so can only bid for a fraction of the spectrum that is to be auctioned for 5G. But O2 and Three pushed for a much tighter cap. Today, their combined spectrum allowance still falls behind BT’s allocation despite Three increasing its capacity by a third following its purchase of fixed-wireless provider Relish.

In the US the situation is even worse. At the Brooklyn 5G Summit in late April, Dave Wolter, assistant vice president of radio technology at AT&T, said: “There is currently no spectrum allocated to 5G in the US with the exception of 3.5GHz. That band has rules that don’t make it attractive as a base 5G layer.”

Operators are pushing ahead, looking at ways to introduce 5G in the face of a massive spectrum shortage and buying up parcels of spectrum that they expect to be able to deploy for the new protocol. T-Mobile, for example, snapped up 600MHz bands in the US originally allocated to 4G with the aim to quickly convert them to use 5G. The 3GPP standards body behind the huge pile of specifications for 5G has put much of its effort into defining ways to eke the most out of the radio spectrum operators can cobble together, whether it has been recovered from analogue TV stations or pushes into frequencies previously considered too difficult and expensive to use.

The Winter Olympics next year in PyeongChang, South Korea, will provide one of the first large-scale tests of whether the 3GPP ideas will work and a demonstration of the kinds of services we can expect from 5G.

HongBeom Jeon, executive vice president at Korea Telecom, says: “There are technologies we can introduce to the world, new kinds of video services. They are not just 4K video but new kinds of experience.”

Previous Olympics technology demonstrations such as those for 3D television have been a continuation of the ‘outside the track’ view that an ordinary spectator enjoys. With its 5G trials, Korea Telecom wants to go inside the skating rink with its own version of the bullet-time effect popularised by ‘The Matrix’ and the ability to give spectators the opportunity to see the views in front of skiers and skeletoneers (see ‘Games go virtual’).

Korea Telecom aims to be one of the first mobile operators in the world to offer 5G to consumers, announcing at Mobile World Congress (MWC) in March it would offer the first services in 2019. That means doing development in parallel to the 3GPP standardisation effort. Although the standards group has accelerated its plans for a subset of the complete specification to make early deployment possible, for its Olympics trials Korea Telecom had to start without it.

“We defined a specification last year with global partners and developed systems and services to provide the trial,” Jeon says. “The aim of the specification was to provide a maximum of 20Gbit/s throughput and less than 1ms latency.”

To get there, Korea Telecom adopted the same policy that US operators seem destined to employ in the absence of uncontested spectrum in conventional bands – aim higher in terms of frequency and move into the millimetre-wave bands. The operators opted for an 800MHz-wide band around 28GHz.

Jeon says that when the company started work, “for mobile services, there was no data on its performance both for outside and indoor use. So we started channel modelling.”

Korea Telecom set up a limited test in the centre of Seoul. Engineers deployed three basestations that would communicate with a bus fitted with 28GHz transceivers. The biggest issue with RF communications as frequency increases is that the waves propagate more like visible light than radio. If the energy is sent in the wrong direction, the receiver will receive practically nothing.

The directivity of high-frequency communications has a key advantage for mobile operators when it comes to delivering gigabits per second to many users in a cell. The basestation can aim beams at individual users instead of dividing up transmissions into tiny packets that are allocated in round-robin fashion to receivers. But this beamforming relies on the use of many antennas and the application of high-​performance digital signal processing (DSP). And, very often, even taking into account reflections from buildings, it is not possible for a single basestation to serve the same users continuously. Instead, basestations need to cooperate so that if there is no path to a user another one that is in range can take over.

Even with the three basestations arranged around Seoul’s central boulevard, Korea Telecom found the signal strength to and from the bus varied much more than with LTE, with the biggest drops often occurring as the bus turned a corner.

“Millimetre-wave is very delicate and very difficult to commercialise in these very crowded areas,” Jeon says, adding that the Seoul trial provided enough data for the installation of similar basestations around the Olympic village.

“We are also developing the user terminal,” Jeon explains. “In 2015, the user terminal was big, designed to sit on top of a car. We are now developing SoCs [systems on chip] with our partners so we can provide tablets and mobile handsets for next year’s Olympics. We hope that devices will be ready at the end of this year.”

Full-5G handsets may have to wait longer, although concerns over the transmit power needed to support high transmission rates back to the basestation would not just severely limit battery life – a problem that slowed the introduction of 3G in the early 2000s – but exceed safe RF exposure limits. Intel senior fellow Ken Stewart says: “There are significant issues to be resolved that are conditioned on the standards process.”

One advantage to operators who deploy 5G early is that the millimetre-wave spectrum could support the infrastructure needed to deploy smaller lower-frequency basestations into dense urban areas more cheaply. The ‘self backhaul’ approach feeds data received on sub-6GHz bands onto broadband millimetre-wave links between basestations. This overcomes the problem that early millimetre-wave systems are likely to have dealing with mobile users. More importantly, the operators do not have to dig up roads to lay fibre to each of the small-cell basestations.

The demonstrations planned by Korea Telecom aim at one corner of a triangle that describes the range of services operators expect to be able to cover with 5G. The scope of 5G goes much further according to Ken Budka, senior partner at Bell Labs Consulting. A less obvious part of the 5G specification, that is potentially far more troublesome than delivering gigabits per second of data to consumers, is the need for the network to respond to each transmission within 1ms. “Low-latency services have the power to transform human existence,” Budka claims. “It is not just industrial control applications that require low latency and high bandwidth. With emerging virtual reality applications we need to deliver up to half a gigabyte per second to a headset or device and do that with very low latency.”

Virtual reality needs low latency because of the reflex in humans to move their eyes in the opposite direction to the rotation of the head. If the image painted by a computer reacts too slowly to this, people suffer from motion sickness. Low latency is critical to another application operators see as a major opportunity for them: autonomous driving.

The result is a reworking of the way in which terminals and basestations communicate. In 3G and even more in LTE, data is encoded to try to maximise efficiency for many users and the bits are interwoven in a complex sequence of transactions. 5G reverses that entirely. As soon as each small packet of data turns up from a single user, it is processed and acknowledged immediately. And this change is among the first to hit following a decision in March to accelerate the deployment of the New Radio (NR) standard for 5G.

Other parts of the standard will not be available until around 2020. The result is that 5G-NR services will piggyback on the 4G infrastructure. The standalone 5G network, which will deliver the ability to have basestations cooperate much more closely than is possible today, will slowly replace the older 4G core and let operators switch over to protocols that make more efficient use of their existing spectrum.

The requirement for low latency will have one effect on the design of the 5G network from day one. Although reworking the protocols for 5G-NR cuts processing time, the laws of physics make the 1ms target very difficult to achieve with a conventional structure. “Every 100km of distance costs you about a millisecond,” Budka explains. “In this new network, the only way we can deliver low latency is to bring the network closer and closer to the user. Processing will typically be within 100km of where your devices are.”

At MWC, Ericsson president and CEO Börje Ekholm said: “5G will not be built like 4G. 5G will be cloud-based. We think even the network will be in the cloud in the future.”

Cloud servers that manage platoons of autonomous vehicles, for example, will most likely have to be running in the same country and not devolved to remote offshore facilities. Similarly, groups of robots in factories that connect to the network using 5G will take orders from servers nearby, with only long-term planning managed internationally.

In sectors such as construction, Budka adds, “networks might be temporarily deployed to a job site”. Similar local 5G networks might provide a boost to applications such as mixed-reality gaming, where players wearing headsets today need to wear a 20kg backpack containing the computers. A fast video-link would reduce the wearables to just the electronic visor, says Durga Malladi, senior vice president of engineering at Qualcomm. 

The final vertex of the 5G triangle lies far away from the requirements of the other two. Already the subject of two additions to the LTE standard, low-power devices for the Internet of Things (IoT) are likely to get more efficient access to cellular in 5G. These devices will, for the most part, use spectrum taken away from TV companies in the 700MHz region, taking advantage of the ability of lower frequencies to penetrate walls and floors to provide direct access to meters hidden away in cupboards and cellars. That called for more changes to 5G-NR but not a fundamental shift.

Although some researchers have argued the technique has flaws, the industry has stuck with the orthogonal frequency division multiplexing (OFDM) modulation format for transmitting data used in LTE, as well as wired systems such as broadband over copper.

The problem with using OFDM was that it was difficult to find a suitable compromise in terms of the width and spacing of the many frequency channels a band might be split into. The answer was to make the values scale with frequency, with very narrow, low data-rate channels below 1GHz but broadening out as they push past 3GHz. Along with other protocol changes to help manage thousands of devices per cell that are only actively sending data intermittently, 5G is meant to offer better energy efficiency than the new IoT-specific modes of LTE.

With the applications spanned by the 5G triangle, operators are expecting 5G to become the core of the future internet. “I believe over the next ten years we will see a massive transformation through customer adoption of 5G,” Ekholm claims. “Look at new capabilities such as virtual reality, augmented reality and artificial intelligence: you will see why we need 5G.”

Henning Schulzrinne, former CTO and now an adviser to the US Federal Communications Commission, sounds a note of caution on 5G. The process carries echoes of 3G’s painful birth. “What worries me about 5G is that it could in its market-driven rush be a missed opportunity. The one thing that has been constant across [cellular] generations is that our collective ability to predict what would happen has not been so good. Take 2G – we thought it would offer better voice quality.

“What really happened, and what drove 2G beyond being just a bad landline replacement, was SMS. It wasn’t planned or predicted but it became a major driver.”

With 4G, Schulzrinne says, the emphasis was on high-quality videoconferencing and a better voice codec. The users embraced push notifications and YouTube videos.

“With 5G today, the big thing I’ve been hearing is low latency. It will allow autonomous vehicles or whatever. Pardon me for being a bit sceptical; it’s not that these predictions were wrong, its just that these things didn’t drive adoption,” he argues. “One constant across the five generations is in different ways, each new arch is branded and sold on quality of service and it turns out that that prediction is far less important than was indicated in keynote talks. We have a history of making networking a quality-of-service story. We should by now have learned whether that truly pans out.”

Seizo Once, CTO of NTT DoCoMo, argues that even if the applications predicted for 5G fail to emerge, operators can still win by deploying the newer technology as, despite the incredible complexity of the protocol, it does promise more efficient spectrum usage. “Data-capacity enhancement can be one of 5G’s killer services,” he says.

Schulzrinne takes the same view of what will emerge: “What’s going to drive it is dollars per gigabyte cost. That’s what people are looking for: high capacity at low cost.” 

Games go virtual

In its trials of 5G services at the 2018 Winter Olympics, Korea Telecom’s focus is on delivering interactive synthesised images to users armed with purpose-built smartphones and tablets. Among the features is the ability to zoom in on skaters on the indoor rink and rotate the image much like the bullet-time effect used in the movie ‘The Matrix’. To do it, servers will pull images from a hundred cameras hung around the rink, reconstituting them on the fly to create a view customised for each subscriber. For the ability to spin a skater through 360° in high definition, the system needs to be able to deliver 400Mbit/s, according to HongBeom Jeon, executive vice president at Korea Telecom.

On the ski courses, the Omni Point system will let viewers track the progress of their favourite competitors. Each skier wears a GPS receiver that passes their location live to the Korea Telecom servers. Using images pulled from cameras around the course, the servers synthesise the view the competitors see in real-time. Unable to use fibre to connect the cameras, Korea Telecom has fitted them with wireless transceivers. A more conventional view from the cockpit will be available for events such as the bobsleigh, with a live view of the course as teams barrel through it.

Visitors to the games can use the telecom operator’s self-driving 5G bus to move between skiing events around the resort. The bus will have screens inside to display the multimedia demonstrations, some using 3D display technologies.

Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.

Recent articles