Week in Review

5G: Atomic clocks, Massive MIMOs and sliced networks

Bernard Hickey travels to Dusseldorf and Milan, courtesy of Vodafone, where he unravels and explains the technologies, the acronyms and the power of 5G

You know a technology has become exponentially more powerful and precise when it needs a whole new timing system to keep it on track.

Understanding why 5G networks need atomic clocks helps explain the quantum leap in its power and the potential it offers. Working out why 5G needs antennas able to follow a phone or sensor with its own ‘beam’ of data and can ‘slice up’ networks for particular tasks or customers further illustrates the scale of the change.

Vodafone is currently rolling out 120 5G base stations in Auckland, Wellington, Christchurch and Queenstown, including 20 ‘COWs’ (Cell on Wheels) so it can switch on its 5G network in December. It will be the first 5G network in New Zealand, which will be the 15th country in the world to go live with this next generation of mobile telecommunications. Vodafone is using Nokia’s 5G equipment and will be the 10th Nokia 5G network to go live.

But just how different and big is 5G?

As I wrote in the first 5G article in this four-article series on what 5G means for New Zealand, the leap has the potential to transform the global economy in the same way the arrival of James Watts’ steam engine in 1775 combined with industrial-scale steelmaking and the telegraph to create the industrial revolution of the 1800s.

Here’s how they’re very similar: they both needed new ways to measure time.

The development of longer-distance railways created the need for standardised time zones through the second half of the 1800s so trains could literally run on time.

Previously, each town and city had their own sun dials and clocks set to their own particular sunrises and sunsets and it didn’t matter exactly when you left or arrived these towns, because you typically couldn’t walk or gallop faster than the sun. But train passengers, drivers and schedulers began missing their deadlines through the 1800s as trains got faster, as this Quartz explainer shows.

The term ‘Greenwich Mean Time’ was coined as the basis for time used by all railways in Britain from 1848. It was based on the solar midday point measured at the Royal Observatory in Greenwich, London and eventually became coordinated ‘Universal Time’, which is now based on time set by a network of atomic clocks around the world.

Why does 5G need atomic time?

Fast forward to 2019 and 5G network builders all over the world are having to do something similar because their regular clocks are just not precise and standardised enough to deal with the massive improvement in the speed and volume of data movement, and the need to synchronise devices and processes over both vast and short distances.

That could range from a remotely-operated surgical device where the surgeon needs to ‘feel’ the tissue in real time and be absolutely sure there will never be any ‘lag’.

The same goes for autonomous vehicles, Internet of Things networks of devices and sensors that need to work in real time. There is no room for even the smallest of hiccups or mistakes if the networks of base stations, sensors and devices aren’t operating in real time.

4G networks and base stations already need split-second time coordination, but they can rely on the usual clocks and on getting the ‘atomic-time’ reasonably regularly from GPS and other satellites, which have such clocks on them. There is enough ‘wiggle room’ between the packets of data to allow the occasional dropout with 4G, in part because the data volumes and speeds aren’t high enough to cause problems and there aren’t that many applications that require always-on real-time responsivity.

Until now.

5G allows data volumes that can be anywhere from 50 to 100 times larger than possible with 4G, with volumes expected to rise from 10 or 20 megabits per second to up to 1024 megabits or 1 gigabit per second at first, with even higher in later iterations of 5G. The ‘latency’ or time it takes for data to travel from one device to another with 5G is as much as 1/50th of 4G. Latency with 4G is around 50 milliseconds, but 5G’s aim is to get latency under 10 milliseconds and close to one millisecond.

It may not seem like a big difference, given the blink of a human eye takes 300-400 milliseconds and it takes 200 milliseconds to recognise human emotion. But one millisecond is essentially what it feels like to be in real time. One example we saw in Vodafone’s 5G lab in Dusseldorf was an air hockey game which showed how quickly a robot opponent reacted at one millisecond vs 50 milliseconds. It was the difference between a lagged loser and a robot that never lost.

One millisecond is also fast enough that if systems are not using the same very-high-accuracy clocks, then they start stumbling over themselves and, most importantly, 5G is being done in such tight blocks of bandwidth and such tight synchronisation of uplinks and downlinks of data that any stumbling will cause problems.

“Clock synchronization is really important end-to-end so that you don't get packets colliding,” says Vodafone’s Technology Director Tony Baird.

“Then you get great retransmissions, and then you get that wheel of death ... so clock synchronisation is really important,” he says.

Baird is in charge of Vodafone’s network and is planning to convert most of its base stations to tell the time from GPS satellites. But it matters so much he’s also buying an iridium atomic clock to tell the network the correct time if the GPS network goes down.

“So we've got a million dollars’ worth of clock. It's really expensive, but I think it's worth it.”

For those worried about an atomic clock in Vodafone’s buildings, it is not radioactive and uses oscillations between the nucleus of an atom and its electrons to measure time.

What is ‘beaming Massive MIMO’?

One reason 5G technology is able to direct such large volumes of data so quickly at devices is the radio antenna is exponentially more sophisticated in how it can ‘beam’ data to a customer’s device or sensor.

Antennas used in 4G networks typically broadcast just one wide blanket of coverage. Antennas used in the first 5G networks will use 8 by 8 arrays of ‘dots’ of antennas that are able to create 64 ‘beams’ to and from multiple devices from the one base station. Think of it like the adaptive and responsive headlights on a modern car that can ‘steer’ around corners. These antennas are able to direct beams of coverage that follow a device, including being directed upwards into a building.

These antenna arrays, which are typically no larger than the 4G ones, are called Massive MIMO (Multiple Input Multiple Output) antennas.

This ability to beam back and forth in such an intense synchronized way allows the massive increase in data volumes and speeds. It also allows special events, such as a rugby test at Eden Park with 65,000 fans wanting instant video replays to be covered by a Massive MIMO antenna, rather than many, many regular ones.

“With massive MIMO, you can do beam forming inside the stadium during the game and then outside the stadium after the game,” says Baird.

“So as people were migrating out (of the stadium) you can move the capacity around as you need it.”

All this will allow the network to handle connections to up to one million devices per square kilometre, which is ten times larger than for 4G.

The network slicing effect

The other thing that 5G does differently is allowing so-called ‘network slicing’, which allows the virtual ‘slicing’ and ‘dicing’ of the services and plans offered on the same network. It’s the difference between 4G’s shotgun blast and a series of laser guided missiles.

This means a particular customer can buy the particular version of speed and volume of data that suits them best over exactly the same network as another type of customer. For example, an autonomous car company would rely on vehicle-to-everything (V2X) communication, which requires very high speed data travelling from the car-to-devices-and-the-cloud, but doesn’t necessarily require massive volumes of data. A car manufacturer could design its own package of low latency, low bandwidth services in two or three specific locations in a ‘slice’ of the new network.

Meanwhile, a 4K video streaming service to the same car would require high throughput or volume of data, but could handle slower ‘latency’. The same network and the same antennas would virtually slice and coordinate the two services so the passenger could seamlessly watch the video in a car that didn’t crash.

What about 5G and ‘Mobile Edge Computing’?

All this extra data and speed open up the possibility of putting processors at the edge of the network that handle the ‘grunt’ work of processing data from sensors and using algorithms to make decisions in real time for robotic operations.

This is described as ‘Mobile Edge Computing’ or MEC and is seen as crucial to the development of industrial automation, autonomous vehicles, augmented reality, video monitoring and analysis, distributed Blockchain ledgers and predictive maintenance of equipment.

This would mean network operators such as Vodafone would install processors in or close to the base station that would talk with bigger databases in the cloud, and do the actual computing needed by the likes of industrial robots, autonomous cars and electricity networkers.

“That's putting caching and control right at the edge of the network, which will be really useful for the real low latency applications as well,” says Baird.

Google, Netflix and Amazon already have caching servers in many of the exchanges around the country to ensure the fastest download speeds for streaming, searching and buying. But edge computing would take that to a whole new level and enable industrial applications such as robotic landscaping or construction.

3.5 ghz first, then 26 ghz

The first versions of 5G being rolled out in New Zealand by Vodafone will be around the existing 3.5 gigahertz spectrum, which is being used for 5G testing and is already used for 4G and WIFI networks at the lower end.

Vodafone has a combined 80 megahertz in four blocks in this part of the spectrum that it will use in a first wave of development over the next two to three years.

But the next wave of development around the 26 gigahertz is where 5G really starts cooking with gas - because the higher the frequency of the radio wave, the more information can be packed in.

Vodafone already has some of that spectrum through its acquisition of TelstraClear and could also install 5G antennas in the 660 streetside cabinets it inherited from TelstraClear in inner city areas. This is the part of the spectrum known as the ‘millimetric’ band and is able to pack in the most information in the Massive MIMO beams for very short distances from the cabinet into surrounding buildings.

“We've got fibre to those cabinets, which is a key ingredient for network slicing,” Baird says.

“We've got D.C. power in those cabinets, so we'll pull out all of the old legacy switching equipment and put a 5G millimetric radio in it and it will go one to two hundred metres,” he says.

“So it's really going to be great because it's high, high frequency. It'll be great for a drop into this building for, you know, 1 to 10 gigabytes per second of bandwidth so you don't need a fibre tail anymore.”

This explains part of the strategic rationale for Vodafone’s big push into 5G. That sort of data volume allows it to sell fixed wireless products to consumers and businesses that would be able to replace optic fibre plans that are currently mostly provided wholesale by Chorus.

“This would be a real competitive threat for Chorus and that's what we will be building. A lot of this will be around fixed wireless for us.”

The rights to these two blocks of spectrum (3.6 Ghz and 26 Ghz) are due to expire in New Zealand in 2022 and the network operators hope the Government will be able to re-auction those chunks of spectrum well before then, although it is also dependent on a separate but contingent Treaty of Waitangi spectrum negotiation. An auction is expected some time next year.

There is another third section of spectrum around 600 megahertz and 1400 megahertz that could also be used in New Zealand, and is being used overseas. The spectrum around the 600 megahertz band is currently used for Digital Terrestrial Television by Maori Television from 500 megahertz and at higher levels by the types of radio microphones used in conferences and concerts. The 1400 megahertz chunk is currently used for direct radio links by the likes of the Police, Airways and Chorus. These chunks of frequency are seen as the third most likely to be used, although their lower frequency allows greater distance and penetration.

So is it safe? Yes.

Understandably, some uninformed consumers are asking whether, if the new 5G networks are up to 100 times more powerful than 4G, that means the electro-magnetic waves are therefore 100 times more ‘dangerous’.

In short, no. The frequency of the radio waves used in 5G is low enough not to be in the ionising radiation part of the spectrum, as shown here in this BBC explainer, and in this New York Times explainer. This WHO report concludes radio waves from 0 to 300 Ghz (5G is 3.6 to 26) “do not produce any known adverse health effect”. This report and graphic below from the GSMA, the global industry standards body, also goes into the non-existent health effects of the electro-magnetic fields used in telecommunications.

A key thing to know about the radio spectrum used by mobile networks is that the higher the frequency of the band, the more information can be packed in. The shorter the wave, the lower the frequency of the wave, the lower the amount of information and the further it can travel. It’s why we can listen to short wave radio from the other side of the world. The higher the frequency, the shorter the distance the waves can travel and the less powerful the radio signal has to be.

That means that high frequency radio waves pack more data in, but can travel less distance and also use less energy and power for each of the little antennas in the Massive MIMO antenna array. That’s why, counter-intuitively, each little 5G ‘rifle shot’ antenna uses less energy (2-20 watts) than the energy used by 20-40 watt antennae used in the ‘shotgun’ 4G networks.

So 5G’s arrays of MIMO antennas are individually smaller and less electrically powerful than the existing 4G antennas. They ‘beam’ over shorter distances. They also collectively use 25-30 percent less electricity than 4G networks, which is another reason why network operators such as Vodafone are keen to move to them.

So paradoxically, a network technology that is 100 times more powerful in terms of data volumes and speeds, is actually less powerful in terms of electro-magnetic activity.

Bernard Hickey travelled to Dusseldorf and Milan to visit Vodafone’s 5G labs and Nokia R&D centres courtesy of Vodafone, which is a foundation sponsor of Newsroom.

Can you help our journalists uncover the facts?

Newsroom is committed to giving our journalists the time they need to uncover, investigate, and fact-check tough stories. Reader donations are critical to buying our team the time they need to produce high-quality independent journalism.

If you can help us, please donate today.


Newsroom does not allow comments directly on this website. We invite all readers who wish to discuss a story or leave a comment to visit us on Twitter or Facebook. We also welcome your news tips and feedback via email: contact@newsroom.co.nz. Thank you.

With thanks to our partners