The arrival of Dense Air has big implications for New Zealand’s cellular market, apart from anything else it will spice up the next spectrum auction.
London-based Dense Air has purchased a considerable amount of New Zealand wireless spectrum from Malcolm Dick’s Blue Reach business and Cayman Wireless.
The company intends to set up a wholesale small cell mobile network. Dense Air says it will not compete direct with existing mobile carriers. It says it can begin operation “almost immediately”.
Dense Air now has rights to 70 MHz of spectrum in the 2.5GHz band. Of this, it acquired 30 MHz from former CallPlus owner Malcolm Dick who previously talked about running a similar wholesale cellular operation using his Blue Reach brand. The rest comes from Cayman Wireless which is a part of Craig Wireless, a Canadian company.
The New Zealand Herald reports Dense Air paid a total of almost $26 million for the spectrum. That is about 13 times the amount the owners paid for the spectrum in 2007.
There are implications for prices when the government decides to auction 5G spectrum some time in the next 18 months or so. If Dense Air decides to enter that auction it will push prices higher and could edge out cash-strapped 2degrees and Vodafone.
Dense Air is unknown in New Zealand. The company began operation in February of this year and part of US-based Airspan.
The company says it is a new class of wholesale network operator. It aims to “enhance and extend” coverage and capacity for existing mobile carriers and says it will run as a “carrier of carriers”.
Small cell sites
In practice this means Dense Air will build and run a series of 4G and 5G small cell sites. The aim is to compliment existing networks. It says that in most cases these will extend existing networks in places that need denser coverage. This might be places such as shopping malls, office parks, campuses or sports stadiums. Dense Air says its small cell approach can dramatically improve performance and capacity.
That said, Dense Air has more than enough spectrum to compete with all three carriers in New Zealand. Should it choose to do so, it could offer MVNO (mobile virtual network operator) services. This could be of interest to telcos such as Vocus or MyRepublic, both wish to offer mobile services but own neither spectrum nor their own cellular networks.
Higher frequencies means more bandwidth. This can deliver faster data and more connections per square kilometre.
As a rule, higher frequency radio signals travel over shorter distances. Higher frequency sites will be useful in areas of high population density. In some cases they may be only a few dozen metres apart.
Cover every street
When cell sites are a few dozen metres apart, you need a lot of them. They will, in effect, need to go down every street in the country. The antennae don’t need to be as high as today’s cell towers. You can install high frequency cell sites on telephone and power poles or the sides of buildings.
Compared with today’s cell sites each one will cost a lot less to build. The hardware is smaller and less of an eyesore so the planning requirements will be simpler. And there will be some incremental upgrades.
Yet there will be so many new sites that the total cost of a 5G network could be as much as the earlier mobile. It all depends on how far New Zealand carriers intend to push the technology. It’s possible we won’t get the same 5G service as customers in say, Shanghai, Paris or New York.
Fibre is the 5G backhaul answer
Connecting lots of cell sites is tricky. Today’s cell sites often connect back to hubs using fibre connections. This is the best technology.
When Telecom, now Spark, built its XT mobile network it made a big deal of its towers using fibre backhaul. That’s the name engineers give to the practice of getting signals back to major centres.
Fibre backhaul gave the XT network a clear performance edge over Telecom’s rival. At least it did once Telecom ironed out the initial teething troubles.
Carriers don’t have to use fibre for 5G backhaul. In my NZ Herald interview Alex Wang said self-backhaul would be a feature of 5G. That is the towers link to each other in a wireless mesh network to get traffic back to a central hub.
Wireless backhaul is possible, but it limits overall network performance. You need a lot of bandwidth to backhaul thousands of 10 or 20 Gbps data streams.
It needs to be line-of-sight and it often uses higher power signals. Cue the protests and renewed fear of microwave signals causing health problems.
In practice fibre is a better way to handle 5G backhaul. It’s the most practical way to deliver the promised performance.
And that’s where the New Zealand mobile telecommunications industry hits a potential problem. There is already a nationwide fibre network for UFB.
Fibre companies already have fibre running down every urban street. It cost more than $5 billion to build that network.
You could argue that building three more nationwide fibre networks would waste resources.
It would also add a lot to the cost of using a 5G network. Add in the cost of new antennae, site fees and network controllers. It could add up to more investment than carriers spent on earlier mobile generations.
In practice there’s little chance of carriers building three more nationwide fibre networks. In theory the carriers could build a shared network.
There are arguments why this should not happen. For a start it could shut out any new competitors. There’s also a fear that three carriers owning shared mobile infrastructure could become a cartel. That’s also bad for competition and terrible for customers.
You can assume the Commerce Commission wouldn’t sign-off on shared infrastructure unless it is open access and otherwise regulated. The alternative is anti-competitive and would stifle innovation.
One third of a lot of money is still a lot of money
Even if carriers build a shared fibre 5G backhaul network, the cost per carrier would still be one-third of a big sum. It is more money than Vodafone or 2degrees appear to have today. This is before they need to spend on towers, antennae and the other kit needed to run a 5G mobile network.
Spark could raise the money for its share. The company has little debt. But even its investors might baulk at the cost of a nationwide fibre 5G backhaul network.
As we’ve already mentioned, a 5G network may need many more towers than the 4G networks that are in place today. Each site is likely to cost a lot less than the cost of a 4G site. The number of 5G sites needed to blanket cities and towns means the capital expenditure is going to, at least, be on a par with the investment in 4G. In reality it is likely to cost more.
A billion here, a billion there
Carriers don’t like to talk about the cost of building their networks. In round numbers each has spent in the region of NZ$1 billion on mobile network infrastructure.
Sure, that’s a back-of-an-envelope calculation. The exact numbers aren’t important. They have also invested many millions in buying spectrum.
The three carriers’ total capital spend on 4G to date is on a par with the amount needed to build the UFB network. They will also need to find the thick end of billion or so to build the extra sites needed for 5G.
This would be fine if there was a chance of getting customers to pay a premium for 5G mobile. That’s not going to happen. We’ll look closer at the business case for 5G in another post.
The open access model
New Zealand already has a tried and tested model for a separate wholesale layer. It’s called UFB.
The big telcos don’t like that model because by law wholesalers treat them the same as small ISPs. Spark can’t go to, say, Northpower and ask for a special deal “because we’re your most important customer”. That grates with the big carriers.
They also resent the wholesale charges. Remember the copper tax debate? It annoys telcos that the wholesaler gets 40 percent of each customer’s subscription.
Never mind that sum means the wholesalers gets a fair return on their investment. The regulator decides what’s fair.
The Chorus proposal
Which explains why the four big telcos scorned Chorus CEO Kate McKenzie’s proposal. She suggested that Chorus could provide the fibre 5G backhaul. They fear loss of control and they fear having their tickets clipped. The cost per mobile connection for such a service would be tiny. It would be far less than the cost of building a new network.
In reality one or more of the mobile carriers may end up using some Chorus fibre to backhaul. They may also use NorthPower, UFF or Enable resources. What they don’t want is another wholesale network muscling in on their turf.
Yet, it looks like they will end up with either Chorus or a regulated Chorus-like wholesale organisation. Only Spark could go it alone. But it has better capital expenditure options on than overbuilding a fibre network.
Disclaimer: Chorus pays me to edit the Download magazine and a weekly newsletter. It didn’t pay me to write about 5G backhaul. Indeed, this post doesn’t reflect anyone’s opinion other than my own. No one vetted or otherwise approved this. Any mistakes are down to me. Your corrections or alternative opinions are welcome.
Telecommunications Commissioner Dr Stephen Gale says:
“We believe the power to regulate remains an important competition safeguard, especially with 5G networks and potential new entrants on the horizon”.
Money go round
In the past government spectrum auctions work by dividing available frequencies into blocks. Bigger blocks give carriers more bandwidth to play with. In simple terms more bandwidth can mean faster data speeds.
Spectrum auctions can make a lot of money for governments. Past auctions have poured gold into the public sector. The recent UK 5G spectrum raised £1.3 billion, around NZ$2.5 billion.
It may look like a windfall. Governments often treat the money that way. But it is more about moving money from one place to another. When telcos pay a lot for spectrum the cost is passed onto customers.
If they overpay, they may spend money that would otherwise be used to build towers and extend the network’s reach. Overpaying often means a network roll-out is slower.
Given the value of cellular communications to the wider economy, squeezing out the maximum amount of cash in a spectrum auction can be counterproductive in the long term.
New Zealand’s last spectrum auction took a more sensible approach.
The government realised the economy could be better served in the long term by a good mobile network than by a windfall. So carriers were offered a fixed price well below what it might have made in a competitive auction.
Not everything sold so one remaining block of spectrum was then auctioned off.
In the past different cellular services have run in different frequency bands.
This can still happen. Yet one of the features of 5G is that carriers are able to mash together greater amounts of bandwidth from different bands. Or to use an engineer’s language: they can aggregate spectrum.
While this already happens a little with 4G, Spectrum aggregation is central to 5G. How that works in practice will be interesting. It will be a challenge for phone makers.
Most people in the telecoms business expect 5G to use higher frequencies than today’s mobile phones. Depending on who you talk to, the options go all the way up to 95GHz.
This brings us to another challenge carriers face. Radio waves have different properties in different bands.
Low frequencies are useful for communicating with submarines or in mines. Shortwave radio is good for broadcasting over long distances. And so on.
Dealing with this is an engineering problem. There are also political challenges. In some cases existing spectrum users may have to give up their rights or move services to different frequencies. It can be disruptive.
Compared with some other countries, New Zealand is well placed to deal with these challenges.
UHF – ultra-high frequency
Almost all of today’s mobile telephone traffic takes place in what is known as the ultra high-frequency band or UHF. This is the spectrum from 300 MHz to 3GHz.
Some of the spectrum that will be used for 5G is in the next band up: super high frequency or SHF. That runs from 3 to 30 GHz.
UHF and SHF frequencies are microwaves. Which means the band is used by microwave ovens. It’s also used by Wi-Fi and other home wireless devices, satellite communications, radar and radio astronomy.
As you move into higher spectrum bands radio signals run into a different set of physical problems. At 5GHz and above signals get absorbed by solid objects.
The signals don’t propagate so well. So antennae cover shorter distances. In other words, you need to build more towers to give carpet coverage.
Bluetooth devices operate in part of this frequency band.
The devices have low signal power levels compared with cellular phones. They are only designed to work over a short distance.
Even so, you a taste of what to expect from a 5G cell site operating at this frequency by thinking about Bluetooth’s limitations around your house. The signals may pass through wooden walls, masonry can block them. So can metal frames.
When outdoors, microwave signals don’t tend to pass through mountains or hills. In effect, they only work in line-of-sight. A cell site operating at higher microwave frequencies that works for a customer in winter might struggle in summer when there are leaves on the trees.
Go beyond 30GHz and radio signals are affected by water molecules. That means rain — satellite TV users will already know about rain fade. From about 60GHz oxygen molecules get in the way.
This tells you something about the risks, although the power used for cellular phones would be many times lower than any weapon.
To keep things simple, let’s leave it at this: higher frequency radio waves are harder to use. On the other hand, they offer much more bandwidth and that means higher potential data speeds.
As a rough rule of thumb, higher frequencies mean faster data, but over shorter distances. Typically higher frequency sites will be in densely populated areas and will be only a few dozen metres apart.
When cell sites are a few dozen metres apart, you need a lot of them. They don’t need to be big. You could put them on existing telephone or power poles.
In New Zealand
For now, talk of higher frequencies and the problems using them is largely academic. Most of the planned 5G action here in New Zealand is in or around frequency bands already used by mobile phones.
When Spark managing director Simon Moutter outlined his companies plans he called for more spectrum below 1 GHz.
He says it will be needed to provide 5G services in rural areas. This will almost certainly mean the 600 MHz band, which is already in the government’s sights. Signals in this frequency band can travel over long distances.
Moutter also identified the “two most likely spectrum bands”. Spark wants the mid-frequency C-band and high-frequency mmWave band to be ready as soon as possible so it can put its 5G network in place in time for the 2020-21 America’s Cup in Auckland.
This shouldn’t be difficult in principle.
Is there enough for 5G?
There should be enough usable spectrum in the 600 MHz band and the C-band to give New Zealand’s three big mobile carriers all they need to build viable 5G networks.
Yet they are not the only possible bidders for 5G spectrum. Wisps — wireless internet service providers — do a fine job filling in the gaps in regional broadband coverage.
Wisps could also make good use of more spectrum. And the spectrum of most use to them happens to be the spectrum the carriers are keenest to buy.
Small regional service providers lack the financial clout of the mobile carriers, but they can argue the service they offer is as deserving. Maybe more, after all, wisps service New Zealand’s exporters.
Economic logic says a competitive auction is a way of ensuring spectrum goes to the bidder who stands to gain the most. This, the argument goes, means the most economically efficient use is made of each block of spectrum.
In practice, some bidders sit on unused spectrum. The last NZ auction made that unlikely as it included a use-it-or-lose-it clause.
Some less well-heeled organisations find it hard to buy the spectrum they need. How these issues will be addressed will become clearer when the auction terms are formally announced.
The confirmation comes after the company conducted trials earlier this year. Spark says the Wellington outdoor trial was a success with customers getting download speeds of up to 9 Gbps. An indoor trial in Auckland saw speeds as high as 18.2 Gbps.
While some telcos overseas are building new networks from scratch, Spark says it will start by adding 5G services to its existing 4G and 4.5G networks.
Spark says it will extend this when there is enough demand.
With existing cell sites there’s a smooth upgrade path. At least there is if a carrier sticks with the same equipment supplier.
Spark managing director Simon Moutter says the company is working on mapping expected cell site densities to learn where there is a need for new cell sites.
He says: “We have already begun a build program to increase the number of cell sites in our existing mobile network – which will enable us to meet near-term capacity demand as well as lay the groundwork for network densification required for 5G.”
No extra CapEx
The company says it is expects to fund its network through its existing capital expenditure programme. This does not include buying any extra spectrum needed for 5G.
Spark spends around 11 to 12 percent of its revenue on capital expenditure. Spark’s 5G briefing paper says:
As Spark responds to demand we will be investing just ahead of it. Cost efficiency that will deliver ever-greater output with the same investment inputs is the primary driver of early 5G deployment.
By 2020, we expect our wireless-network specific capex to be between 25-35 percent of Spark’s overall capex envelope. This implies intended annual wireless network investment of approximately $100m to $140m, compared with an average of just over $100m for the past five years.
This excludes spectrum purchases and any material move towards widespread rollout of new cell sites using mmWave band spectrum. During this period, we expect our total capex (excluding spectrum) will remain in line with our desired range of 11 to 12 percent of revenues.
This is something of a surprise.
5G network equipment tends to be less expensive than 4G hardware. But to deliver the next generation network’s full promise, a carrier needs more spectrum and at higher frequencies it will need more small towers.
Many of these towers will be smaller than existing 4G towers – in some cases they can fit on lamp posts or telegraph poles, but even so, Spark’s comment about capital expenditure suggests one of two possibilities.
It won’t happen overnight
The first possibility is that Spark’s network roll out will be incremental and relatively slow. This follows the pattern of the company’s roll-out of 4.5G.
It is two years since Spark first installed a 4.5G tower in the centre of Christchurch. There are more today, but coverage is far from nationwide.
It looks likely the 5G roll out will begin before Spark has upgraded every worthwhile cell site to 4.5G. Presumably many sites will go straight from 4G to 5G.
The second possibility is that Spark isn’t aiming for the same high density network being planned for large urban centres elsewhere in the world. At least not at first.
Neither of these are important in the short-term.
Indeed, today’s mobile phone users can’t tell the difference between using a 4.5G tower and a 4G tower. There’s no pressing need to upgrade the network on their behalf.
And places like Eden Park in a test match aside, New Zealand doesn’t have the density of people you might find in Hong Kong or New York.
Spark may want to push forward on plans to offer 5G-driven fixed wireless broadband as an alternative to fibre. It already does this with 4G. This is a strategic business decision. If there’s enough demand for more fixed wireless then the internal business case for increased capital expenditure is easy to make.
5G innovation lab
Spark plans to open a 5G Innovation Lab later this year in Auckland’s Wynyard Quarter. This will let companies test their applications on a private 5G network before the full roll-out.
The company says:
“Providing early access to a pre-commercial 5G network through our global relationships with leading equipment vendors like Huawei, Cisco and Nokia will give our local partners a competitive boost, fast-tracking these businesses’ 5G developments.”
Significantly Spark has not named the network equipment provider it will work with on the programme.
The company used Huawei to build the 4G network and has previously worked on 4.5G and its test site with the Chinese equipment maker. Huawei has to be in consideration for the contract despite the political problems the company faces getting business in the US and Australia.
Yet Spark deliberately named Nokia and, surprisingly, Cisco. The latter is not known as a technology provider for cellular networks. This could be a way of putting pressure on Huawei in order to get a better deal.
Spectrum is a potential concern.
In a briefing paper Spark called on the government to make more spectrum available. All the carriers are pushing hard. They have a case.
This is already in motion, but the company wants this done in time for the new network to be running ready for the 2021 America’s Cup in Auckland. Hence the earlier comment about the need to get this wrapped up in the next 18 months or so.
Spark says it needs large blocks off spectrum in the C-Band, that’s 3400 to 4200 MHz. It says it needs at least 80 MHz blocks and preferably 100 MHz blocks to build networks with 5G performance. It also calls for even larger blocks at higher frequencies.
In 2015 I travelled to Shenzhen in China to learn more about 5G mobile technology at Huawei’s headquarters.
Huawei’s brand is best known in New Zealand for phone handsets. That is only part of the company’s story. Huawei is also the world’s largest telecoms-equipment-maker and a world-scale economic powerhouse.
Spark New Zealand and 2degrees use Huawei kit to power their cellular networks. Moreover, Huawei is leading the charge towards next generation mobile networks.
5G was always going to happen
Everyone in the phone business always knew there would be a generation to follow 4G. Cellular technology is far from done.
Yet at the time of my visit 5G was still a new idea only starting to take shape. In 2015 many telcos around the world were still finishing their 4G networks.
The hype machine hadn’t kicked in and technologists were still batting ideas around.
Some concepts were just that: concepts.
Huawei’s 5G perspective
During the visit I got my first comprehensive overview of the Huawei’s perspective on the technology from Alex Wang, the company’s VP of wireless marketing.
This is from the NZ Herald story I wrote about the trip:
“Dealing with more connections is one reason telecommunications companies need 5G. Wang says the formal definition of 5G has yet to be agreed, but one of the items of the list is for it to support massive connectivity.
The goal is for cell sites able to cope with one million connections in a square kilometre — effectively that means one mobile device per square metre. By comparison today’s 4G cell sites might handle 1000 to 3000 devices.”
Wang also said the goal was to get latency down to 1 ms and to support data speeds of up to 10 Gbps. This second goal has since been changed to 20 Gbps. Most of the other numbers remain as planned in 2015.
Phones at the speed of light
As any physics student will be able to tell you, light, or radio waves, travels through a vacuum at about 300 kilometres in a millisecond. The speed through air is not much different.
The original 5G target speed of 10 Gbps is ten times the speed of today’s fastest home fibre connections. The newer 20 Gbps target is twenty times faster.
Without getting deeper into electromagnetic physics or engineering, these goals are ambitious.
You can’t push wireless data that fast with the existing mobile radio spectrum. There isn’t enough free bandwidth for three carriers to hit these targets in densely populated areas.
More spectrum needed
Which means carriers need to find new spectrum if they are to deliver 5G. Or, more to the point, government’s have to reorganise existing spectrum allocations. In most cases they then sell it to carriers in an auction.
Contrast this with the UK where bidder spent £1.4 billion buying 5G spectrum. This was more than twice the anticipated cost. The winning bidders spent money they could have used for the capital expense of building a network. It’s likely to mean a slower build and higher costs for users.
In the past different services have run in different frequency bands.
One of the features of 5G is that carriers will be able to mash together greater amounts of bandwidth from different bands. Or to use their language: aggregate spectrum.
This already happens a little with 4G. Spectrum aggregation is central to 5G. Aggregation opens the door to merging what now may seem like different technologies, in particular cellular and wi-fi. How that works in practice will be interesting.
In the next post on 5G we’ll look more at the spectrum issue.