There are disputes about almost every aspect of 5G, but one area of agreement – the cost of spectrum will have to go down to make most 5G business cases viable. Shared and unlicensed spectrum will play a far greater role than before, because while these are a double-edged sword for established MNOs, their need to keep costs under control will outweigh the threat of lowering barriers to entry for new rivals. Long before most will think about 5G, regulators and operators are already accelerating their work on new spectrum strategies to improve the economics of 4G and mobile data.
At the recent Mobile World Congress, attention was certainly shifting from the whizzbang technologies and futuristic 5G use cases of previous years, towards the actual economics of deploying LTE-Advanced, LTE-A Pro and 5G. Timotheus Höttges, CEO of Deutsche Telekom, got considerable attention when he cited calculations which suggest the cost of deploying 5G in high frequency spectrum across Europe (or the US) could amount to between €300bn ($318bn) and €500bn ($530bn).
Concentrating on sub-6 GHz bands, which require less dense infrastructure, will reduce that bill, and outside a few enthusiasts for millimeter wave fixed wireless, this will be the main area of investment in the early 5G years. But operators are still looking at very dense, high capacity zones of 5G to enable new revenue streams, so as Höttges begged regulators, spectrum needs to be priced at “reasonable rates”. In the 3G auctions, German operators paid $46.2bn for their spectrum licences, mainly in 2.1 GHz, while in the 4G auctions, they paid €4.4bn ($4.6bn) in 2010 and another €5.1bn ($5.4bn) in 2015. But even at that lower rate, DT’s spectrum bill in 2015 was equivalent to 20% of its annual EBTIDA in Germany.
Regulators need to think hard about whether auctions are the most appropriate way to allocate 5G spectrum, and whether spectrum should be regarded as a short term revenue boost for the Treasury, or the enabler of a longer term, but far bigger, boost to the whole economy via enhanced mobile broadband access.
Most regulators are still at the stage of issuing experimental licences and assessing 5G radio behaviour as well as socio-economic impacts. But some are discussing auctions already, and in many cases, in worryingly conventional ways. Some, like India’s TRAI, have leftover spectrum from 4G processes which they are now labelling 5G; the same can be said for Europe’s ‘second digital dividend’ in 700 MHz, or the US’s 600 MHz incentive auction. These were old-fashioned auctions which were not considered ‘5G’, but the timing means that many operators will use these airwaves for next generation networks.
The relatively low prices paid in recent sales, however, reflects the fact that MNOs have other choices rather than pay sky-high fees for their own long term licences. That system gives them control and a powerful weapon against rivals in unlicensed bands, but the costs are becoming unsustainable. MNO reluctance to take part in auctions, because they have other options, such as shared spectrum, will force regulators to think more creatively about how to allocate the next wave of airwaves to be opened up.
Amit Nagpal, a partner at spectrum advisory group Aetha Consulting, believes this will be one factor in reducing the amount operators pay for 5G spectrum, as will the relatively large amounts of spectrum available in the higher bands. “There is a fair amount of spectrum at 3.5 GHz and below that can be used for 5G – including by refarming spectrum used for 2G, 3G and eventually 4G services – and so I don’t see any one particular band raising massive amounts of money,” he said. “There may be a small premium for 5G pioneer bands since the first 5G handsets will initially be designed to work in a limited set of bands. However, I expect this won’t last for long and that within two or three years handsets will support 5G in all the existing 4G bands.”
Another important factor will be the development of easier ways to refarm spectrum, or to use the same bands flexibly for different technologies. Huawei’s CloudAIR is a particularly advanced example of a system which allows a band to be shared between two generations of radio technology in a flexible way depending on the requirements of users. There will not be the same clear line between ‘4G bands’ and ‘5G bands’ as there currently is between 3G and 4G. “Operators will buy kit that can largely be used for 4G and then gradually move to 5G when devices come into the market,” Bengt Nordström, CEO of the Northstream consulting group, told LightReading. “It is realistic to expect that a lot more 4G will happen over the next five years.”
In both 4G and 5G, spectrum sharing will help to reduce costs and improve quality of experience. MNOs are already looking to aggregate shared bands, such as the US’s 3.5 GHz CBRS spectrum, with their conventional licensed airwaves, to increase capacity; though of course, technologies like MulteFire will also enable new entrants, such as cablecos or web providers, to deploy LTE and future 5G themselves, without the need to buy any licences.
“This is the first time we’ve had a band that anybody can go into and they can put any kind of technology they want,” said Preston Marshall, principal architect at Google Access. “The technologies will be competing in this band, the different applications will compete in this band. That competition is going to create a very, very rich ecosystem for innovation, for new ideas. It could never happen if people spent $30bn or $20bn buying spectrum.”
For now, the technologies to aggregate unlicensed and licensed spectrum remain reliant on a licensed host network. LTE-Unlicensed and LTE-LAA are both hovering on the edge of commercial deployment, supporting supplemental downlink in the 5 GHz unlicensed band to augment the capacity of a conventional 4G network.
AT&T’s VP of RAN and device design, Gordon Mansfield, said at MWC that the operator would start to sell LAA-supporting devices “imminently” and fully launch LAA later this year. These will be the first devices to incorporate the 5 GHz Band 46 and once AT&T starts to switch on updates on the infrastructure side, it will be pushing for gigabit speeds using four-way carrier aggregation (its three LTE bands plus LAA).
“With the spectrum positions that the operators have in the US, you’ve got to bring together some unlicensed spectrum as well (to hit gigabit speeds),” Mansfield said. “You can get close, you can get up to several hundred with just the macro network and the licensed spectrum, but the way you get to a gigabit is you use licensed spectrum with 4×4 MIMO and 256-AM and you bring in LAA with unlicensed carriers. And in that case it will actually be 4CA.”
This focus on LAA is in contrast to the decision by Verizon and T-Mobile to move first with LTE-U, following the FCC’s recent decision to authorize this technology in 5 GHz, despite long-running wrangles with the WiFi community over possible interference. LTE-U may prove a shortlived system, since it can only work in the US and a few other countries which do not insist on listen-before-talk (LBT) to avoid interference in the 5 GHz spectrum. LTE-LAA, part of the latest 3GPP standards release, addresses this issue and has greater potential to be globally implemented, and so supported in a wide base of devices.
Verizon said recently that it would start rolling out LTE-U this spring, calling it “an example of yet another great innovation using unlicensed spectrum”. T-Mobile also plans to launch LTE-U from this spring using equipment from Ericsson and Nokia. “T-Mobile already has more capacity per subscriber than AT&T and Verizon, and the addition of LTE-U will only extend that lead and further improve the Uncarrier’s blazing-fast speeds,” the company said in a statement.
New FCC chairman Ajit Pai said of LTE-U: “The excellent staff of the FCC’s Office of Engineering and Technology has certified that the LTE-U devices being approved today are in compliance with FCC rules. And voluntary industry testing has demonstrated that both these devices and WiFi operations can co-exist in the 5 GHz band. This heralds a technical breakthrough in the many shared uses of this spectrum.”
Then there is the CBRS band, whose innovative, if rather complex, three-tiered allocation system is setting important precedents for how spectrum can be shared between incumbents and MNOs, and between different wireless providers. All four national MNOs have joined the CBRS Alliance, which is working to define standards to drive interoperability and a broad base of certified devices, both essential to commercial roll-outs.
The Alliance says that it is seeing such momentum that it expects the first commercial LTE deployments in the shared portion of the 3.5 GHz band to take place next year. It expects the FCC to start certifying devices in mid-2017, while there will be network tests in the second half of this year, leveraging new Spectrum Access System (SAS) databases. These are the method by which interference will be avoided in the CBRS spectrum, and build on previous experience in the unlicensed TV white spaces (TVWS) airwaves.
Indeed, the WhiteSpace Alliance (WSA) believes that the completion of the 600 MHz incentive auction in the US will accelerate deployments in the white spaces within that broadcasting band. “The completion of the incentive auction will remove regulatory uncertainty and free up more than 80 MHz of spectrum for licence-exempt wireless Internet access,” said Apurva Mody, chairman of the Alliance. “We have seen innovative applications in medical, military and home markets from use of unlicensed spectrum in 2.5 GHz ISM bands. This newly available spectrum will support additional digital infrastructure development across the United States, and help foster the Industrial IoT.”
While several organizations applied to build SAS systems for CBRS, only two have announced actual platforms – Google and Federated Wireless.
The 3.5 GHz band is split into three tiers of access – priority access for federal incumbents; licensed access; and shared access. Auctions for the licences are likely to start in 2018.
Paul Challoner, VP of network product solutions at Ericsson, said the creation of a shared spectrum framework in 150 MHz of the band will usher in a wide variety of providers. Large cable operators, which are working on various ways to add wireless services to their offerings, are prominent in the CBRS Alliance alongside the MNOs, with Comcast and Charter Communications both discussing the potential to build their own subnets of 3.5 GHz small cells to support their core services, relying on MVNO deals only for wide area roaming.
The enabling technologies and standards still need work to support a fully commercial platform, but progress is accelerating. Last month, Google Access announced the first end-to-end CBRS demonstration with 3.5 GHz mobile devices; and a new Trusted Tester program to help vendors test their products with Google’s SAS.
And Nokia is partnering with Federated Wireless on an LTE solution for CBRS, while Ericsson has already tested its Radio System Architecture with Federated Wireless’ SAS. Federated has also been doing trials and demos with other companies, including Ruckus Wireless, SpiderCloud, Siemens, Telrad and Airspan, among others. Its spectrum controller recently received conditional certification from the FCC, and is likely to achieve full approval before mid-year, and commercial customers by the end of 2017. In addition, Google and Federated have cooperated to ensure their SASs are interoperable.
The cellular industry will learn many lessons from WiFi as it places shared and unlicensed spectrum at the heart of its agenda. And some techniques which have been pioneered for WiFi will now find that they are relevant in the LTE world too, opening up new opportunities.
One of these is BandwidthX, which developed a cloud-based marketplace system to match service providers with available WiFi capacity on an on-demand, pay-per-use basis. Now the company will move its BxMarket offering into the cellular market too, allowing MNOs to define their value for their mobile capacity, when this is in shared or open spectrum, and automatically trade across ownership boundaries.
“With the ongoing expansion of mobile data network types, spectrum licensing models, and diversification of network ownership, the benefits of BxMarket’s functionality, including real time policy controls and automated connections, have become even more significant in the mobile data ecosystem,” said the company.
Systems like this will increasingly be driven by machine learning (see separate item) and will evolve to support fully flexible, on-demand allocation of bandwidth, with approaches like network slicing. To maintain quality of service, and service differentiation, there will also need to be significant enhancements to current optimization and network management systems. As Gabriel Brown, principal analyst at HeavyReading, wrote recently: “For shared spectrum to be most effective, it is vital to have effective and reliable mechanisms to optimize sharing. Tightly coordinated systems perform more efficiently under load than contention-based shared access. Co-ordination, however, requires information sharing between operators, which in turn requires some form of standardization. There are a number of ways to do this from the radio interface itself through to centralized database systems, or combinations of both.”
Vodafone CTO damps down 5G hype:
The latest mobile bighitter to try to manage 5G expectations in the face of a wave of hype is Johan Wibergh, CTO of Vodafone Group. “I’m worried that people think 5G is going to be gigabit speeds for everyone and solving all problems including world hunger,” he told an audience at Mobile World Congress.
And he told Mobile Europe: “Sometimes when you talk about something it may not be possible to do it at the beginning.”
Adrian Scrase, CTO of ETSI, added: “5G is not going to be a wow factor more than you have in an advanced LTE network. High performance and low latency will probably be achievable halfway through the lifecycle. You’re not going to reach the extremes on day one.”