Spectrum is like land in the world of real estate: everyone wants it, but nobody can make more of it. There is growing demand from IoT and high-bandwidth-intensive services such as video streaming, while even low-bandwidth services such as two-way radio struggle to find the spectrum they need for new business in urban areas.
To complicate the situation further, Dean Bubley, founder and director of Disruptive Analytics, says: “There’s a general recognition among the regulators that it’s going to be difficult to clear existing bands of incumbent users.” He therefore believes that “there’s going to be relatively less clearance and reuse and more varying forms of sharing and coexistence”.
Another factor is the way that the cellular industry is looking to millimetre wave technologies to deliver high bandwidth and use previously relatively untouched areas of the radio spectrum. However, given spectrum above 3GHz’s poor in-building penetration, the high probability that the bulk of IoT devices will be present in indoor environments and the way in which most users spend the majority of their lives indoors, Kalpak Gude, president of the Dynamic Spectrum Alliance (DSA), asks if traditional spectrum models make sense.
“Think about the unnecessary friction [they] create. The only entity that has the right to deploy in a building is the building owner. The licensee, if you go down the path of exclusive licensing, holds the spectrum [but cannot enter] the building without the building owner’s permission, so what was the value of giving the right to that spectrum to a third party? Let the building owner have that right, he has the incentive and the capabilities to go build.
“You reduce and eliminate the friction of creating some sort of rights deal between the two if you can have a neutral host that allows building owners to provide LTE-based connectivity and through striking a deal with MNOs allow seamless connectivity and easy billing, and in the absence of such a deal allow the building owner to still monetarise the connectivity, via their tenants, guests or just through increasing the value of their property.”
He also says that MNOs don’t have the capacity to negotiate with every building owner.
Gude adds: “What we’ve seen is regulators looking at the past model of allocation and licensing and projecting forward as if the spectrum [and its propagation characteristics are] the same. [That’s only the case] if the fundamentals are the same, but the fundamentals are not the same.”
One potential issue with Gude’s suggestion is that it would be hard to see how mobile operators would pay as much for spectrum if it excluded in-building use. He believes the amount of money that governments earn from mobile spectrum auctions does have “a significant impact on the way that governments approach [spectrum management]”, but while they want to monetarise as much as possible, they are looking at the overall economic impact and their citizens’ need for connectivity.
Gude adds: “Carriers’ ability to monetarise spectrum in the same way they have in the past is changing. In the US, they’ve moved to unlimited plans largely across the board that fundamentally disconnect CAPEX investment from revenue. That has a real long-term impact on their ability and willingness to continue to pay billions of dollars for spectrum to do that.”
He notes the relatively low level of participation in the recent 600MHz auction in the US. “Verizon didn’t play at all. Sprint for a variety of reasons also didn’t, while AT&T played on the margins. Auctioning the spectrum is not going to have the same kind of return as it has had in the past because [carriers’] ability to monetarise that [spectrum is in decline].”
Databases to the rescue
At present, there is something of a holy trinity when it comes to the efforts to make greater use of shared and unlicensed spectrum over LTE: the Citizens Broadband Radio Service (CBRS) in the US, Licensed Assisted Access (LAA) and MulteFire. As LAA is essentially a way for carriers to leverage their licensed spectrum with additional unlicensed spectrum, I’ll focus on CBRS and MulteFire.
CBRS uses a three-tiered approach to managing 150MHz of spectrum in the 3.5GHz band, with different tiers for incumbents, Priority Access License users and General Authorized Access users. It uses a database in concert with a Spectrum Access System to manage the spectrum in a dynamic way, to allow the most efficient use of the spectrum.
“Unlike some of the other approaches to sharing, CBRS is unusual in that it seems to appeal to pretty much every constituency,” says Disruptive Analytics’ Bubley, as it has something for everyone. “The US military seems happy that the people keep out of its way, it gives the carriers access to a slab of cheaper spectrum for use in certain places and the new service providers have access to geographically limited areas. The general access tier gives innovators the potential to do cool new stuff.
“Everyone seems to be quite pleased with CBRS, all the vendors seem to be aligned, whereas some of the other approaches to sharing have tended to be a little bit more antagonistic to one or another group – so [CBRS is] a good model for the future.”
Gude expects that CBRS roll-out will take place over the next 12 months and he says that while it is currently a US regulatory structure, there is some discussion in Europe regarding the 3.6-3.8GHz and 3.8-4.2GHz bands. Given the incumbents that are present in these bands, Gude sees great value in the way in which it takes care of incumbency rights and allows new users to operate around them. The three-tier sharing approach is one that can be applied to a variety of spectrum bands.
However, Jaime Fink, CTO and co-founder of Mimosa Networks, says he is “fairly negative about CBRS in the long term, it’s a very small amount of spectrum that’s going to be shared across a lot of different [users]. 150MHz sounds great, but when you look at [it in the] context of delivering services, [there] is not enough spectrum to move the needle on 100Mbps-plus and gigabit services. So, we desperately need more spectrum, which is why we’ve been very active in asking to open up another 500MHz directly above it.”
Bubley strongly disagrees with this, given that CBRS is likely to be used with indoor small cells, and adds: “There’s an awful lot that’s overstated with this idea of gigabit broadband, that’s a headline for 5G but it is mostly irrelevant in my view.”
Moving away from CBRS to the general advantages of a database-based approach, Gude believes it could solve some of the issues around trying to share spectrum between different technologies and usage types. “Ultimately, those can be solved, because you just need to know the protection criteria that the individual rights holders require. Those can be uploaded into a database and, [as] the channel assignments are happening dynamically, if the database understands that a service requires a greater guard band, a greater protection level, it can assign channels around it accordingly.”
Bubley wonders if a blockchain-based approach might be suitable for managing spectrum allocations that are made on relatively long timescales.
Firing up
Turning to MulteFire, which is part of LTE and, according to Bubley, will also have an equivalent in 5G, “where it differs from LTE-U and LAA flavours of 4G is that it doesn’t require SIM authentication or a licensed spectrum anchor, so in theory, while it’s probably not going to be as easy as Wi-Fi, anyone can run a MulteFire network with enough technical expertise”.
Bubley expects this to lead to “some interesting hybrid approaches” and “a greater diversity of service provider types”. He suspects that “a lot of people are going to underestimate how complicated it is. There will probably be a lot of scope for intermediate specialists: private network roaming specialists or integrators or installers, people who do the RF planning for MulteFire or CBRS; there’s a plethora of virtualised core network providers who are doing various flavours of vEPC.”
One of these is Core Network Dynamics (CND), a mobile core network start-up, which joined the MulteFire Alliance in November last year.
“MulteFire is gaining real traction,” says its CEO, Carsten Brinkschulte. “[We’ve just been] contacted by one of the world’s largest network infrastructure suppliers for the enterprise worldwide and they are interested in CND’s OpenEPC precisely because of MulteFire.
The market is still building but it has broken through the barrier of just standardisation and talk, it’s about building business, building products and starting to sell it over the next two or three years. By the end of this year, we’ll have the first chipsets by Qualcomm shipping to market with MulteFire support and it won’t take long until you have the first USB sticks that you can plug into a laptop… or custom-designed devices to communicate over MulteFire. And, of course, the radio vendors, like the base station and the small-cell vendors, are getting ready.”
Brinkschulte adds: “The main use-case [for MulteFire] will most likely be IoT and industrial IoT to connect sensors and actuators, which is good as it [reduces the importance of] handset availability – otherwise private LTE would be dependent on Samsung or Apple adding support for MulteFire to their handsets.”
Slicing the pie
This focus could be argued to put MulteFire in competition with 5G, particularly given the latter’s network slicing capabilities, which at first glance combined with licensed spectrum would appear to be a difficult combination to beat. However, Bubley says that in his conversations with MNOs: “Most will admit that while the headline is that 4G and 5G can pretty much do everything, when you speak to them privately, there’s a bunch of stuff they recognise that they probably cannot or will not be able to do, certainly on their own.
“So, you’ll probably find that a lot of the national operators will pick and choose which of the industrial sectors they focus on – they all recognise that there are gaps that will need to be filled either by private owners of networks or new classes of service provider.
“Network slicing and 5G more generally are massively overhyped. Both will be important, but there’s a huge amount of hope above reality across the board.” Bubley expects that pure 5G networks won’t arrive for probably another decade, with initial 5G deployments existing as part of a hybrid 5G/4G/3G network. “It’s entirely unclear what slicing does when you go across those boundaries,” he says.
Bubley adds that it is also unclear whether 5G networks will be able to be sliced to the level of granularity that people require. He gives the example of self-driving cars, where there might be 10 applications, each having very different network quality requirements, “so you would end up needing 10 slices for a vehicle – it makes no particular sense to group them all under the slice called ‘car’ when they’re all very different”.
He has yet to see an Amazon Web Services-style interface for developers that would allow users to create their own network slices – “it’s still going to be very much handcrafted in the same way that MVNO deals are today” – and Bubley suspects that the initial uses for network slicing will be to serve the different business units within large MNOs. Another issue Bubley sees with network slicing is the potential headaches if it involves having to negotiate different deals “with all the operators around the world who will all have slightly different mechanisms, pricing and capabilities”.
Gude also highlights the cost of 5G networks, citing the CTIA’s estimate that it would cost $275bn for a US-wide 5G network. He believes the MNOs do not have the ability to invest on that scale, given the question marks that still hang over the business case for 5G.
Bubley also notes the current business model in which enterprises’ local area networks aren’t run as a service and don’t involve a subscriber relationship with an MNO. This, together with businesses’ current use of Wi-Fi, implies there may be resistance to the 5G/operator approach. “There is going to be a desire for short-range, high-reliability, on-site, on-premise connectivity in licensed spectrum that is not run by a national operator,” he adds.
From Gude’s perspective, if there is a question around the use of shared spectrum and the new ways of licensing spectrum, “it’s how the business models are going to play out. For example, in IoT, are the factory and building owners, the hotels and stadiums, going to jump in and go do this? We think so. But until they do, regulators are holding back and saying, ‘Well, wait a second, we want to see what has happened before we start exporting this regulatory structure to other bands.’ Once this does move forward, it’s going to be [difficult not to] use a similar model elsewhere because the economics and the efficiencies [of shared spectrum] are going to take this to a different [level].
