Speaking Engagements & Private Workshops - Get Dean Bubley to present or chair your event

Need an experienced, provocative & influential telecoms keynote speaker, moderator/chair or workshop facilitator?
To see recent presentations, and discuss Dean Bubley's appearance at a specific event, click here

Monday, January 07, 2019

Private cellular networks - why Ofcom's UK spectrum proposals are so innovative

On December 18th 2018, Ofcom announced two consultations about new 5G-oriented spectrum releases (link), and potential new models for spectrum-sharing, rural mobile coverage and related innovation (link). 

I've already commented briefly on Twitter (link) and LinkedIn (link), but it's worth going a bit deeper in a full post on this - particularly on the aspects relating to private networks and spectrum-sharing.

NOTE: this is a long post. Get a coffee now. Or listen to my audio commentary (Part 1 on the background to private mobile networks is here and Part 2 on the Ofcom proposals is here)

My view is that 2019 is a key breakout year for new mobile network ownership and business models - whether that's fully-private enterprise networks, various types of neutral-host, or a revitalised version of MVNO-type wholesale perhaps enriched by network-slicing. 

This trend touches everything from IoT to 5G verticals, to enterprise voice/comms & UCaaS. I'll be covering it in depth. I also discussed it when I presented to Ofcom's technology team in November (see slides halfway down this page), and it's good to see my thinking seems to align fairly closely with theirs.

This was the future, long ago

Localised or private cellular networks - sometimes called Micro-MNOs - are not a new concept.

Twelve years ago, in 2006, the UK telecoms regulator Ofcom made an unusual decision - to auction off a couple of small slices* of 2G mobile spectrum, for use on a low-power, localised basis for a number of innovative service providers or private companies' use. (Link). A few launches occurred, and the Dutch regulator later did something similar, but it didn't really herald a sudden flourishing of private mobile networks. 

*(The slices were known as the DECT Guard Bands, which separated GSM mobile bands from those used for older cordless phones, widely used in homes and businesses)

Numerous practical glitches were blamed, including the costs / complexities involved in deploying small-cells, the need for roaming or MVNO deals for wide-area coverage, and the fact that the spectrum was mostly suitable for voice calls, at a time when the world was moving to mobile data and smartphones. 

Unfortunately, there was also no real international momentum or consensus on the concept, despite Ofcom's hope to set a trend - although it did catalyse a good UK-based cottage industry of small-cell and niche core-network vendors.

Going mainstream: private / virtualised networks for enterprise & verticals

At the start of 2019, the world looks very different. There is a broad consensus that new models of mobile network are needed - whether that is fully-owned private cellular, more-sophisticated MVNOs with their own core networks, or future visions of 5G with privately-run "network slices". 

There's a focus on neutral-host networks for in-building coverage, proponents of wholesale national "open" networks, and a growing number of large non-telecoms enterprises wanting more control and ownership.

It is unrealistic to expect the main national MNOs to be able to pay for, deploy, customise, integrate and operate networks for every industry vertical, indoor location or remote area. They have constraints on capital, personnel, management resource, specialised knowledge and appetite for risk. Other types of network operator or service provider are needed as well.

In a nutshell, there is a wide recognition that "telecoms is too important to just leave up to the telcos".

I've been talking about this for several years now - the rise of unlicensed cellular technologies such as MulteFire or Huawei's eLTE, the growing focus on locally-licensed or shared spectrum for IoT or industry use, and the specific demands of rural, indoor or industrial network coverage and business models.

(As well as non-MNO deployed and owned 4G/5G networks, we will also see a broad range of other ways to deliver private capabilities, including various evolutions of MVNO, mobile SD-WAN and future network-slicing and private cores. But this particular consultation is more about the radio-centric innovations).

Where is the action?

But while there has been a lot of discussion in the UK (including my own presentations to Ofcom, the Spectrum Policy Forum and others), the main sources of action on private (licensed) cellular have been elsewhere. 

In particular, the US push on its CBRS 3-tier model of network sharing - expected to yield the first local service launches in 2019 - and German and Dutch approaches to local-licensed spectrum for industry, have been notable. Unlicensed cellular adoption is (fairly quietly) emerging in Japan and China as well.

Plenty of other trials and regulatory maneouvring has occurred elsewhere too, with encouragaing signs by bodies like ITU, BEREC and assorted national authorities that private/local sharing is becoming important. 

In the UK, various bodies including Ofcom, National Infrastructure Commission, DCMS (the ministry in charge), TechUK/Spectrum Policy Forum (link) and others have referenced the potential for shared/private spectrum - and even invited me to talk about it - but until now, not much concrete has happened.

What use-cases are important here?

From my perspective, the main focus in actual deployment of private LTE has been for industrial IoT and especially the ability for large enterprises to run their own networks for factories, robots, mining facilities, (air)ports or process plants. Some of these also want human communications as well, such as replacing TETRA mobile radio / walkie-talkie units with more sophisticated cellular smartphone-type devices, or links to UCaaS systems.

These are all seen as future 5G opportunities by vendors too. They are also often problematic for many MNOs to cover directly - few are really good at dealing the specialised demands of industrial equipment and installations, and the liability, systems-integration and customisation work required.

Together with big companies like GE and Bosch and BMW, there has been some lobbying action as well. CBRS has had a broader appeal, with numerous other categories showing interest too, from sports stadium owners, to cable operators looking for out-of-home coverage for quadplay, or fixed-wireless extensions.

But I'd say that rural coverage, and more generic in-building use-cases, have had less emphasis by regulators or proponents of Micro-MNO spectrum licensing. That's partly because rural uses are often hard to generate business cases and have fragmentary stakeholders by definition, while in-building represents an awkward mix of rights, responsibilities and willingness-to-pay.  

Yet it is these areas - especially rural - that Ofcom is heavily focused on, partly in response to some UK Government policy priorities, notably around rural broadband coverage.

What has been announced?

There are two separate announcements / consultations:
  • An immediate, specific proposal for 700MHz and 3.6-3.8GHz auctions to have additional coverage conditions added to "normal" national mobile licenses, especially for rural areas. This includes provisions for cheaper license fees for operators that agree to build new infrastructure in under-served rural areas, and cover extra homes in "not-spots" today.
  • A more general consultation on innovation, which focuses on various interesting sharing models for three bands: the 1800MHz DECT guard bands (as discussed above), the 3.8-4.2GHz range and also 10MHz around 2.3GHz.

The first proposal is essentially just a variation of "normal 3.5GHz-band national 5G licenses", similar to the earlier 3.4-3.6GHz tranche which has already been released in the UK. Some were hoping that this would have some sort of sharing option, for instance for neutral-host networks in rural or industrial sectors, but that has been sidelined. 

Unlike Germany, which has just 3 MNOs and a powerful industrial lobby wanting private spectrum, the UK has to squeeze 4 MNOs' 5G needs into this band, with a big chunk already belonging to 3/UK Broadband. So, it has stuck with fairly normal national licenses. Instead, there's some tweaks to incentivise MNOs to build out better rural coverage. This helps address some of the UK government's and voters' loudly voiced complaints, but doesn't really affect this post's core theme of private/novel network types.

It is the second consultation that is the most radical - and the one which could potentially reshape the mobile industry in the UK. There are two central elements to its proposals:
  • Local-licensed spectrum in three "shared" bands, with Ofcom managing authorisations itself, with a fixed pricing structure that is just based on cost of administration, rather than raising large sums for the Treasury. There are proposals for low-power and mid-power deployments, suitable respectively for individual buildings or sparsely-populated rural areas.
  • Secondary re-use of existing national licensed bands. In essence, this means that any existing mobile band could be subject to 3rd-party localised, short-term licensing in areas where there is no existing coverage. This is likely to be hugely controversial, but makes inherent sense - essentially it's a form of "use it or lose it" rule for MNOs. 

Local licensing in shared bands
The local licensing idea has numerous potential applications, from industrial sites to neutral-hosts to fixed-wireless access in rural districts. It updates the 1.8GHz 2006 low-power wireless licenses to the new approach, and adds in the new bands in 2.3GHz and 3.8-4.2GHz. 

While I'm sure that some objections will be raised - for example, perhaps around the low-cost aspects of these new licenses - I struggle to find many grounds for substantive disagreement. It is, essentially, a decent pitch for a halfway-house between national licenses and complete WiFi-style unlicensed access. Like CBRS in the US (which is much more complex in many ways) it could drive a lot of innovative network deployments, but at smaller scale, as CBRS is aimed at county-sized areas rather than local areas as small as 50m diameter. 

There are numerous innovations here - and considerable pragmatism too, and plenty of homework that's been done already. The medium-power band, and the rural restrictions for outdoor use, are both definitely interesting angles - and well-designed to ensure that this doesn't allow full national/mobile competition "on the cheap" by aggressive new entrants. The "what if?" consultation sections on "possible unintended consequences" and ways to mitigate  them are especially smart - frankly all governmental policy documents should do something similar.

Ofcom also discusses options for database-driven dynamic spectrum approaches (similar to CBRS, white spaces and others) but thinks that would take too long to develop. It essentially wants a quasi-static authorisation mechanism, but with short enough terms - 3 years - that it can transition to some DSA-type option when it's robust and flexible enough. 

(As an aside, I wonder if the ultimate version is some sort of decentralised blockchain-ish decentralised-database platform for dynamic spectrum, which in theory sounds good, but has not been tried in practice yet. And no, it shouldn't be based on SpectrumCoin cryptocurrency tokens).

Secondary licensing of existing bands

This is the really controversial one.

It basically tells the MNOs that their existing - or future - national licenses don't allow them to "bank" spectrum in places where it's not going to be actively used. If there's no coverage now, or credible mid-term plans for build-out in the future, then (as long as it won't create interference) then other parties can apply to use it instead, as long as Ofcom agrees that there's no risk of interference. 

Unlike the shared-band approach (except for 1800MHz), this means that devices will be available immediately, as they would operate in the same bands that already exist. It would also potentially apply for the new 5G bands, especially 3.4-3.8GHz. 

There's a proposed outline mechanism from Ofcom to verify that suggested parallel licenses should be able to go ahead, and again a fairly low-cost pricing mechanism.

Clearly, this is just a broad outline, and there are a lot of details to consider before this could become a reality. But the general principle is almost "use it or lose it", although more accurately it's "use it, or don't complain if someone else uses it until you're ready".

There are a few possible options that have been suggested in the past for this type of thing - leasing or sub-licencing of spectrum by MNOs, or some form of spectrum trading, for instance. In some countries / places this has worked OK, for example for mines in the Australian Outback running private cellular, that have been able to do a deal with one of the national MNOs. But it's complex to administer, and often the MNOs don't really have incentives or mechanisms to do this at scale. They're not interested in doing site-surveys, or drawing up unique contracts for £1000 a year for a couple of farmhouses or a wind-turbine on a hilltop. Plus, there are complexities about liability with leasing (it's still the original licensee's name on the license).

While there will be costs for Ofcom to manage this process, it thinks they should be reasonable - it's pricing the licenses at £950 for a 3 year period. 

All this is pretty radical. And I expect MNOs and industry bodies to raise blue-murder about this in the consultation. Firstly, they will complain about possible interference, which is valid enough, but can be ruled out in some locations. They'll talk about the internal costs of the acceptance process. And above all, they may talk about "cherry-picking" and perceived competitive distortions.

The most interesting aspect for me is how this changes the calculus for building networks indoors, in offices, factories or public buildings. This could limit the practice of MNOs sometimes insisting that enterprises pay for their own indoor systems, for delivery of the MNOs' network coverage and capacity. It could incentivise operators to focus on indoor coverage, if they want to offer managed services for IoT, for example.

There's a lot of other implications, opportunities and challenges I don't have time to address in this post, but will pick up on, over the next weeks and months. There are technical, regulatory, commercial, practical and political dimensions.

I'm really curious to read the responses to this consultation, and see what comes out of the next round of statements from Ofcom. I'm probably going to submit something myself, as I can see a bunch of questions and complexities. Let me know if you'd like me to brainstorm any of this with you.

Spectrum is not enough

One thing is definitely critical for both proposals. The availability of local-licensed spectrum is not enough for innovators and enterprises to build networks. There are many other "moving parts" as well - affordable radio infrastructure such as small-cells, inexpensive (likely cloud-based) core and transport networks, numbering resources, SIMs, billing/operations software, voice and messaging integration, and so on. The consultations cover numbering concerns and MNC (mobile network codes), at least up to a point.

In some cases, roaming deals with national networks will be needed - and may be hard to negotiate, unless regulatory pressure is applied. As I've been discussing recently (including in this report for STL - link) this ties in with a wider requirement for revisiting wholesale mobile business models and regulation.


This is all very exciting, and underscores a central forecast of mine, that mobile network business / ownership models will change a lot in the next few years. We'll see new network owners, wholesalers and tenants - even as normal MNOs consolidate and merge with fixed-lie players.

I'd like to think I've played a small part in this myself. I've advised clients, presented and run many workshops on the topic, including my own public events in May and November 2017 (link), and numerous speeches to regulators, industry groups and policymakers. Industry, rural and in-building users need both more coverage and sometimes more control / ownership of cellular networks in licensed bands. 

There will need to be customisation, systems integration and a wide variety of "special cases" for future cellular. The MNOs are not always willing or able to deliver that, so alternatives are needed. (Most will admit, privately at least, that they cannot cover all verticals and all use-cases for 4G/5G). WiFi works fine for many applications, but in some cases private cellular is more suitable.

We're seeing a variety of new network-sharing and private-spectrum models emerge around the world, and a general view that they are (in some fashion) needed. What's unclear is what is the best approach (or approaches). CBRS, German industrial networks, Dutch localised licenses, or something else. I'd say that Ofcom's various ideas are very powerful - and in the case of the secondary re-use proposal, highly disruptive.

Edit & footnote: rather than "secondary re-use", perhaps a better name for this proposal is "Cellular White-Space", given that it is, in essence, the mobile-spectrum equivalent of the TVWS model.

If you'd like to discuss this with me - or engage me for a presentation or input on strategy or regulatory submissions - please reach out and connect. I'm available via information AT disruptive-analysis DOT com

Also, please subscribe to this blog, follow me on Twitter and LinkedIn - and (new for 2019!) look out for new audio/podcast and YouTube content. 

There are two audio segments that relate to this blog post:
Part 1 covers the general background to private cellular (here)
Part 2 covers the specific Ofcom proposals (here)