The more I look at enterprise mobile, especially its focus on verticals and IoT, the more I'm convinced there needs to be a change in industry structure, regulation and network ownership/operation. And that means new spectrum policy, as well.

In particular, private licensed-band wireless networks will be essential - that is, networks (using cellular, WiFi, LPWAN or other technology) that can be directly managed by organisations that are not traditional MNOs (mobile network operators), to provide high-QoS, reliable wireless connections. I'm thinking large companies running their own networks, industrial network specialists, local cooperatives, perhaps new government-sector initiatives, and various other aggregators, outsourcers and intermediaries. These will mostly be in-building / on-campus, but some may need to be genuinely wide-area, or even national, as well.

This is in addition to enterprise-centric initiatives in the MVNO/E space, vertical activities by fixed telcos and MNOs, unlicensed-band WiFi and LPWAN deployments, future "network slicing" concepts and so on.

 There are three main models for licensing radio spectrum today:

• Exclusive licenses: Dedicated access to certain bands is very common today, for example for mobile networks, fixed microwave links, broadcasters, satellite access and many government-sector uses, such as military radios and radar. Particular organisations have rights to solo access to particular frequencies, in a given country/region, subject to complying with various rules on power and so forth.

• Unlicensed: (also license-exempt): Beyond some basic rules on power and antenna siting, some bands are essentially "open to all". The 2.4GHz and 5GHz bands used by technologies such as WiFi, Bluetooth and many other technologies are prime examples, as well as bands used for consumer walkie-talkies and various medical and automotive applications.

• Shared spectrum: This covers various models for allowing multiple users for certain frequencies. It could involve temporary usage (eg for event broadcast), bands that haven't been "cleared" fully and still have incumbent users that newcomers need to "work around". It might be spectrum assigned in geographic chunks, or at low power levels and mandating "polite" protocols so that multiple users can co-exist. We've seen TV "white spaces" where under-used bands are opened up to others, and so forth.

The latter approach of sharing is becoming much more important - despite continued clamour for exclusive licenses, especially from the mobile industry. Given that the demand for spectrum is rising from all sides - mobile, WiFi, utilities, broadcast, satellite, Internet and many others - and each has a different demand profile (global / national / regional and subscription / private / amenity etc), a one-size-fits-all model cannot work, given limited spectrum resources. More spectrum-sharing will be essential.

More models are now emerging for sharing spectrum bands. Depending on the details, these open up opportunities for a greater number of stakeholders. The US' innovative CBRS model (see link) for 3.5GHz is worth examining, and perhaps replicating elsewhere, especially Europe and also developed Asia. It is much more sophisticated - but more complex to implement - than the Licensed Shared Access (LSA) that Europe has leaned towards historically. In Disruptive Analysis' view this extra complexity is worthwhile, as it allows a much broader group of stakeholders to access spectrum, fostering greater innovation. 

The important differentiator for CBRS is that there are three tiers of users:

• Incumbents, primarily the military, which gets the top level of access rights for radar and other uses in the band

• Licensed access providers which can get dedicated slices in specific geographic areas. These are "protected" but subject to pre-emption by the top tier. They will also generate revenue for the government in terms of license fees - although awards will be for shorter periods than normal bands (3 years is being discussed).

• General access - basically this is like unlicensed access, but it has to work around the other tiers, if they are present.

To make all this work, the CBRS system needs databases of who is using what spectrum and where, and sensors to detect any changes in the top tier's usage. (The military, as incumbents, isn't keen on spending any money to actually tell the system what it's doing - it needs to be securely automated).

When all this is up and running, there will be many potential user groups for shared spectrum such as this, using either the priority licenses, or general access tiers:

• Incumbent mobile operators needing more capacity in specific areas

• MVNOs wanting to "offload" some traffic from their host MNO networks, onto their own infrastructure, without the expense of full national coverage. This could work either alongside, or as an alternative to, WiFi-based offload or WiFi-primary models.

• Enterprises wanting to deploy private cellular networks indoors or over large campuses (eg across an airport or oil-refinery for IoT usage)

• Potentially, large-scale WiFi deployments in new bands, with less subject to interference than mainstream unlicensed bands - although this would require devices/chipsets supporting new frequencies that are currently outside the proper WiFi standards.

• Various "neutral host" wholesale LTE models, for example run by city authorities for metropolitan users, or cloud-providers for enterprise - or as a way to provide better indoor coverage for existing incumbent "outdoor" operators, without their needing individual infrastructure in each building. This could allow the pooling of back-end / administrative functions and costs across multiple local LTE networks in shared bands. Imagine an Amazon AWS approach to buying cellular capacity, on-demand.

• Various approaches to roaming or "un-roaming" providers - for example, a theme-park operator or hotel owner could offer its foreign guests "free LTE" while on-site.

• Potential new classes of cellular operator, such as an Industrial Mobile Operator (imagine GE or ABB integrating cellular access into machinery & plant equipment), various IoT platform providers, and integration opportunities with Internet, healthcare, transport and other systems.

This approach may not work for enterprise wireless users requiring national (or very broad-area) coverage, such as utility companies or transport providers. There are separate arguments for utility and rail companies getting slices of dedicated spectrum, or some other model of national sharing.

Importantly, CBRS means that LTE-U variants like MuLTEfire can be used to create private cellular networks. Coupled with cheap, virtualised (& probably cloud-based) core networks, this means that mobile networks are much more accessible to new entrants. The scale economies of national licenses will no longer apply to lock out alternative providers.

In other words, we will see consolidation of national MNOs, but fragmentation of localised MNOs or (PNOs as some are calling private networks). 

While some MNOs and their industry bodies may be concerned at more competition, privately many of them acknowledge that a lot of the use-cases above cannot realistically be offered by today's industry. 

Even large MNOs can probably only pick 2 or 3 verticals to really get deep expertise in - maybe smart cities, or rail, or utilities, say. But they cannot get enough expertise to effectively build customised, small networks in all the possible contexts - car factories, ports, hospitals, mines, hotels, shopping malls, airports, public safety agencies, universities, oil refineries, power stations and so on. Each will have its own requirements, its own industry standards to observe, its own systems to integrate with, its own insurance/liability issues and so on. They need wireless for all sorts of reasons from robots to visitors - but today's MNOs will not be able to satisfy all those needs, especially indoors.

For many governments' visions of future factories, cities and public services, good quality wireless will be essential. But it will need to be provided by many new types of providers, with business models we can only guess at.

While CBRS is still at an early stage, and will be tricky to implement, we need something similar to it - with multiple tiers including a "permissionless" one - in Europe, developed Asia and the rest of the world. Enterprise and private cellular networks (and other licensed-band options for WiFi and LPWAN) are critical - and policymakers and regulators need to acknowledge and support this.