The feasibility of frequency spectrum sharing arrangements is becoming increasingly of interest to ICT regulators for many reasons. Among these reasons, two standout. First, the frequency spectrum requirements to deliver broadband services is getting larger. This requires opening new frequency bands where existing users have legitimate needs. Second, the utilization of spectrum varies widely in time and geography leading to inefficiencies in exclusive licensing. This is becoming more critical in high frequency bands that have localized coverage, such as the ones meant for 5G and beyond. So what options are available to regulators and how can they go about balancing the needs of multiple users?
Not a new phenomenon
Frequency spectrum sharing is not a new phenomenon. It dates back to over a decade at least, and applies to both exclusive (licensed) and unlicensed spectrum regimes. For instance, Wi-Fi implements DFS which requires a Wi-Fi access point to sense and vacate channels in the 5 GHz band to avoid interfering with Doppler weather radars. DFS entered in 2003. Similarly, the AWS-1 band, which was auctioned in 2006, required manual coordination of adjacent bands between MNOs and US government users. Later, TV whitespace developed the notion of geolocation database. Today, there are many shared bands and techniques for sharing spectrum. The CBRS band, auctioned last year, was the first to auction shared spectrum on multi-tier basis.
| Band | Year | Country | Regime | Type of sharing | Parties involved |
|---|---|---|---|---|---|
| TVWS: 600 MHz | 2010 / 2017 | US, Canada | Unlicensed | Automated: Geolocation database | Wireless microphones and medical telemetry devices; TV broadcast |
| CBRS: 3550 – 3700 MHz | 2020 | US | Licensed & Unlicensed | Multi-tier: SAS, Geolocation database, ESC | Government users |
| 3650-3700(1) MHz | 2007 | US, Canada | Lightly licensed | Database/Manual | Common users of band |
| AWS-3 | 2015 | US | Licensed | Manual coordination | Commercial and government users (DoD, DoJ) |
| 2300 MHz | EU, UK | Various incumbent users | |||
| 5 GHz | 2003 | US, others | Unlicensed | Automated sensing/DFS | Radars (weather, aviation), Wi-Fi access nodes |
| 6 GHz (U-NII-5 & U-NNI-7) | 2022 | US | Unlicensed | Geolocation database/AFC | PTP microwave, satellite (uplink), Wi-Fi access nodes. AFC specifications is work in progress. |
| 1800 MHz (1780 – 1785 MHz and 1875 – 1880 MHz) | 2019 | UK | Licensed | Manual coordination; local license | CSA licensees |
| 2390 – 2400 MHz | 2019 | UK | Licensed | Manual coordination; local license | MOD, Amateur radio |
| 3.8 – 4.2 | 2019 | UK | Licensed | Manual coordination; local license initially; automated database (future) | Fixed links |
| 24.25-26.5 GHz | 2019 | UK | Licensed | Manual coordination; local license; Indoor use only | Fixed links |
| (1) US band is grand-fathered into CBRS. Canadian band will be auctioned in the 2023 C-band auction and lumped into 3.7 – 3.9 GHz band. | |||||
Types of spectrum sharing
Sharing could be a manual coordination activity as is the case in some bands. This works where there are few users with infrequent and planned usage activity. The UK spectrum sub-licensing is an example of this arrangement. Automated techniques, however, enable scalability and coordination between different user groups. The implementation of automated coordination could take different forms: geolocation database (TVWS), a geolocation database with spectrum coordination and RF sensing (CBRS), or RF sensing alone (DFS). This is by no means an exhaustive classification of sharing techniques since the implementation of the sharing scheme leads to further differentiation. [See here the insightful paper by John Leibovitz & Ruth Milkman for additional details.]
Planning for sharing regimes
Regulators have to answer a few critical questions in designing a sharing scheme. Some of these questions include:
- Who are the current and future users of spectrum? Who are the incumbents needing protection, and to what degree? What new users and applications will sharing enable?
- What are the technical characteristics of the band?
- Where will the sharing happen?
- When will the sharing happen?
- How will sharing be authorized, and by whom?
- How to resolve conflicts in a sharing regime?
- Is harmonization with existing implementations needed, and will the ecosystem be able to support the scheme?
The last point is especially important for small countries with low volume demand on equipment in case selecting automated techniques and sensing capabilities. In such cases, it makes most sense to harmonize with existing implementations in order to access equipment with the appropriate certification.
Levers for spectrum sharing
There are a few levers regulator could control in implementing shared spectrum access. Location (indoor/outdoor), geography (boundaries for the sharing regime) and RF power level are a few. Other levers include additional restrictions on deployment scenarios such as limiting the height of antennas or putting restrictions on the radiation patterns.
Operators dislike sharing
Operators want exclusive licensing schemes and dislike sharing frequency spectrum. There are some legitimate reasons for this. For instance, tracking interference is a difficult and time-consuming job, especially when its coming from 3rd party users. It is both less costly and more efficient for MNOs to be in sole control of their spectrum assets.
On the other hand, there are legitimate questions on the sustainability of exclusive licensing schemes for the reasons mentioned earlier. Regulators looking to balance among the need of multiple players will have to consider sharing schemes.
Here, I like to bring an additional perspective which I believe will increasingly factor in the tradeoffs regulators need to consider in the future.
A new angle: changing finances
Proceeds of frequency spectrum auctions have mostly decreased in recent years. Regulators can no longer expect the returns on frequency spectrum auctions as they did in prior years for the low frequency bands. The value of spectrum in terms of $/MHz-PoP is decreasing as frequencies increase. The increase in bandwidth (MHz) is not sufficient to compensate and bridge the shortfall in auction proceeds.
There is an exception to this trend: the US C-band and Canadian 3.5 GHz auctions have generated some of the highest proceeds and valuations. I believe that the reason for this is the competitive dynamics of the US market at the time of the C-band auction, and regulatory design – or mis-design – of the Canadian 3.5 GHz auction. The evidence is in favor of my point by comparing the proceeds of the three US mmWave spectrum auctions which raised a total of $10.3 billion to the proceeds of the 600 MHz and AWS-3 auctions which raised $19.3 billion and $41.3 billion, respectively. The situation is more clear outside of North America where many mid-band and mmWave failed.
A reason for lower future auction valuations is that, in most parts of the world, the number of bidders is limited to the incumbent MNOs. Greenfields are rare as it is almost impossible to compete with the incumbents (Rakuten Mobile financial performance is an example; we shall see how well Dish does!) This limits competition and leads to prices that barely exceed the reserve price, as happened in many recent mid-band auctions. As the cost of building out mobile networks in higher frequency bands increases to the point where many service providers forgo interest in auctions, regulators will have more incentive to balance the competing needs for spectrum. This makes the case for shared spectrum more compelling than ever before.
Sharing to balance competing claims
One area that will see competing claims for spectrum is between satellites and terrestrial mobile networks. We already see this in the competing claims for the 12 GHz band in the US where T-Mobile and Dish are battling against SpaceX and OneWeb. The previous and upcoming WRC offer additional examples where the mobile and satellite industries clash over spectrum rights. Frequency spectrum sharing will be integral to develop the future hierarchical network that include NGSOs. [As a side note, direct-to-handset constellations will need to collaborate and coordinate with MNOs to deliver their services.]
Concluding Thoughts
The mobile industry, looking for its self-interest, will make claims for all types of spectrum: spectrum assures a moat that makes it very challenging for any potential competitor. Some are already calling for terahertz spectrum in 6G applications. I am yet to see how a mobile network operator could deploy such a short-range technology in any any meaningful access play. I use this example to highlight the need for an active role by regulators to define suitable sharing strategies. Spectrum sharing will increasingly take on a more important role in future generations of wireless networks – it is a matter of time!