As recently as a few years ago, the idea of major carriers (particularly cellular carriers) using shared spectrum was almost unheard of. Whilst some had dabbled with running their own Wi-Fi networks which used the unlicensed 2.4 and 5 GHz bands, these were always considered a nice-to-have; their mission-critical cellular networks operated only on their own private, licensed spectrum.
It’s not hard to see why: if you were able to choose between operating a network in uncontested private spectrum versus shared spectrum, it’s an easy choice. Even counting the cost of acquiring private licensed spectrum, for a major network operator the benefits of doing so outweigh that expenditure over time, with guaranteed availability and performance paying large dividends.
The problem, however, is that the amount of data these networks are required to send and receive has been increasing at an alarming rate. According to Cisco’s Visual Networking Index, global mobile data traffic exceeded 7.2 exabytes (equivalent to 7.2 billion gigabytes) per month at the end of 2016, compared to 4.4 exabytes per month at the end of 2015.
In 2011, mobile data networks carried only 400 petabytes per month. The amount of data usage per month at the end of 2016 represents a growth of 18 times since 2011 – without an accompanying 18-fold increase in the amount of available spectrum.
More data requires more wireless spectrum, as after a point it becomes infeasible to significantly improve data speeds using the same spectrum due to diminishing returns in increasing the order of modulation and coding in use, and difficulties in densifying networks by breaking one large macrocell down into many small cells when they are all using the same spectrum.
Although some additional licensed spectrum has become available in the traditional sub-1 GHz bands (favoured by the major cellular carriers for their excellent propagation characteristics), there simply isn’t enough spectrum available in these favoured frequency bands to accommodate the projected increases in mobile data usage.
Cisco predicts that monthly data traffic on mobile networks will reach 49 exabytes by the end of 2021, continuing the type of growth seen between 2011 and 2015. The only way for network operators to accommodate this huge increase is to look at alternative spectrum, for the first time seriously considering shared and unlicensed frequencies.
Ignoring the use of Wi-Fi in unlicensed frequencies for mobile data offload, there are many interesting developments occurring on the use of shared and unlicensed frequencies for cellular network usage. Foremost amongst these are the variants of LTE, LTE-Unlicensed (LTE-U) and License Assisted Access (LAA) that use unlicensed spectrum in various ways and degrees, primarily to deliver the downlink (towards the user’s mobile device) data traffic as this is the main consumer of bandwidth in mobile networks.
Doing so frees up the network operator’s private licensed spectrum for uplink data traffic, and other less-intensive users at that point in time. However, it also makes the intensive user’s experience dependent on the unlicensed 5 GHz spectrum in use for the downlink data traffic, subjecting them to the same interference from other users of the spectrum that cellular network operators have spent many billions of dollars investing in private licensed spectrum over several decades to avoid.
Are cellular network operators willing to bet their customer satisfaction on the use of unlicensed spectrum? The answer appears to be yes, prompted by the explosion in mobile data usage. At this point, cellular network operators have little choice – either size up their network capacities using whatever spectrum is available whilst providing reasonable assurances of quality where possible, or lose the race against customer demand to their competitors who are prepared to do so.
One solution is opportunistic and application-based usage of frequency. By switching a user who is watching streaming video to unlicensed spectrum, the network operator can ensure that other users – for instance those making emergency calls – are always able to perform their needed task, by ensuring these low-bandwidth latency-sensitive tasks are kept on the private licensed spectrum.
Millimetre-wave frequencies open up a new discussion on the use of private and shared spectrum. The 28 GHz band in particular in the United States is subject to a licensing model which favours large cellular operators, allowing them to access one or both 425 MHz blocks of spectrum that are available. At the same time, the 60 GHz band is organised similarly to the 2.4 and 5 GHz bands; unlicensed operation means the spectrum is available for use by anyone complying with regulations.
Both of these high-frequency bands are likely to find a home in next-generation cellular networks, used appropriately to ensure that the cellular network’s reputation isn’t unduly harmed by the effects of wireless interference in shared unlicensed spectrum.
The Citizen’s Broadband Radio Service (CBRS) aims to open up the 3.5 GHz band in the United States, traditionally occupied by military and satellite users. The use of this band will rely on a Spectrum Access System (SAS) which co-ordinates usage of the band and ensures these incumbent users still have appropriate access when required. All of the major United States cellular network operators are members of the CBRS alliance, and are clearly as hungry for spectrum in 3.5 GHz as they are everywhere else in the radio spectrum.
A formal shared spectrum scheme such as this is an interesting development between the extremes of unlicensed free-for-all spectrum, and the tightly-controlled private licensed spectrum that the cellular network operators prefer. How successful it will be in providing relief for cellular networks from the rapid growth of mobile data usage is unclear today, but you can expect it to be positive if limited as mobile devices begin to be rolled out with chipsets which support the band and SAS.
One thing is for sure: the days of cellular networks operating only on private licensed spectrum are over. Increasingly, unlicensed and shared spectrum will see an influx of attention from cellular network operators, and a proportional amount of their mobile data traffic as well.
What this means for other users of the band, such as Wi-Fi networks or Wireless Internet Service Providers (WISPs) who have traditionally used these unlicensed frequencies, is uncertain. Testing by Qualcomm appears to indicate that LTE-U is a better neighbour to Wi-Fi than Wi-Fi is to itself, but this has been disputed by some other parties.
In the next couple of years, beginning in early 2018 as T-Mobile looks to deploy LTE-U across the United States, we will see how deployments of cellular technology in the real-world impact existing users of shared unlicensed spectrum.
Here’s hoping that where needed, compromises can be reached between network operators to allow the landscape of global wireless network connectivity to keep growing, supporting many different network operators, users and use cases.