With spectrum prices rocketing in the USA recently (e.g. the AWS-3 auction exceeded $44 billion), the wireless industry is looking enviously at a couple of substantial bands allocated elsewhere. There is one hitch – there is no more pre-cleared broadband-ready spectrum. The cost and time to clear is also out of step with wireless realities. So, could they agree ways of sharing that with existing users co-operatively, and how would that spectrum be incorporated into small cells?
Large chunks of spectrum in two particular bands at 3.5GHz and 5GHz have been in the news over recent months.
- 3.5GHz (more specifically 3400MHz through 3800MHz), which is currently used by some commercial, naval and military radar systems.
- 5GHz (more specifically 4915MHz through 5825MHz) currently specified as unlicenced and widely used by Wi-Fi.
The limited range of these frequencies makes them well suited to indoor small cells, offering high reuse and relatively short range. They will be of most value in those dense urban areas with the highest traffic density demands. Very high peak speeds are feasible with expectations and tests showing up to 1Gbps.
3GPP standards already define Bands 42 and 43 at 3400 to 3600 and 3600 to 3800MHz of TD-LTE for this purpose. In the US only, 802.11y can be used for Wi-Fi between 3550 and 3695 MHz.
In the US, this is just the initial part of a broad shared spectrum allocation policy. 1000MHz of mostly Federal, contiguous spectrum between 2.7 and 3.7GHz has been earmarked to become a new "spectrum superhighway". This was targeted by President's Council of Advisors on Science and Technology on Spectrum Management (PCAST) report of July 2012 (summary slidedeck). This band has the opportunity to be internationally standardized thus further increasing the role for small cell access outside the US.
The same sharing technology developed in the 3.5GHz band could also be used in lower frequency bands. An additional 200MHz was identified in the PCAST report as potentially share-able.
Commercial availability for LTE at 3.5GHz varies between regions
The UK boasts the world's first (and I believe only) commercial LTE network using this band. UK Broadband offers a fixed broadband service of up to 50Mbps in Central London for just $30/month (unlimited). With 120MHz of spectrum at their disposal, they have plenty of capacity to offer but the service only makes sense in dense urban areas which is compatible with the short transmission range at those frequencies.
The Japanese regulator has also been very keen to make use of this spectrum, proposing to give (free of charge) 40MHz of TDD bands to each of the three national operators in return for a commitment to deploy it across at least 50% of the urban areas that need it. They believe that 1Gbps peak rates are quite achievable, even without Carrier Aggregation at other frequencies. I believe this proposal would use the same 3GPP band plan above and could potentially happen fairly quickly.
In the USA, the major problem is that a lot of naval military radar already uses this frequency and can't be shifted elsewhere. If you exclude those coastal areas completely, this would block use for 60% of the population base of the US. So the FCC have proposed a 3-tier priority scheme which gives preference in the following order:
a) The existing government users (e.g. radar), then
b) Licenced access, obtained through auction, no interference to existing services, then
c) General best-efforts access, no interference to higher tiers
Current thinking is for the FCC to allocate some 100 to 150MHz of TDD spectrum from 3550MHz up – which doesn't align with the 3GPP allocation above. Verizon is currently trialling this in their labs.
Accurately modelling external RF interference protects existing use
A lot of work has been done to identify methods of achieving mutually acceptable sharing. The simplest involves looking up a central database with your current location to find the permitted frequencies, RF power levels etc. To ensure maximum density with least interference, a more complex approach involves measurement of building containment loss plus terrain path-loss models which are both necessary to estimate interference accurately. In-building RF transmissions are shielded and contained by external walls and if measured may allow substantial re-use, whereas transmissions outside or on the rooftop typically will not.
Sensing interior building loss is also critical to increase system re-use. One company, iPosi, has developed a capability to determine the building RF isolation at each small cell site as part of its 3D location and synchronization through its extreme sensitivity A-GNSS solution. These are two more mandatory features required to comply with the FCC's shared spectrum proposed rules issued in April 23, 2014.
In theory, you can transmit anything you like in an unlicensed band – there is no requirement for it to conform to a particular global standard – although regulators expect you "behave". That means listening before transmitting to reduce interference to others, limiting maximum radiated RF power and complying with guard bands and spectral power masks etc.
So if the LTE system could adopt a listen-before-talk mode, some would therefore argue that it's quite legitimate to use LTE in those bands. The lower RF power would tend to limit use mostly to indoors, but this is where the highest traffic demands are to be found.
With smartphones already using this band for Wi-Fi, it may not be too much of a stretch to use it for TD-LTE instead. That's certainly the proposal from Qualcomm and others over recent years. LTE in the unlicenced spectrum is termed LTE-U and this is aggregated and controlled by LTE running in licenced bands, a solution termed Licenced Assisted Access. The main control channel and basic voice or data calls would remain on much lower frequency licenced LTE spectrum, but during peak traffic periods, supplemental data channels would be added using unlicensed spectrum.
We can only hope that the solution if adopted is a global standard and not country specific.
Small Cells with unlicensed spectrum
Dedicated small cells that operate only using the 3.5GHz band are quite feasible to design and build today – it would just involve alternative RF components on the front end.
The Licenced Assisted Access mode requires Carrier Aggregation - simultaneous use of two bands, typically one FDD and one TDD. The concept has been proven by Nokia using macrocells already, and should be quite feasible in a small cell. I wouldn't be surprised to see a demo of that at MWC this year. Ericsson have been one of the first to announce product availability of LAA Small Cells by end 2015 and demonstrated it publicly at the recent CES show.
As with any new band or feature, it would take time for new smartphone devices to support it. This isn't a software upgrade – the RF hardware is band specific – so would be rolled out slowly through the usual upgrade cycle. That means that small cells would continue to support existing, standard frequencies and modes alongside it.
A possibility is for a combined 3G/LTE multimode small cell to be upgraded to become LTE/LAA or also use in bands at 2.6GHz and 3.5GHz, the latter is already favouring LTE according to Qualcomm.
This isn't a single, dominant choice for any product manager to include into their roadmap, whether for small cells, Distributed Radio, Cloud RAN or DAS products. However, both bands are too high in frequency to be handled by coax based DAS networks.
My guess is that these frequencies and modes will initially be more suitable for a small cell product set, where the larger number of nodes allows high frequency reuse offer to achieve the highest speeds and can be targeted at the most demanding locations.
Expect to hear a lot more about both of these options during 2015.
Excellent free 20 page report on Shared Spectrum from 4G Americas, Oct 2014