While the press hype up and distract you about the long term potential of 5G, there is a quiet revolution in progress. Almost everywhere, you will find most mobile operators around the world are quietly reallocating their spectrum, upgrading from 2G to 3G and 4G. Some have switched off 2G entirely. This has implications for smartphones, small cells and infrastructure choices in the coming years ahead.
What is spectrum refarming?
Each newer generation of mobile network introduced significantly improved radio technology, increasing capacity, data speeds and performance. 4G/LTE can penetrate further indoors than 2G, achieving 100x data rates with high definition voice. Often each new generation comes with its own dedicated spectrum, 2G at 900MHz, 3G at 2100MHz and 4G at 700/800MHz.
4G also opened up many new bands such as 2.3GHz and 2.6GHz, offering both TDD and FDD mode that means it can be assigned to many more dormant bits of spectrum. We’ll even see it introduced into 3.5GHz this year and 5GHz thereafter.
But for existing networks it can be attractive to reassign frequencies previously used for 2G or 3G and operate them with 4G/LTE. You would normally expect new LTE spectrum to be used first, expanding the capacity of each existing cellsite. Higher frequencies (say 2.6GHz) are most useful to add capacity in dense/busy areas where shorter range and spectrum reuse is important, whereas lower frequencies (900MHz) are more relevant in sparsely populated areas requiring larger coverage and in-building penetration.
Refarming isn’t so worthwhile until a substantial proportion of the user smartphones are compatible, so there is a market timing issue where operators would want to monitor and even encourage upgrades while matching infrastructure progress to take advantage of it.
Mobile operators have been looking ahead at this issue for several years, and most have re-equipped their entire basestation inventory with software configurable radios from Ericsson, Huawei and Nokia. These can be remotely assigned to 2G, 3G or 4G as required, although major band changes (e.g. from 2100MHz to 800MHz) would require antenna replacement. There may also be some physical adjustment to antenna tilt because of the longer range of 4G.
Examples of refarming to date
We’ve previously reported that several leading operators planned to switch off their 2G service altogether. Japan had done this some years ago (their proprietary PDC system was similar to but incompatible with GSM). After many years of advanced notice on 1st January 2017, AT&T started turning it off across the USA. There had been many years of advanced notice (and a quiet backroom deal with T-Mobile to take over some machine-to-machine contracts), so it was relatively painless. Nonetheless, one San Francisco bus company suffered problems with it’s traffic information App – some of their buses with outdated GSM trackers “disappeared” and so the App gave incorrect information about when the next bus was due. They may also be losing out on some potentially lucrative inbound roaming visitors, where their home networks smartphones don’t support AT&T’s 3G or 4G bands.
Most of the European operators I’ve spoken to are quite keen to retain a sliver of GSM coverage, even if just a few hundred kHz, because virtually every mobile phone has this baseline capability to fall back to. Some M2M systems would be quite costly to upgrade.
More of the 900MHz spectrum originally used for GSM has being reassigned for 3G, with many smartphones supporting this band. It’s not always entirely straightforward, since a matching 5MHz downlink and uplink band has to be found and cleared. Not all GSM spectrum allocations were made with this in mind. Some swaps between operators have been needed (with regulatory approval).
From 2016, we saw more operators introducing LTE in the 1800 and 2100MHz bands, refarming from 3G. Sometimes this has been quite aggressive. This UK study explains how Vodafone UK allocates 5MHz for 3G and 10MHz for 4G at 2100MHz today, having migrated 5MHz of paired 900MHz spectrum from 2G to 4G.
The long term trend over the next 5 to 10 years will be to migrate almost all except a small sliver of 2G across to LTE.
MIMO also a popular strategy this year
Operators remain keen to squeeze the utmost out of their existing cellsites. MIMO (Multiple Input Multiple Output) makes use of parallel transmitters and receivers to send/receive duplicate signals via diverse paths to the user. It needs additional antenna, although it’s also possible to use (horizontal and vertical) polarisation for separation (which is popular for indoor environments). Beamforming extends this feature by directing the signal towards each user in turn.
Huawei announced trials for Vodafone UK of 4x4 MIMO with beam-forming (8T8R) in Manchester using 2.6GHz TDD in transmission mode TM-8, indicative of dual layer beam-forming.
It’s the signal quality improvements from MIMO and beamforming that provide the greatest benefit, allowing higher modulation rates. By encoding a datastream using 256QAM rather than 16 or 64QAM, you can pass through substantially more data at higher speeds and thus increase capacity.
This requires a lot of extra equipment and more visible extra antenna, which may not be viable for some prominent sites due to planning restrictions. I’d also comment that achieving these very high QAM levels is unlikely to be achieved inside buildings, unless a dedicated inbuilding small cell or DAS solution is installed.
After refarming comes densification
But ultimately there are limits to these techniques and densification is inevitable.
I’m seeing analysts report growth of an urban densification layer especially by operators that have fewer spectrum options. Telefonica O2 for example has deployed a great number of microcells in dense urban locations around the UK.
I’d expect these to become almost entirely 4G/LTE and serve a relatively small bubble, reusing the spectrum efficiently. The short range and high signal quality between small cells leads to highly efficient QAM, especially Enterprise systems deployed indoors. This means they don’t have to match all of these performance squeezing features of their macrocell siblings, since they are each serving fewer end-users. There will be some industry debate about the tradeoff between the value of high performance from each site versus the cost of deploying larger numbers of simpler small cells.
While urban small cells are mostly LTE only, it’s quite possible to deploy multi-mode 3G/LTE products which can themselves be refarmed to become LTE only in the longer term once smartphones are widely compatible.
I’ve also heard it said that in some areas, especially third world countries, where GSM remains predominant, that users would invest in LTE phones if they had a network to connect to. Arguably Reliance Jio of India has confirmed that to be the case, having launched an LTE only network aimed at providing a very low cost service.
Definitely a trend to watch as the year unfolds.