There are various ways to make use of LTE with unlicensed spectrum of which LAA seems to be the most popular and imminent. It boosts capacity and speed of standard LTE data streams by augmenting using one or more data channels in unlicensed spectrum at 5GHz. Read on for more detail including implications for product design, permitted frequency combinations and realistic timescales.
Brief Recap of options for Unlicensed Spectrum with LTE
- LWIP: You can just run parallel data paths through the existing LTE and Wi-Fi connections. This uses Multi-TCP to re-sequence and reassemble the parallel data streams at an anchor point in the core network.
- LWA (LTE Wi-Fi Access): As above, but runs the LTE Layer 2/3 stack on top of the Wi-Fi Physical Layer 1.
LTE-U (LTE in Unlicensed Spectrum): Runs a relatively unmodified form of LTE in the unlicensed 5GHz band, so is only legal in a few countries (including USA, Japan but not Europe). Intended as a precursor/quicker to market form of LAA, Qualcomm have said their chipsets will be software upgradable from LTE-U to LAA and could even support both simultaneously. Likely to be superceded by LAA within a few years.
- LAA (Licensed Assisted Access): Runs a modified form of LTE in the unlicensed band, boosting downlink speeds by aggregating with an existing standard LTE connection. Dynamic channel selection (to find the least used sub-band) and Listen-Before-Talk (to avoid trampling over other users) have been introduced to co-exist with other unlicensed users.
- eLAA (enhanced Licensed Assisted Access): Adds an uplink in the unlicensed band, so can boost uplink speeds too.
- MulteFire: Uses LAA and eLAA to run an LTE network entirely in the unlicensed spectrum.
3GPP standards for LWIP, LWA and LAA are formally issued as part of Release 13 (June 2016). eLAA will be included in Release 14 (2017). The MulteFire Alliance expects to publish their own standard (building on the 3GPP work) Q4 2016.
The 5GHz band is the primary target for all of the above, although shared spectrum bands (specifically the US CBRS band at 3.5GHz) are also of interest. This is defined as Band 46 (5150 to 5925MHz) and is TDD mode.
Many countries have regulations regarding fair behaviour to unlicensed spectrum. European publishes EN 300 893 which specifies maximum permitted RF power and listen-before-talk. By complying with that (and a few other regional specific rules), truly global products can be developed which would work legally anywhere.
LWA vs LAA
Several vendors have been backing both these options, arguing that there are different benefits depending on whether a mobile operator has more or less licensed spectrum.
I’d argue that we have too many choices above, and this adds uncertainty and delays take-up of either. LWA combines the worst of both worlds while introducing a lot of complexity. The LTE radio interface is generally thought to be better while the Wi-Fi protocol stack is very widely deployed and low cost. So why not adopt standard Wi-Fi as a relatively simple and straightforward quick fix (using the LWIP architecture), and focus on LAA for the medium term?
Permitted Combinations and constraints
With so many different LTE frequency bands, there has to be some constraint on which combinations are supported otherwise handsets would become unmanageable. 3GPP specification 36.101 has a set of tables that does just that.
The current version of that spec permits aggregation of Band 46 (5GHz) only with Band 1 (1.9/2.1GHz) or Band 2 (1.8/1.9GHz). No doubt more options will be added in the future.
Small cell vendors indicate that where LAA is in operation at 5GHz, then Wi-Fi wouldn’t be used. I can imagine that trying to engineer an RF solution that operated both systems concurrently (on a small cell or a handset) would be extremely difficult.
Early LAA systems adopt a single 20MHz channel within the 5GHz band alongside a 20MHz licensed channel which in theory could provide peak rates of more than 300Mbps in ideal environments. Real world experience in congested areas will differ. Personally, I think this would be more than enough but further carrier aggregation of 20MHz licensed plus up to three 20MHz sub-bands are in the roadmap. ZTE thought this could achieve 600Mbps downlink data rates.
eLAA will boost uplink speeds, such as for sending streaming video and photos, and follow a year or so behind downlink only LAA.
Where small cells are deployed by multiple operators within the same building, each will operate in its own licensed frequency bands. Where LAA is also used, these are unlikely to overlap, with each small cell dynamically sensing and allocating the least used channel.
Small Cell technology is the most suitable
Today’s DAS systems are designed to handle licensed frequencies up to 2.7 or 3GHz. Support for 5GHz bands would need radical replacement of both RF radio heads and central equipment. The shorter range of the 5GHz band would require many more radio heads, adding substantial cost. A possible architecture would be to deploy LAA only small cells to augment an existing DAS deployment, boosting capacity and peak speeds. I’ve not seen that commercially proposed by DAS vendors at this stage.
Small cells, already relatively low cost and easily deployed in volume, are far more suited for LAA. Many commercial products have a Wi-Fi option which I’m not convinced is used very often. Instead, repurposing this 5GHz radio for use with LAA makes a lot more sense.
Distributed Radio Systems, such as Huawei Lampsite and Ericsson RadioDOT, will support LAA, repurposing the built-in 5GHz radio from Wi-Fi to LAA.
The higher peak capacity of each small cell, carrying data traffic from 2 or 3 radios, should just about be handled by the 1Gbps Ethernet backhaul connection commonly used. (Today, its often backhaul to the building itself is the limiting factor). Faster data transmissions technologies already exist for use with Cat 5/6 cables should the need arise – we’ve looked at the emerging NBaseT technology before.
While I'd expect most LAA deployments to be integrated into the same small cells, it's also possible to use macrocells for the licensed channels alongside small cells for LAA only. This may be a potential option for operators with residential wireline broadband modems, who could use this to boost (and offload video traffic) from their macrocells, provided that the domestic broadband was adequately dimensioned.
Is it a reason to delay?
Enterprise small cells already offer huge capacity and service quality improvements. They don’t need any additional spectrum to achieve that and are compatible with existing handsets.
Future systems are likely to introduce LAA to provide higher capacity and performance. New handsets/smartphones will be needed to make use of it, causing a lag in take-up and resulting benefits.
However, this does paint a picture of a clear long term roadmap as we move from today’s 3G/4G multi-mode, to 4G only, LAA, eLAA and multi-channel LAA. This gives huge potential capacity and capability as LTE continues to evolve and meet our needs for the future.
Perhaps there’s a reason it’s called Long Term Evolution.