A new emphasis for unlicensed LTE, but the effort still has fatal flaws

LTE-Unlicensed has a new name and the first concrete demonstration that it works in the real world. The technology, which allows LTE-Advanced to run in unlicensed and shared bands, notably 5GHz, has been badged ‘LTE LAA’ (Licensed-Assisted Access) by the 3GPP. This rebranding, and a demo by Huawei and NTT Docomo, seem designed to present 5GHz LTE as a non-threatening and workable technology, distancing it from some of the more excitable debates about the 3GPP community trying to squeeze WiFi out of its most important spectrum band.

After all Qualcomm’s work, over the past year, to establish LTE as an option in 5GHz, several 3GPP players are now keen to moderate some of the tone of the debate. This is strictly a supplemental downlink approach, they insist, with the primary carrier always remaining in the licensed band. Therefore it is no threat to WiFi as the pre-eminent technology in most of the 5GHz band and should not be presented as a cellular vs open networking dispute. (The other motivation behind this change of emphasis is to avoid signalling to regulators that operators will need less licensed spectrum if they can harness 5GHz.)

The language may have calmed down, but the issues of harnessing 5GHz to augment LTE capacity in licensed spectrum are no less contentious than if the 3GPP had, indeed, decided to mount a full challenge to IEEE technologies in licence-exempt frequencies (and who knows where this opening salvo, if successful, might take the cellular industry in future anyway?)

In some ways, it would be more honest to be attempting to promote LTE as an official technology for 5GHz, where it could take its chances alongside WiFi and some proprietary options. After all, WiMAX had a 5GHz implementation which (partly because it was an IEEE standard) was seen as potentially complementary to WiFi ‘ but in fact, had WiMAX thrived as a ‘4G’ platform, could have turned into a carrier-controlled threat to it.

The approach of LTE-LAA raises different concerns, because it is not competing for air as an alternative open technology to WiFi ‘ as various 2.4GHz standards do ‘ but is completely under the control of the operators. Should LAA be adopted by device and access point makers, it will still not be an option that can be adopted by any operator in licence-exempt spectrum (as 5GHz WiMAX would have been). Nobody envisages an open LTE ecosystem, because of the patent and supply structures of this market, but any chance of opening up the gates is excluded when LAA can only run while linked into a licensed-band network, which by its nature must be owned by a cellco.

So stressing the SDL aspects actually make LAA sound more, not less, like an attempt to hijack at least some of the 5GHz spectrum for the LTE carriers’ exclusive use, an outcome which would go against the spirit, if not the practice, of unlicensed bands, and which could, if successful, reduce the capacity available for WiFi and increase congestion, thus damaging the attractiveness of the technology.

That would be a victory for the 3GPP enclave but a poor result for the wireless market at large. The GSM experience showed how ecosystems and quality of experience are enhanced when a single technology dominates a band, and this will be true of unlicensed spectrum too. WiFi is clearly in pole position to be that standard in 5GHz, and that would further expand the range of interoperable devices and services available to consumers, and the capacity to support them. However, the scenario where LTE-LAA presents a real squeeze on WiFi’s space in the 5GHz band is far-fetched, outside of the wildest dreams of Qualcomm executives. It would, of course, depend on LTE-LAA being used very extensively, and that seems improbable. There are still all kinds of political, technical and IPR issues around LTE-LAA which will create uncertainty for most carriers as they formulate their HetNet plans, and which will lead to tensions with the WiFi community. Given the scale and power of the WiFi ecosystem, and the long timelines involved in any new 3GPP standard acquiring a significant base of devices and deployments, we expect those issues to be near-fatal to the LTE-LAA effort. (The technology will be standardized as part of 3GPP Release 13, officially due to be frozen in March 2016, though timelines are lengthening ‘ see below).

In the real world – away from industry politics and the cellular equipment vendors’ desperate need to hang on to the power and profits of a platform they control ‘ the benefits to operators seem to marginal, when they will be fully occupied in rolling out working HetNets using a rising variety of licensed bands plus carrier WiFi. Supplemental downlink (SDL) will be a useful tool to boost capacity, but can be achieved using other bands than 5GHz, and ones with less congestion. And while, some years ago, WiFi was a b’te noire for cellcos, these days the more powerful ones are often finding a way to break in the wild horse, tying carrier WiFi networks into their mobile cores and their policy engines, to make them behave almost like cellular.

Therefore, WiFi has become more asset than threat for many mobile carriers ‘ and for those facing the threat of competition from wireline or over-the-top rivals, armed with WiFi-first, grabbing a bit of 5GHz for LTE downlink will hardly be enough to turn the tables.

However, while this is mainly a self-interested vendor-driven initiative, there is involvement from some influential operators, such as Docomo, Verizon and China Mobile, and there is real investment going into the platform, so it cannot be written off entirely.

Huawei ‘ the most vocal supporter of LAA after Qualcomm ‘ has announced tests conducted with NTT Docomo, which it claims show superior coverage and capacity in 5GHz than 802.11n WiFi. The two companies have been working on LAA R&D for the past six months and now proclaim that ‘LAA is a viable technology for LTE and future LTE-Advanced’, as Seizo Onoe, CTO of Docomo, put it in a statement.

Fine words, but of course there are many caveats. Supplemental downlink using LTE will not be standardized ‘ and therefore not commercially viable outside a few operators’ hotspots ‘ until 2016, so it is not relevant to LTE, only to LTE-A. And that also means that a comparison with 802.11n is facile, because in two years’ time, there will be extensive deployments of the new 802.11ac gigabit WiFi release, and another extension, 802.11ax, geared specifically to dense environments, will be in draft form.

However, Docomo is making a statement of confidence in LAA, and Onoe continued: “We aim to contribute to the standardization of this technology, which inherits the highly advanced features of LTE, to further enhance the global user experience with wireless broadband.”

The partners made the same bandwidth available to LTE and WiFi and said they saw ‘several times throughput improvement of LAA over WiFi for both cell-medium and cell-edge users in most of the scenarios. In addition, the cell capacity gain was approximately 1.6 times that of a single-cell scenario, with better coverage extension by LAA over WiFi also observed.’

In many ways, these results, while interesting, miss the point. The debate is not really about which technology is superior ‘ of course, LTE has some advantages over WiFi, but the real issues are whether the two can coexist (because however superior the 3GPP platform can be shown to be, WiFi will always be there); and whether it is appropriate for a closed standard, available only to a small set of operators, should be working in unlicensed spectrum.

Huawei and Docomo acknowledged that the next phase of their work would need to focus on coexistence, and the 3GPP has also placed emphasis on this, since it will be vital to regulatory approval for 5GHz LTE devices and equipment. Some vendors might be secretly pleased if their technology reduced the performance of WiFi, but most operators, increasingly making WiFi part of their HetNet mix, would not.

The coexistence problems revolve around the fact that LTE transmissions are scheduled by the base station, while WiFi implements ‘listen before talk’ (carrier sense multiple access) to support sharing between different users. That has led to fears, even within the 3GPP, that LTE-LAA would interfere with, or even block, WiFi transmissions. Qualcomm recently updated its proposal with several coexistence features, designed to reduce the risk to WiFi, particularly in China, Korea and the US, whose regulations make conflict particularly concerning. And interference mitigation is a high profile element of some of the other submissions to the 3GPP, from players like Verizon and Ericsson.

All this is not to downplay the importance of SDL, and of cellular offload to networks in unlicensed spectrum. Both are significant in the cellcos’ attempt to balance capacity and cost, but should not meet in the 5GHz band, which is vital to enabling global WiFi, something of value to all kinds of service providers and end users. Other bands may support the 3GPP’s aims ‘ indeed, the body is careful to emphasize that LAA will be standardized as a frequency-agnostic technology which could be targeted at various licence-exempt or lightly licensed areas.

One vision for LTE-LAA is that it supports a network of small cells in 5GHz (since small cells work well in high bands, with their low range and high capacity). These would all be linked to the primary cells in the licensed spectrum, achieving the HetNet concept of a separate layer for macro and small base stations, and reducing the problems of coexistence when the metrocells are in the same spectrum as the macro. However, this idea could equally be implemented in another, less contentious band, 3.5GHz (licensed but underused in most countries outside north America) being the prime candidate. And simple SDL has been tested in various frequencies, including unpaired 700MHz (AT&T and Qualcomm) and the L-band mobile satellite spectrum (Orange France). Indeed, European standards body CEPT is harmonizing the L-band for SDL purposes.

Nevertheless, the early focus of the 3GPP’s work on standardizing LTE-LAA is firmly on 5GHz. A workshop was held in June to share ideas and was described as a ‘fruitful contribution to the start of work in the project’ by Dino Flore, chairman of the workshop and of the 3GPP TSG-RAN effort (from Qualcomm, naturally). Interested parties are now working on proposals to define the scope of the work and this should be approved at a meeting next month, kicking off the real process.

Important results from the June workshop were that the early focus would be on 5GHz, but the core technology should be frequency agnostic; the solution should be as globally applicable as possible; it should work indoors and outdoors; and it should enable fair coexistence with WiFi and other technologies. Most vendors and operators believe the first work should define a secondary cell used for supplemental downlink only, and then move onto a version that could also support supplemental uplink capacity. Flore insisted, in his summary of the event: ‘Initial results suggest that, when augmented with the appropriate coexistence mechanisms to operate in unlicensed spectrum, eg Listen-Before-Talk, LTE can effectively coexist with Wifi and outperform it in terms of spectral efficiency.’

Companies which gave presentations at the workshop, and so are likely to be among those driving the standard, included Nokia, Huawei’s HiSilicon unit, InterDigital, Qualcomm, Sony, Texas Instruments, Alcatel-Lucent and its Chinese venture ALU Shanghai Bell, Samsung, Hitachi, Intel, CATT, ZTE, Broadcom, Cisco, BlackBerry, Fujitsu and LG.

Operator submissions reflected the weighting of this effort towards the US and east Asia, with contributions from AT&T, T-Mobile USA, China Unicom, Docomo, KDDI and KT, as well as a lone European representative, TeliaSonera (though Europe’s standards body ETSI presented too). There was also a presentation of potential use cases, jointly given by Verizon, China Mobile, Huawei and Ericsson; and a review of regulatory situations by most of the key vendors (ALU, Ericsson, Huawei, HiSilicon, IAESI, LG, Nokia, Qualcomm) plus Docomo. And the cable community, which is often spearheading the use of carrier-class WiFi as a weapon against cellcos, also had its voice heard, in a presentation by CableLabs and WiFi vendor Ruckus Wireless.

The more items are added to the long menu of options which cellular operators need to keep their LTE models competitive, the more complicated the standards process becomes, at a time when speed is often of the essence (the WiFi standards have the same issue of course, though their community has achieved better ways of supporting and even certifying pre-standard equipment without leading to dangerous fragmentation). The bloated nature of the upcoming LTE standards is sure to create delays in commercial equipment, and indeed, the date for 3GPP Release 12 to be frozen has been pushed from September until at least December, or possibly March 2015. That is likely to have a knock-on effect on Release 13 (12 and 13 are the core LTE-Advanced releases), which is currently slated to be frozen in March 2016. Standards-based equipment and devices usually start to appear from a year after the freeze date.

Said Tatesh, 3GPP standards leader at ALU, said in an interview that Release 12 has become particularly complicated. Work began in December 2012 but turned into a ‘very long release’ compared to Release 10, on which the 3GPP spent only 12 months, and Release 11, which took 18 months. Release 12’s key areas of focus include some very complicated, though potentially business-critical, technologies, grouped in three broad categories ‘ HetNet, multi-antenna/MIMO, and support for diverse traffic types (including internet of things). Tatesh told FierceWireless that the main sticking points had been HetNet dual connectivity (in which a device can link simultaneously to macro and low power layers); and device-to-device communications including automatic discovery and peer-to-peer links.