Your browser is not supported. Please update it.

27 July 2018

EHT effort could challenge 5G latency and push WiFi into 6 GHz band

The latest project within the IEEE 802.11 standards effort (the basis for WiFi) could challenge 5G’s projected latency, throughput and coverage targets, even in unlicensed spectrum. However, the EHT (extreme high throughput) goal of delivering these capabilities to affordable equipment would depend on 6 GHz being freed up for unlicensed usage in most of the world – and it will only be ratified in 2023.

EHT shows how the 3GPP and IEEE are pursuing similar goals, even if they seem as unprepared as ever to converge their work.

A couple of years ago, there were real hopes that there might be meaningful cooperation between the main standards bodies active in wireless connectivity – 3GPP and IEEE 802 – to achieve a more coherent multi-RAT 5G platform. In fact, the situation remained the same as in 4G – completely separate, if often parallel, tramlines for cellular and WiFi.

The two technologies will continue to be intertwined, by operators using a combination of both, with a probable move towards common multi-RAT cores and infrastructure, helped by virtualization and slicing. And of course, there are plenty of cooperations between the 3GPP and IEEE communities on tactical technologies to enable coexistence, even if they remain very separate on core standards.

There are some who argue that, now 4G and, one day, 5G can work in unlicensed bands, there is no need for WiFi. But WiFi has its own strengths and a well-established open ecosystem which it would take many years for the 3GPP community to emulate.

So while a common physical layer platform may have eluded us again, many of the goals which 5G New Radio (NR) and the next wave of 802.11 standards are pursuing are similar, which should help operators to combine the two technologies in the future in various ways, depending on their use case, their access to spectrum and so on.

The more WiFi becomes a central part of the ‘5G’ landscape, the more it will seek new spectrum options, just as 5G is; and the more its own chip vendors will take a place in the once-closed mobile ecosystem, introducing their own economics to the space.

The ‘5G’ era WiFi standards will start with 802.11ax and 802.11ay, multi-gigabit specifications for 5 GHz and 60 GHz bands respectively. The first draft of 802.11ax has been approved, though final ratification will take at least another year, and 11ay is behind that.

Now, 802.11 engineers are already working on a new standard, currently called EHT (extreme high throughput), which would be the first to push WiFi into the 6 GHz band. The green light to start working on this extension to the 802.11 family was activated in May.

In voting to go ahead with EHT, the group only set out high level goals – boosting throughput and capacity by four times while improving efficiency and latency. Detailed goals will be set by a study group over the coming months and official work will begin in May 2019.

According to Intel, the latency goals may be more aggressive than those of 5G NR, at least in its Release 15 guise. An Intel analysis found that WiFi could be capable of delivering millisecond latency, but to do so, it needs emission and synchronization controls to maintain that level in unlicensed spectrum – so, even when there is interference or multiple users.

EHT is also interested in adding features to allow access points (APs) to coordinate to achieve better coverage, reflecting cellular work on technologies like CoMP (coordinated multipoint). In WiFi’s case, it would be likely to formalize and extend work on master and slave APs in the WFA, which targets whole-home coverage.

The new standard is also expected to support 16 streams, twice as many as 11ax; and the ability for APs to support multiple simultaneous transmissions across multiple bands.

However, since EHT is unlikely to be completed until 2023, its supporters are pushing for an interim option to accelerate WiFi’s readiness to move into 6 GHz. The proposal is to devise standard way to retrofit 802.11ax to work in 6 GHz in 2020, when it is hoped the band will be cleared for unlicensed use as well as cellular, in the USA and some other areas.

This would extend WiFi’s range in an upwards direction from 5 GHz, and allow far wider bandwidths to support very high data rates and device densities (the latter is a key focus of 11ax, for scenarios such as stadiums). Full EHT, when it arrived, would provide even better performance because it will have been designed from scratch to use the 5 GHz/6 GHz combination optimally.

The push for interim solutions reflects the increasing misalignment between slow standards-setting processes and the speed of change in the wireless market. The 11ax spec has taken four years to get to first draft – the point at which companies can make pre-standard chips and equipment without significant risk that these will not be easily adaptable to the final specs. That is likely to make it one of the longest 802.11 projects ever, though the last three major WiFi standards did take at least five years each.

While technologies are increasingly complex, these long processes frustrate the industry at a time when customers are demanding ever-higher data rates, and the WiFi ecosystem feels the pressure – with LTE moving into 5 GHz and other shared bands – to keep pace with the 3GPP.

Laurent Cariou, a WiFi standards specialist at Intel, told EETimes: “With EHT, we are trying to put in place a new process to get to a two-year cadence.”

That would mean a steady drip of releases, which might help address the risks of fragmentation, when companies produce equipment too far ahead of a standard set of specs; or an industry shift to alternative processes including open source.

Already, the WiFi Alliance (WFA) works to make technologies useful to its base before they are fully ratified and often tests and certifies interim subsets of standards, or adds extensions of its own (for security, for instance).

It will kick off its certification program for 11ax based on its draft specs in 2019, and another, based on the final standard, in 2021. Some engineers are pushing for this pattern to become embedded in the IEEE 802.11 process, with a two-year cycle, each carrying WFA certification.

“There’s a consensus that people want to move faster,” said Michael Montemurro, director of standards at Blackberry. “Some want to move to two years and others are not so committed, so it’s hard to say how it will sort itself out.”

The other potential roadblock for EHT would be failure to get 6 GHz open for unlicensed use. There are proposals to open up as much as 1.2 GHz in this band for this purpose but some of the spectrum is already in use, and hotly defended, by satellite and public safety networks as well as being used for mobile backhaul. Cariou said discussions with regulators in the USA and Europe are looking promising and some EHT members are already working on APs that simultaneously send and receive data over 2.4 GHz, 5 GHz and 6 GHz bands.

The 3GPP is also working on an implementation of 5G NR for unlicensed bands which could include 6 GHz by the time 5G-Unlicensed is ratified. Qualcomm has been driving this effort, with hopes of getting it into Release 16 next year, though it may also have to wait until Release 17. “I think it’s in a study item, but they will move fast, probably faster than 802.11,” Cariou commented to EETimes.

The 802.11ax standard:

The first draft of the 802.11ax specs have been approved. They will boost data rates and be particularly efficient in very dense environment, which means they are addressing some of the key issues – multi-gigabit speeds and extreme device density – that 5G is also tackling.

But the IEEE standard has had a rockier road than Release 15 in many ways. Its first and second drafts failed to get the required 75% approval from the 11ax working group in November 2016 and September 2017, dashing some vendors’ hopes of being close enough to a full standard that they could ship early products by the end of 2017.

It is now more than four years since the 11ax process kicked off, longer than the current high speed standard, 802.11ac, took. It will still take until late 2019 to complete the final ratification of the standard, including a sponsor ballot, though at this stage, the specs will be sufficiently frozen for vendors to feel safe to develop products, without the risk of significant reworking to conform to the final standard.

The physical layer specs have been stable for some time, enabling chip designers like Intel, Qualcomm and NEC to be sampling pre-standard 11ax units already. Most recent changes have related to the media access controller (MAC) and any further modifications should be able to be implemented with software updates.

The average time for an IEEE process is three years, though some WiFi generations have taken longer – for instance, 802.11n, the predecessor to 11ac, which was attempting a major step change in performance and architecture, and was also beset by the politics of rival vendor approaches.

Like 11n, 11ax is very complex and has ambitious objectives. It is targeting an increase in real user data rates of up to 30%, along with a fourfold reduction in latency. It also promises to deliver up to four times more overall data in the same spectrum as 11ac.
That means harnessing cellular-style OFDMA modulation with 1024QAM, plus expanded support for multiuser MIMO.

Access points based on the draft specs are expected to ship in the last quarter of this year. However, the late start means the technology will wait until 2018 to gain scale. Researchers at dell’Oro have reduced their forecast for enterprise 11ax AP shipments from 750,000 during 2018 t0 250,000. They expect that first wave of sales to be driven by lower cost, pre-standard units and also by China, where Huawei and H3C, a spinout from HPE China, have expressed plans to be first to market with 11ax.