Your browser is not supported. Please update it.

19 November 2020

Will WiFi 7 arrive too late for ultra-low latency consideration?

The ability of WiFi to handle the ultra-low latency applications addressed by the 5G Ultra Reliable Low Latency Communications (URLLC) use case category will determine whether it can expand significantly from its current enterprise LAN base and make an impact in the Industrial IoT (IIoT). That certainly seems to be preoccupying standardization efforts for the next generation, WiFi 7, due to start coming on-stream late in 2024. Currently the preceding generation, WiFi 6E (the extension of WiFi 6 adding support for the 6 GHz band), is only starting to be deployed following regulatory approvals.

To an extent, WiFi 7 can be seen as completing unfinished business or extending capabilities introduced under WiFi 6, such as improving efficiency of OFDMA. This has been available for cellular since 4G and was introduced for WiFi 6, allowing a number of users each to transmit data split across multiple sub-channels at dispersed frequencies to improve throughput and maximize utilization of overall spectrum, also reducing power.

While OFDMA does not of course magically conjure up additional spectrum out of nowhere, it effectively increases capacity by reducing susceptibility of a whole larger channel to fading, which attenuates signals more at some frequencies than others. By splitting a user’s transmission across multiple sub-carriers at different frequencies, the impact of fading is confined to just one or a few of those. WiFi 7 will improve on this further by allowing these multiple sub carriers to be allocated more flexibly to individual WiFi stations, or users, effectively.

Similarly, MIMO technology, already supported under WiFi 6, will be doubled in capacity under WiFi 7, which will support 16 single-user MIMO (SU-MIMO) and multi-user MIMO (MU-MIMO) spatial streams, rather than eight. Taken together, the improvements will increase theoretical maximum data rate by about three times to 30Gbps under WiFi 7. Claims that WiFi 7 will boost peak data rates five times need taking with a pinch of salt and in any case users are concerned with real world rates, which are unclear at this stage, but a threefold boost would seem to be the consensus view.

It is true, though, that further effective improvements could be delivered through another feature that will be new to WiFi 7, although already available for cellular as carrier aggregation, which came in under 4G. Indeed, carrier aggregation will be key to coexistence between 4G and 5G services, by giving operators the flexibility to combine 4G carriers with each other, or with 5G to boost capacity as demand for 4G falls off.

Under WiFi 6E, the equivalent of carrier aggregation, called multilink operation, became desirable because now there were three frequency bands, 2.4 GHz, 5 GHz and 6 GHz, to choose from. Currently users or their devices still have to select between these bands, so their existence just increases overall capacity rather than individual data rates.

Under WiFi 7, a client device will be able to connect across all three bands at the same time using the available access points (APs). As well as increasing data rates, this will also allow the links to be combined to support different use cases, so here again there is an element of catch-up with 5G.

It was notable, however, that at the Wireless Broadband Alliance’s recent Wireless Global Conference, the drive to match 5G’s ultra-low latency capability seemed to top the agenda, or at any rate list of concerns to be addressed. Intel’s CTO of wireless communications, Carlos Cordeiro, was among those pushing this point, explaining that as critical real time data continued its migration closer to users with edge compute, in turn the wireless access domain comes under growing pressure to deliver as low latency as possible on its side.

Cordeiro therefore underlined the importance of deterministic low latency for applications that require a quantifiable upper delay bound, such as UAV control, extended reality (ER) and remote surgery.  This deterministic low latency will be delivered in part through the other features, such as multilink operation, which will allow greater scope for prioritization and optimal use of spectrum for the most time critical applications. Multiple AP Coordination is another impending WiFi 7 feature that will help out, by enabling network wide optimization of traffic, rather than the current contention between users or APs.

Some other new features will also assist, such as Hybrid Automatic Repeat Request (HARQ), which combines forward error correction (FEC), where redundant bits are added to cater for a degree of loss on the fly, with ARQ (Automatic Repeat Request), which allows receivers to detect whether corruption has occurred but without ability to reconstruct the data. Under ARQ on its own, receivers detecting corruption request a message or data packet to be resent, while with FEC on its own re-transmission does not occur.

In HARQ, the data is initially encoded with an FEC code and the parity bits that allow corruption to be detected under ARQ are only transmitted themselves upon request when an erroneous message is detected. The idea is to maximize the chance of transmitting data correctly in one pass and minimizing subsequent retransmissions, which naturally bears down on latency.

The one fundamental change specific for latency under WiFi 7 will be incorporation of time-sensitive networking (TSN) capabilities, which are also coming into 5G. This includes techniques for time synchronization, traffic shaping, scheduling, ultra-reliability, and resource management, tailored towards WiFi capabilities such as the multilink operation and the multi AP coordination.

Traffic shaping will be especially important for TSN, through two mechanisms in particular, non-preemptive prioritization and especially pre-emptive prioritization. The principle is simple enough, non-preemptive prioritization involving promotion of any urgent packet that arrives in the queue ahead of non-urgent ones already waiting. Pre-emptive prioritization goes further still by interrupting an ongoing transmission of non-urgent packets the instant an urgent one arrives in the queue.

A study published late 2019 at the Pompeu Fabra University in Barcelona, Spain (Adame, Toni & Carrascosa, Marc & Bellalta, Boris ‘Time-Sensitive Networking in IEEE 802.11be: On the Way to Low-latency WiFi 7’), concluded that with non-pre-emptive prioritization policies, latency values over WiFi could be cut to 3.83ms, and as low as 2.58ms with pre-emptive prioritization. This is low enough for most time-sensitive communications, and with low variance the demand for deterministic low latency would be met.

There is the prospect then of WiFi stepping up to the plate for ultra-low latency operations, but whether it will be too late is another matter. Currently WiFi is rarely considered for such time-sensitive applications.