Downloading the newest Wi-Fi protocols: 802.11ax and 802.11ay explained

[nir]

Wi-Fi protocols are more confusing than ever, so here's the TL;DR.

 

Look, Wi-Fi still kind of sucks. And marketing excesses aside, its worst problems all revolve around airtime distribution among multiple devices.

 

Unlike LTE (the protocol cellular data uses), 802.11 WI-Fi is a protocol with no central management, which leaves all nearby devices duking it out for airtime like angry, unsupervised toddlers. There's only so much you can do to fix this problem without radically overhauling and replacing 802.11 itself—but as new 802.11 protocols emerge, they do their best.

A brief overview of the alphabet soup

If you don't deal with this stuff for a living, it's easy to get lost in all the different Wi-Fi protocols in the ether today. New additions have been released in sort of alphabetical order, but some are backwards-compatible and some aren't. Some are "mainstream" and have broad consumer device support, and some are offshoot technologies rarely to be seen in anything you can buy at a big box store. It's kind of a mess.

 

If all this isn't bad enough, the Wi-Fi Alliance has not-so-helpfully decided to replace some—not all!—of the 802.11 designations in consumer marketing with a supposedly simpler scheme. 802.11ac, which most of us are using now, becomes "Wi-Fi 5" under this new scheme. 802.11ax will be marketed as "Wi-Fi 6." This new numeric designator conveniently ignores some protocols, unfortunately: neither 802.11ad nor 802.11ay will get "Wi-Fi Numbers" at all.

 

Designation	Spectrum	single-MIMO PHY	notes
802.11a	5 GHz	54 Mbps	There was almost no consumer device adoption; it was prevalent in very early 2000s enterprise.
802.11b	2.4 GHz	11 Mbps
802.11g	2.4 GHz	54 Mbps
802.11n	2.4 GHz / 5 GHz	144 Mbps / 300 Mbps	802.11n devices must have a 2.4 GHz radio; one or more 5 GHz radios are optional.
802.11ac	5 GHz	433 Mbps	802.11ac protocol is 5 GHz only, but in practice all 802.11ac devices also offer a 2.4 GHz 802.11n radio.
802.11ad	60 GHz	~5 Gbps
802.11ax (draft)	2.4 GHz / 5 GHz	~500 Mbps	It's a draft protocol scheduled to be ratified in 2019. It covers 2.4 and 5 GHz, with provisional support for 1-6 GHz at a later date.
802.11ay (draft)	60 GHz	~ 40 Gbps

 

Confused about the PHY column in the table above? You probably should be. PHY is the "PHY"sical transport layer speed of a Wi-Fi connection—but you can't actually move data across the link that fast. Actual data transmission rates can be anywhere from 1/3 to 2/3 of PHY on a completely healthy link in reasonable transmission range. And as you move further away from an access point, you can rapidly see transmission rates falling to 1/10 of PHY or worse.

 

Ideally, you'll also see the PHY itself fall off as you move further from the nearest access point—a lower QAM means lower PHY and throughput, but longer reliable connection range—but the connections your devices negotiate between themselves frequently aren't optimal.

 

Adding to the confusion, many of the protocols in this table support varying channel bandwidth settings, with higher bandwidth meaning higher throughput, but fewer available channels and more problems with interference (and multiple MIMO streams as well). The table above assumes a single MIMO stream and the most common (not necessarily the largest) QAM and channel width settings.

About MIMO streams

MIMO is an acryonym for Multiple Input / Multiple Output; it's a way of using multiple antennas to send multiple spatial streams of data from a single radio on a single channel. Broadly speaking, there are two types of MIMO— SU-MIMO, and MU-MIMO. The SU stands for Single-User, and no matter how many streams a device has available, it can only talk to one other device at a time. Got an 8-stream 802.11ac router that's currently talking to a single-stream, non-MU-MIMO 802.11ac device? Tough; you're only getting a single stream worth of transmission to that single device no matter how many other devices you've got clamoring for airtime.

 

MU-MIMO is Multi-User MIMO, and as the name suggests, it means that an access point can divide up its available MIMO streams between multiple clients. For example, a 4x4:4 access point (four 2.4 GHz MIMO streams, four 5 GHz MIMO streams, and four antennas) can simultaneously "talk" to one 2x2:2 laptop, and two 1x1:1 phones or tablets.

 

There are quite a few catches with this; the biggest is that all devices currently "talking" must support MU-MIMO. 802.11ac's implementation of MU-MIMO is download only; so while an 802.11ac MU-MIMO router can simultaneously deliver data to several MU-MIMO enabled client devices, any time one of them wants to request data from the router, all other traffic comes to a screeching halt.

 

Enlarge / MIMO is a scheme allowing the use of multiple antennas to simultaneously transmit or receive multiple spatial streams of data, using a single radio and channel.

All of this MU-MIMO stuff already exists in the real world, but vanishingly few of us have ever benefited from it. MU-MIMO capable routers are increasingly common, but MU-MIMO capable client devices are still rarer than hen's teeth. There's a reason for that, as Chuck Lukaszewski demonstrated in a WLPC presentation from 2016: MU-MIMO requires significantly increased power usage compared to SU-MIMO. And that makes it much less attractive in battery-powered devices.

802.11ax, the next "normal" Wi-Fi

Spoiler alert—despite the obvious congruency of 802.11ax and 802.11ay, one isn't a successor to the other. 802.11ax is the protocol which will succeed 802.11ac as the next mainstream Wi-Fi protocol we all use at home and in coffee shops, hotels, and so forth. If you're into the Wi-Fi Alliance's new, supposedly simpler marketing, it's Wi-Fi 6, compared to today's Wi-Fi 5... assuming we conveniently ignore the Wi-Fi 4 that's on the 2.4 GHz radio all of our Wi-Fi 5 devices also have.

 

Just as 802.11ac is backwards-compatible with 802.11n (which is itself backwards-compatible with 802.11b and 802.11g), 802.11ax will be backwards-compatible with 802.11ac. You won't have to go replace all of your devices when you buy a new 802.11ax router, or vice versa. That's the good news; the bad news is that nearly all of 802.11ax's significant improvements require support on both the access point and the client device side, and you won't be able to buy 802.11ax client devices (phones, laptops, accessory cards) for quite some time after the routers are everywhere.

 

802.11ax is still a draft protocol. While it won't be fully-ratified until sometime in 2019, a few routers supporting the 802.11ax draft are already publicly available, including Netgear's Broadcom-powered RAX-80 and Qualcomm-powered RAX-120 models. (I hope to have an RAX-120 available for testing soon—but neither you nor I are likely to have any 802.11ax devices to test it with, unfortunately.)

Netgear's RAX-120 departs from the company's usual "upside-down spider" motif, opting instead for "Darth Vader's Imperial Shuttlecraft".

Netgear stock product image

Why PHY?

There isn't much improvement in PHY going from 802.11ac to 802.11ax; instead, the big improvements are more subtle than that. Thankfully, these improvements are also more useful. The biggest problem most people have with Wi-Fi, increasingly, isn't so much "it doesn't go fast enough" as is it "it can't handle multiple devices well enough." This is exactly where 802.11ax should shine brightest, thanks to several new or improved features. The most important ones are OFDMA, bi-directional MU-MIMO, trigger-based random access, spatial frequency reuse, and target wake time.

OFDMA

In a nutshell, OFDMA allows an access point to offer each connected client device one or more Resource Unit (RU). Each RU consists of several extremely narrow-bandwidth "tones," or subchannels, within the overall Wi-Fi channel itself. And this makes it possible for multiple client devices to transmit simultaneously, whenever they feel like doing so, without having to worry about packet collisions. This also greatly mitigates the problem of extremely long-range, low-throughput devices that nevertheless end up hogging all the airtime because their QAM rate is so low, or their error-and-retry rate so high.

Bi-directional MU-MIMO

What OFDMA does for the spectrum, bi-directional MU-MIMO does for the available spatial streams. 802.11ax's new implementation allows multiple client devices to share the available streams from the access point for both download and upload. I hesitate to make any bold, sweeping predictions without real-world devices in hand, but this stream-sharing should stack multiplicatively with OFDMA's channel-sharing, potentially allowing (number of available RUs) × (number of spatial streams) total clients to simultaneously transmit without interfering with one another.

Trigger-based Random Access

Trigger-based random access works with OFDMA to improve situations with multiple client devices even in the absence of directly assigned RUs. In addition to directly assigning RUs to individual clients, the access point can designate multiple RUs as "randomly" available. A device which wants to upload a little bit of data but doesn't have a directly assigned RU can randomly pick one of these pooled RUs. With luck, they'll pick different RUs from the random pool even when several devices try to transmit simultaneously. If two particularly unlucky clients do decide to transmit on the same RU at the same time, they'll fall afoul of CSMA/CA and have to back off and retry after a randomly determined pause, just like they would with current protocols.

Spatial frequency reuse

Spatial frequency reuse is yet another feature which improves Wi-Fi performance and predictability in RF-dense environments. Without this feature, a device or access point must remain "silent" while any other device transmits, even if that device doesn't belong to the same network. With spatial frequency reuse, coloring allows a device to decide whether simultaneous transmission is permissible.

 

Let's say there are four devices in play; A, B, X, and Y. Device B would like to transmit to access point A, but device Y is already transmitting to access point X on the same spectrum. This isn't normally permissible, but if B can determine that Y isn't transmitting to A, then B can transmit to A simultaneously—so long as B adjusts its output power low enough that the B-->A transmission won't interfere with the Y-->X transmission.

 

Maybe this won't help someone with 50 devices and one router who lives in a farmhouse with no competing networks, but it's a serious upgrade for someone in an apartment complex surrounded by competing routers on all sides, or someone in a big house with tons of devices and multiple access points.

Target Wake Time

TWT makes me hopeful we might actually see the rest of these features implemented in the small, portable devices we actually use (unlike the current generation of MU-MIMO). It gets a little hairy trying to explain how and why it works, but basically it schedules devices to explicit, non-competing (or at least less-competing) timeframes. This in turn allows battery-powered devices to leave the radio "asleep" for longer periods, significantly decreasing how much power is consumed by active Wi-Fi connections.

802.11ay, the next "Wi-Gig"

Despite some very hopeful claims I've heard from chipset manufacturers, I don't really expect 802.11ay to ever become "normal Wi-Fi." Like 802.11ad (aka "Wi-Gig") before it, it operates solely in the 60 GHz spectrum. The great thing about 60 GHz is that it punches through clear air very easily, making high-throughput, long-range connections much simpler than they are with 2.4 GHz or 5 GHz radios. The unfortunate thing about 60 GHz is that it can't punch through solid obstructions such as walls, furniture, or human bodies very well at all.

 

Assuming you've got a clear line of sight from access point to client and no interference from competing 802.11ay devices on the same channel, it's not unreasonable to expect overall PHY rates of 10 Gbps or more, with actual data transmission rates of well over 1 Gbps—as fast or faster than current wired Ethernet connections. You should, repeat should, be able to sustain a reasonably reliable 802.11ay connection on the other side of a single wall from the access point, but you won't see transmission rates anywhere near that high. A human also walking through the path would be enough to further degrade or even completely kill throughput.

Enlarge / This is the now-defunct Wireless Gigabit Alliance's "Wi-Gig" marketing logo. Like 802.11ad itself, it never saw broad adoption.

Wikimedia Commons

It's pretty hard to find 802.11ad devices; the majority of the ones I've seen are proprietary "wire replacement" devices such as laptop docks or wireless HDMI bridges.

 

Netgear includes an 802.11ad radio in their stratospherically-priced Nighthawk X10 router, but I never could find a usable 802.11ad network card to connect to it with. I bought an Intel 17265 from a third-party vendor on Amazon shortly after receiving an X10 from Netgear for testing, but I never could source the 60 GHz antenna assembly required to make it work after shoehorning it into a laptop that had never been built for it. You can bridge two X10 Nighthawks together over their 802.11ad radios... but it's not going to do you any good unless, for some reason, you have a space to span that's too long range for 2.4 or 5 GHz, but doesn't have any interior walls.

 

In the real world, it's difficult to imagine a Wi-Fi bridging use case better suited to a pair of X10s with 60 GHz backhaul—at an eye-watering $900 or more for the pair—than to a more typical mesh kit with a high-performance 4x4:4 5 GHz backhaul, such as Netgear's Orbi RBK-50, Plume's Superpods, or the newer Gryphon, all of which use Qualcomm's QCA-9984 chipset to achieve near-gigabit throughput even when walls, furniture, and people are in the way.

 

802.11ay is, for the most part, just an even-faster 802.11ad, but with the same limitations. 802.11ay can bond channels and streams, but it still fares poorly if it can't get a clear line of sight.

 

I received an enthusiastic Broadcom presentation envisioning a ceiling-mounted 802.11ay access point with wired backhaul in the ceiling of every room of an apartment or house, but I wouldn't hold my breath waiting for that reality. I think we're more likely to see 802.11ay in long-distance directional Wi-Fi links from specialty devices than in general-purpose, omnidirectional Wi-Fi. It might very well be cheaper and easier to deliver high-speed broadband Internet to underserved rural areas with 802.11ay and tall towers than with fiber. The low penetration and high throughput might also make for a compelling Bluetooth replacement. Eventually.

Should you throw everything away and start over?

The short answer is "no, probably not." If you're not already in the market for a new router, don't bother. But if you are already in the market, you might want to consider shopping specifically for an 802.11ax draft router, though I wouldn't even make that a prime consideration at this point. Most users will be far better served by a good 802.11ac mesh product than a single 802.11ax draft router. Once the 802.11ax protocol is formally ratified in 2019 and we start seeing it pop up in the majority of routers and Wi-Fi mesh kits, it'll get much more important to make sure you're future-compatible—and once most of your devices support it, you won't want to live without it.

 

802.11ay, like 802.11ad, is mostly a curiosity for now. If you're in the market for a high throughput line-of-sight connection that won't work through walls, and you see a device in both 802.11ay and 802.11ad flavors, by all means pick the 802.11ay if it doesn't cost too much more. Just don't expect an all-Wi-Gig home anytime soon, if ever.

Source