Next-generation 802.11ac wi-fi: The state of play

The first routers based on the first draft of the next-generation wi-fi standard, 802.11ac, appeared at CES in 2012, sporting multiple antennas (the technical term is MIMO, standing for Multiple Input Multiple Output, which means sending and receiving more than one signal at a time).

In January 2013, manufacturers added one of the key 802.11ac features — beamforming — to their access points. Beamforming is what allows 802.11ac routers to deliver a wireless signal straight to a device rather than bathing the entire surrounding area in the signal intended for that device. Although it's supported in the previous-generation 802.11n, the new standard is more efficient — in part because it only includes one way of doing it rather than supporting several possible options.

The 802.11ac standard has yet to be formally ratified by the IEEE, but the Wi-Fi Alliance launched its certification program at the end of June, and the first 802.11ac-compliant devices are now reaching the market (all the latest MacBooks, including the Air, plus the HTC One and Samsung Galaxy S4 for example).

However, that doesn't mean you'll get all the features of 802.11ac with even these new devices. And while the standard promises the major improvements in battery life as well as bandwidth, there's still some way to go according to Matthew Gast of the Wi-Fi Alliance (also director of product management at Wi-Fi supplier Aerohive Networks).

Getting a new wireless standard to market takes time, but 5GHz 802.11ac has been a model of cooperation compared to the lengthy and argumentative progress for 802.11n, its 2.4GHz/5GHz predecessor. Instead of starting with a complete proposal and arguing until 75 percent of the members of the standards committee finally agreed (which meant including four different ways to do carrier signals and multiple implementations of beamforming, for instance), the committee started with something much closer to a specification requirements document.

"We said, 'We need this many encoders, we need this form of MIMO'," as Gast puts it. Deciding on the design together turned out to be a much more efficient way of agreeing on the standard, and the collaboration meant there was "a lot more consensus about what was going to be in the standard" much earlier on.

What's taking the time is actually building the new technology.

Still to come: multi-user MIMO

Some of the improvements come from newer, faster processors in newer access points, or just from having had the experience of building 5GHz radios for 802.11n devices for a few years (so when we connected a Buffalo 802.11ac access point to our small office network, we saw both faster and more reliable wi-fi with 802.11n devices and a speedup on our BT Infinity connection because the access point could accept data from the line faster).

But the rest of it is physically targeting wireless connections to different devices using different amounts of bandwidth and juggling connections with different priorities: that's called multi-user MIMO, and the wireless chips to do it aren't ready yet.

802.11ac uses its multiple antennae and beamforming capabilities to send targeted signals to multiple devices; a lower-bandwidth device such as a phone gets a lower-bandwidth signal without forcing the signals sent to a higher-power, higher-bandwidth tablet or a notebook with a full-bandwidth 802.11ac adapter to drop down to the same lower bandwidth the way 802.11n does.

Single-user MIMO, as supported by this year's 802.11ac access points and devices, uses beamforming to send the signal efficiently to one device at a time; next year we'll get chips that can send signals to multiple devices at once — and when you can do that, you save power on your devices because the fast ones don't have to keep the radio on for as long. You can also make the signal more efficient by only sending it in the direction of devices that are transferring data and keeping the "null" areas where any devices that aren't connecting are located (reducing interference, which can slow the connections down).

You won't be able to upgrade today's 802.11ac device to work with multi-user MIMO because it needs more than just a firmware update.

There's also a question of priorities. If you're making VoIP calls, you want the access point to prioritise your voice calls so the line doesn't break up. "But it's not just about voice going out first," Gast says. "If voice goes out first in this direction, what else I can squeeze out in some other direction at the same time?" That might be a connection that's a lower priority but needs to go in a direction where I can send it without interference. An 802.11ac system will need the chip to pull information out of a lower-priority queue to transmit it off to the left while the higher priority signal is still going to the device over on the right.

You won't be able to upgrade today's 802.11ac device to work with multi-user MIMO because it needs more than just a firmware update. For one thing, there's a lot of work to do just measuring where all the devices are physically, both in relation to the access point and in relation to each other (so you can pick the best channel to transmit on to avoid interference if, say, one person is using two devices). When the access point asks one of those devices to deliver a channel measurement, it needs to come back almost in real time, and multiple pieces of information must be sent at the same time. The chips to manage that are still being developed and should be in hardware next year.

Until then, the maximum speeds wi-fi manufacturers are hoping for remain theoretical, although Gast is expecting good results. "You won't get the absolute top possible speed; you have to take some off the table because you're transmitting multiple beams and you have to take out the overhead of the channels — but the aggregate will still be faster. I'd say that's a good bet. And as an industry we're investing a lot of money in that bet!"

Next-generation 802.11ac will enable the more effective use of all the phones and tablets we seem to want to take to work these days. It will make an even bigger difference to the "Internet of Sensors" — as I prefer to call the Internet of Things, since it's the sensors that collect and report the data of intertest. I don't want all the sensors I care about connecting over Bluetooth — even Bluetooth Low Energy — because all I get is a bubble of sensors around me.

What's required is a mesh of sensors reporting all the time, whether I'm there with a phone or not. Wi-Fi is the best way to do that — especially 802.11ac, which Gast suggests thinking of as almost a wi-fi switch. Hospitals are increasingly connecting monitoring devices by wi-fi, and even treatment devices will switch to wireless. Imagine 25,000 drug infusion pumps on a hospital wireless network: getting an efficient signal to lots of devices becomes literally vital in this case. If you don't need that, you might want to wait until 802.11ac isn't significantly more expensive than 802.11n — unless you consider even a little extra wireless speed is worth the price.