This article originally appeared in TidBITS on 2013-06-13 at 1:43 p.m.
The permanent URL for this article is:
Include images: Off

802.11ac Promises Better Coverage, but Won’t Hit Advertised Speeds

by Glenn Fleishman

Apple’s towering revisions to the AirPort Extreme Base Station and the Time Capsule tout the speed bump [1] that will come from its inclusion of the 802.11ac standard, the latest in wireless local area networking. Instead of (up to) 450 Mbps with 802.11n, your data will zip around at (up to) 1.3 Gbps*†ª⚔! (Note the parenthetical, asterisk, footnote, clarification, and proviso — it’s not that simple.)

Now, it is true that 802.11ac can offer higher speeds than 802.11n, and Apple’s implementation is technically capable of 1.3 Gbps. But as Apple notes at the bottom of the page, “Actual speeds will be lower.” I’ll say. In practice, I would wager that most home users and some business users will see only modest improvement in net throughput across their networks.

What 802.11ac should achieve, however, is far better coverage. Those who moved from 802.11g (a 2003-era standard) to 802.11n (released widely in 2007) remember what a difference that was. Dead spots in your home and office were suddenly lit up. Areas in which you had slow data rates but a good signal were now able to communicate at several times the previous data rate.

Apple doesn’t ignore range and coverage, but that’s really the benefit of 802.11ac for most users, not speed. Let’s dig into Apple’s claims.

Closed Course with Professional Drivers -- I object to Apple’s chart because it’s exceedingly misleading about average, typical performance, and sets expectations far too high. It’s as if Toyota advertised that its consumer sedans could travel at “up to 150 miles per hour!†” (“†Closed course with professional drivers on a sunny day with new tires on a flat straightaway.”) Most drivers will be lucky to average 15 to 30 mph in the city.

When Apple says that its implementation of 802.11ac can achieve up to 1.3 Gbps — and other manufacturers with beefier radio systems already say up to 1.7 Gbps — the reality is that a lot of conditions have to be met to achieve that raw data rate. And, as you well know from decades of network-technology advertising, dear reader, a “raw” data rate (often incorrectly called “theoretical”) is the maximum number of bits that can pass over a network. That includes all the network overhead as well as actual data carried in packets and frames. The net throughput is often 30 to 60 percent lower.

The key improvements in 802.11ac that give it the potential for higher data rates are:

If you do the math for Apple’s current 802.11n systems, you get an ideal case of 75 Mbps times two (two 20 MHz channels) times three spatial streams or 450 Mbps — that’s when using the 5 GHz band. With 802.11ac, Apple increases that: using 80 MHz (the biggest channel Apple opted for) with the improved encoding calculates out to 433 Mbps times three streams for roughly 1.3 Gbps. (Some companies that sell to corporations are already using four streams, which gets you that higher 1.7 Gbps advertised top rate.)

But we’ll rarely see that data rate.

Play Nice -- With all Wi-Fi standards, the technology tries to play nice with other networks and potential nearby interferers. Networks slow down when there’s other traffic nearby on the same or adjacent channels, unless the signals can clearly be discriminated from one another. MIMO helps with this, because the multiple antennas allow overlapping signals to be teased out. (Some people argue there’s no such thing as interference, but merely a limit to how effectively a receiver can tease out an incoming signal.)

802.11n had to go a step further. With 40 MHz channels, the chance of stepping on other networks is higher because of the increased radio space. 802.11n is designed to make devices and base stations listen for signals outside of a core 20 MHz channel, such as networks in adjacent businesses, houses, or apartments. If none are detected, the whole 40 MHz channel is used; otherwise, devices stick with the regular 20 MHz channel. Users don’t notice whether a narrower or wider channel is in use, although they might notice the variation in throughput.

802.11ac starts with a disadvantage for its 80 MHz and 160 MHz channels: there are simply not enough of them available without restrictions in most countries to allow multiple networks in the same general vicinity to rely on channels that aren’t already in use. In the United States, only two 80 MHz channels (four contiguous 20 MHz channels) are available: channels 36 to 48 as one chunk and 149 to 161 as another. Europe has even fewer clear channels, with likely just one consistent 80 MHz chunk possible. (Channel numbers are in increments of 5 MHz, so channel 36 comprises 36–39 as a total of 20 MHz, and so forth. Yes, it is confusing!)

Europe and the United States have additional 5 GHz channels that can be used, but with a strict requirement that base stations on these channels (15 more 20 MHz channels in the United States!) must listen for telltale signs of weather-sensing radar used by governments. If detected, the base station has to tell clients that it is switching to another channel and then immediately move to that channel. Manufacturers have told me for years that the detection mechanism suffers from many false positives, and is thus triggered frequently where no radar installations exist! It’s hard to use these channels reliably and introduces complexity in networking software. As a result, Apple has never included these channels as options in its base stations, and most consumer gear avoids it as well. (Meru Networks, an enterprise wireless network hardware maker, has a chart [2] with more detail.)

For the channels in which checking for radar isn’t required, 802.11ac is much cleverer than 802.11n about backing off when it senses there’s traffic on any of the narrow channels it might be combining into one of its wide ones. So it’s quite efficient about using up to 80 MHz (in 20 MHz to 40 MHz swaths) as available for each chunk of data sent. If there are no other networks nearby, or those that exist are mostly quiet, an 802.11ac network can get great throughput. But in places with active networks, throughput drops significantly.

This problem might not exist forever. Future 802.11ac gear could combine 20 MHz or 40 MHz channels that aren’t directly adjacent from different parts of the 5 GHz band.

With Laser-Like Focus -- Don’t let the speed discussion get you down, because 802.11ac does have three distinct advantages: better coverage, better performance at greater distances, and multiple-device simultaneous transmissions.

Let’s take these one at time:

These three new features require new 802.11ac radios in all the gear you want to use, which will disappoint those who hoped for some backward-compatible improvements. Unlike the shift from 802.11g to 802.11n — where 802.11g devices saw improvements merely by talking to an 802.11n-capable base station — you won’t see these improvements without new adapters. It worked with 802.11n because its benefits came from more-sensitive receivers, multiple antennas, and more-directed and -powerful transmitters, even without beamforming. 802.11ac uses essentially the same basic technology as 802.11n in that regard and thus requires that all devices have new chips and radios to take advantage of the improvements.

(For a vastly more detailed and technical discussion, Cisco has a very nice explanation [4] from last year.)

Eventually, 802.11ac Will Be the New Normal -- Just as 802.11n gradually made its way first into base stations, then into desktop and laptop computers, and finally into tablets and then smartphones, 802.11ac will follow the same progression as chips get smaller, require less energy, and become cheaper.

For most of us, 802.11ac isn’t a must-have feature, especially when homes and offices that really need throughput already have cheap gigabit Ethernet everywhere. However, as 802.11ac becomes commonplace, we’ll slowly see improvements in speed and more significant ones in coverage.

We might finally reach a point where, instead of the three 802.11n base stations I need to cover my modest main floor and basement home and office (riddled as it is with signal-blocking older building materials), a single 802.11ac base station could handle the whole shack.