Anyone with a yen for disassembling computers—which turns out to be a disturbingly large number of people—discovered last year that Apple had jumped the gun on wireless standards by including Atheros and Broadcom 802.11n, or “N,” chips into some Intel Core 2 Duo models.
This stole some of Apple’s thunder last week at Macworld Expo when it formally announced its adoption of 802.11n and the wireless networking standard’s 100 Mbps-throughput. But what was more surprising was the company’s willingness to commit to a standard that’s a year from completion.
G, I’m having deja vu
Four years ago, Apple also went with a draft of a wireless standard, in that case 802.11g or “G.” at Macworld Expo 2003, and that didn’t seem so bad—did it?
Both G and N come out of the Institute of Electrical and Electronics Engineers, or IEEE—a standards group that brings groups of engineers together to draft and refine protocols. (A third spec, 802.11a, or “A,” uses a different frequency range than B and G, which will become significant later in this article.)
But the G standard was essentially complete in its IEEE task group when Apple shipped the original AirPort Extreme gear. Several firmware upgrades were required to ensure full compliance with the final standard—approved six months later—and interoperability with other companies’ G hardware.
The N chips that Apple put in last year’s machines were based on a much earlier draft of N. That early version, Draft 1.0, has been substantially overhauled, and Draft 2.0 is slated for approval in March. There’s some concern that chips based on Draft 1.0 won’t achieve the full potential of 802.11n when it’s approved in early 2008.
It’s likely Apple received remarkable assurances about future-proofing from its chip partners, and it’s certain we will see many firmware upgrades over time as N develops. And it’s also possible that a network with N devices that all shipped in mid-2007 will outperform a set of 2006-era N devices.
Let N = faster!
The idea behind N is stated in its charter: Enhancements for Higher Throughput. When 802.11g shipped with its “54 Mbps” rated speed, many were disappointed to find that they were lucky to get 20-25 Mbps of real throughput once networking overhead was removed.
The most basic flavor of N shipped by Apple and others has a raw data rate of roughly 300 Mpbs and net throughput of 100 Mbps. This allows N to slightly exceed 100 Mbps Ethernet, still a standard in many offices. While the ratio of 100:300 seems far worse than 25:54, the number to focus on is the real throughput, not the raw data rate. (Tests of early gear by PC World and other labs reveal lots of incompatibilities among equipment, but have seen 100 Mbps throughput with similar equipment in the best cases.)
To make N work with handsets and gadgets, like the iPhone or a Wi-Fi-equipped camera, the IEEE task group had a grand compromise that allows even faster flavors without breaking compatibility. Faster N devices may hit 600 Mbps in raw speed and perhaps 200 to 300 Mbps in throughput, and will be used in corporations and cost substantially more than consumer equipment. But just like N is backward-compatible with all older 802.11 versions at their fastest speeds, so, too, will variations on N work together at the lowest common denominator.
|Standard||Apple's name||Rated speed||Real throughput||Band||Introduced by Apple|
|802.11b||AirPort||11 Mbps||4-6 Mbps||2.4 GHz||1999|
|802.11a||Mud*||54 Mbps||20-25 Mbps||5 GHz||2007|
|802.11g||AirPort Extreme||54 Mbps||20-25 Mbps||2.4 GHz||2003|
|802.11n||AirPort Extreme||300 Mbps||100 Mbps||2.4/5 GHz||2007|
* Steve Jobs declared 802.11a dead in 2003. It only caught on in companies and for long-range point-to-point connections.
N-thing up my sleeve, and presto
N achieves speeds far above A, B, and G through three techniques: It’s more efficient, it has more radios, and it can use more spectrum.
Efficiency is easy to explain: A, B, and G used more overhead in packaging data to go out over radio waves. Streamlining that added double-digit percentage speed improvements on its own.
As noted in Macworld ’s first look at the new AirPort Extreme Base Station, 802.11n uses MIMO (multiple-in, multiple-out) antenna arrays. The spec requires a minimum of two receiving and two transmitting antennas; it also requires at least two radios. Each radio can send a separate stream of data using the antennas to create and steer a beam.
This allows the same spectrum to be used: Double the radios produces, at most, double the raw bandwidth. The other antenna advantage is that more energy is focused, producing a signal that can be received further away; more sensitivity in receiving signals means a device can “hear” data at greater distances.
N has a final trick up its sleeve, which is using more spectrum than A, B, and G. In most countries in the world, a swath of the 2.4 GHz and 5 GHz spectrum is reserved for unlicensed use—that is, the use of equipment that’s certified by a national regulator, but which works on frequencies that everyone shares, and no one has a unique right to. (Cellular companies have paid hundreds of billions of dollars worldwide for the exclusive rights to their frequencies, by contrast.)
In the 2.4 GHz band, G (and the older B) use 22 MHz wide channels that allow 54 Mbps of raw data to pass; the same is true for A in the 5 GHz band. The N spec will optionally allow 40 MHz wide channels (legal in the U.S. and some other countries), which roughly double that bandwidth.
There’s a fly in the ointment, however. N is designed to avoid interference. In fact, a big delay in finalizing N at the IEEE have been differing proposals and the reconciliation of those ideas for keeping N from stepping on older networks.
So while N is backward-compatible with A, B, and G, it should drop down out of its double-wide channel mode if the base station or N adapter detects other, older networks that it’s stepping on.
This is one reason why 5 GHz suddenly becomes interesting. The 2.4 GHz band has 11 channels available in the United States, but they’re staggered and overlap. Only channels 1, 6, and 11 provide the least overlap and can be used at the same time in the same space.
In 5 GHz, there are as many as 23 channels for use in the U.S., and generally fewer elsewhere in the world. Most of those are restricted to indoor use, and a handful for outdoor. Apple is supporting four indoor and five outdoor channels. (The other 14 channels have additional burdens related to the use of military radar in those bands; one wireless expert suggested Apple or its wireless chip partners hadn’t added the necessary support yet.)
If you’re building an all N network, you might therefore decide that 5 GHz is a better place to build your “home” with fewer users and more channels to choose from in setting up your network. Signals in 5 GHz travel less far than comparable 2.4 GHz signals, which is normally a problem—but good when you’re trying to limit interference from other networks.
If you’re mixing older B and G gear with newer N equipment, or live in an apartment building or an area with city-wide Wi-Fi, you might find your N speeds average more like 50 Mbps than 100 Mbps in the 2.4 GHz band.
The last word
With gear shipping based on a draft of a standard, it might seem premature to purchase equipment. If you’re choosing to go all Apple, there’s no doubt you’ll get the highest possible speeds and compatibility. For mix-and-match networks, wait until second quarter 2007 for the greatest odds of compatibility.
[ Glenn Fleishman is a frequent Macworld contributor and blogs about wireless networking at WiFiNetworking News. ]
Sidebar: What’s with those numbers and letters?
The IEEE never intended the general public to wrestle with how it names things, and 802.11n seems perfectly a perfectly reasonable name to a set of engineers.
The 802.11n standard is part of the 802 committee that focuses on local and metropolitan scale networking standards, including Ethernet (802.3). Inside that committee, the 802.11 Working Group handles wireless LANs. Its first standard, plain old 802.11, defined 1 and 2 Mbps networking speeds.
Various lettered task groups have been formed over the years to handle issues: b for the first robust WLAN standard, e for improved video and voice quality, and i for security. (Capitalization counts: lowercase letters define amendments, in this case to the original 802.11 spec; uppercase, standalone specifications.)
The Wi-Fi Alliance tries to make some sense out of these designations by providing friendly names and consistency. As a trade group, the alliance only allows the name Wi-Fi to be put on devices that meet its lab-based tests, which include requirements for interoperability with a set of standard equipment in the lab.
The Wi-Fi Alliance took the long-delayed 802.11i security spec, for instance, and released an interim certification in 2003 called Wi-Fi Protected Access (WPA); they later updated that to WPA2 when the full 802.11i spec was done.
The group expects to certify devices as compliant to the draft of 802.11n that’s expected to approved in March 2007. They haven’t decided on a name for it yet.