For a while, we had unity.
Since 1999, wireless networking has simply been IEEE 802.11b -- dubbed "Wi-Fi" (for wireless fidelity) last year by an industry certification group, in an attempt to make it more user friendly.
For two years, 802.11b reigned supreme, gathering momentum last year as prices plummeted. Wi-Fi operates at 11 Mbps at distances from 50 to 150 feet indoors to well over 1,000 feet line-of-sight outdoors. It works on Windows, Macintosh, and various Linux/Unix/BSD flavors, sending TCP/IP and other packet data as an extension to plain old Ethernet networking.
But 2001 brings four new flavors to the wireless pot, mixing in disparate elements that might spoil the soup. Some of these standards have been kicking around for a year or more and are about to be commercially deployed; others are new, and may supplant 802.11b entirely. These specs include 802.11a and 802.11g from the IEEE, and the commercially supported HomeRF and Bluetooth protocols.
HomeRF, Bluetooth, and 802.11g all share the 2.4-GHz band with 802.11b. They have to co-exist or they threaten to overwhelm one another. The FCC tests and approves devices intended for use in this unlicensed band to ensure that they conform to tight interference requirements; your use of equipment can't impact other users, which may include wireless networking and a variety of industrial, scientific, and medical (ISM) devices. The FCC doesn't ensure, however, that multiple devices you deploy in your own space are non-interfering; that's up to trade groups.
The "g" flavor is the IEEE wireless LAN working group's attempt to boost speeds in this band from 11 Mbps to 22 Mbps without altering other perfectly workable aspects of 802.11b
The IEEE's 802.11 Task Group G had been considering two competing modulations that would allow the faster speed. All these wireless devices use spread-spectrum broadcast technology, which distributes information across many frequencies in an alternating pattern, so you never continuously occupy any one part of the band. Currently, the FCC (and, by extension, most of the regulatory bodies from other countries) allows only two kinds of spread-spectrum use: frequency hopping (FH), employed by HomeRF and Bluetooth; and direct sequence (DS), used by 802.11b.
Frequency hopping switches between preset ranges of frequencies, from several times per second to several thousand times per second, depending on the standard. Bluetooth, for instance, hops among 75 one-Mhz-wide segments in the 2.4-GHz band.
Direct sequence signals are sent in parallel across an entire range of frequencies, using techniques that attempt to prevent the continuous use of any given segment to prevent interference.
We've been having a great time with 802.11b at the O'Reilly Network, linking up laptops, Visors, and even a long-distance long shot. Tell us about your wireless network experiences.
Two types of direct sequencing are competing to become the 802.11g standard. Texas Instruments has been promoting a technology called packet binary convolutional code (PBCC), which it said offered better backward compatibility with earlier versions, ensuring that 802.11b devices would remain useful as organizations upgraded.
Intersil is pushing a modulation called wide-band orthogonal frequency division multiplexing (OFDM), which is used under the same and different acronyms for xDSL transmission and other tasks where a wide band of frequency can achieve higher efficiency through subdivision. Wi-LAN, which holds a Canadian patent on the technology has information on OFDM. Both of these technologies are actually old modulations reinvigorated by the arrival of faster, smaller, and cheaper chips that make them practical for wireless networking.
Recently, the 802.11 Task Group G eliminated Texas Instruments' contender, PBCC, from consideration for the 802.11g protocol. But TI says it will pursue the technology and sell it as a wireless networking chipset compatible with 802.11b up to 11 Mbps. PBCC was already approved for use by the IEEE with 802.11b at 11 Mbps, although it's not what manufacturers agreed to implement.
OFDM isn't the clear winner yet: At least 75 percent of the working group has to approve it, and if it fails, the group can go back to the drawing board for a new 802.11g technology. It has, however, already been approved for use with 802.11a, the 54-Mbps, 5-GHz standard discussed below.
HomeRF is backed by the HomeRF Working Group, a consortium of companies that includes Proxim, Motorola, and Siemens. Intel jumped ship earlier this year, choosing 802.11b for its consumer wireless products.
An FCC ruling last August let the group finish spec 2.0 which brought its speed from 2 Mbps up to 10 Mbps. But devices using HomeRF have only recently been demonstrated at trade shows, and they're expected to begin to ship this summer.
HomeRF had previously been deprecated because of its lower speed and a paucity of cards and hubs. The spec's key advantage is the integration of voice and data into the baseline for data transmission: Voice packets are given a regular slot to make sure conversations aren't broken up. New HomeRF hubs should allow the use of cordless phone handsets side by side with computers transmitting data.
There is some concern that busy 802.11b networks might cut holes out of the hopping pattern of HomeRF devices, or that HomeRF devices could carpet-bomb Wi-Fi channels -- picture SUVs weaving around a highway careening into tractor-trailers occupying multiple lanes. Under their current designs, HomeRF and 802.11b aren't watching out for each other, and their transmissions could block each other, slowing down each protocol's effective throughput.
But the goal of HomeRF is truly "home," and it's unlikely you'll find both 802.11b and HomeRF networks at anyone's house except the most extreme technophile.
I've spoken to a variety of people inside and outside the HomeRF group, and a wait-and-see attitude prevails on HomeRF's chances once it's included in shipping equipment. The three main companies behind it will send out set-top cable boxes with embedded HomeRF hubs, handsets, consumer standalones like Internet radios, as well as PCI and PC cards. Time will tell.
Bluetooth has received considerably more attention than HomeRF, as the specification has been widely talked about for years. The goal of Bluetooth is low-bandwidth, short-range, low-power synchronization and data transfer, not full-blown networking. PDAs, laptops, cell phones, and other small devices could exchange information without requiring a full-blown TCP/IP stack or much configuration. (Note that I'm saying could; security settings might be Byzantine.)
Because Bluetooth is entering the market late, many companies and standards groups worried that 802.11b might completely drown it out. Or that Bluetooth would be a woodpecker in the 802.11b tree, reducing throughput on both systems.
Fortunately, the IEEE's Personal Area Network (PAN) working group, 802.15, had prioritized work on co-existence between Bluetooth and 802.11b, as well as future 802 wireless specs. The 802.15.2 task group recently approved a proposal that allows adaptive hopping, in which Bluetooth transceivers can avoid heavily trafficked frequencies, benefiting both systems.
Of course, the few Bluetooth devices shipping today and in the next months may not be able to take advantage of this new spec, as aspects of it will certainly require chipset modifications, not just a firmware update, even if firmware updates were available.
Bluetooth, with its lower power signature would have been the loser in those battles, anyway: imagine a relentless bulldozer crushing gnats dashing across its path.
There's one more player in this new world order: IEEE 802.11a, which lives in a different band. 802.11a uses the 5-GHz band called U-NII (Unlicensed National Information Infrastructure) in the United States.
The "a" spec operates at 54 Mbps due to the higher frequency and greater bandwidth allotment. However, because the same power limits apply, "a" access points may only offer access within a few dozen feet. Because of this, a greater density of hubs is required, not just a combined "a," "b," and "g" hub.
In early 2001, many of the manufacturers and standards group members I spoke to thought that 802.11a wouldn't arrive until late 2001 or early 2002. They planned to push hard on 802.11b, and leave "a" as a future upgrade.
But the specification was approved more rapidly than expected (it uses the OFDM modulation) and chipmaker Atheros whipped out a chipset to manufacturers early this year. By summer, at least some 802.11a access points and cards are expected.
This caught folks by surprise. In the middle of a push for an 11-Mbps deployment, suddenly being tempted by 54 Mbps makes an IT purchaser sit up and take notice. The greater density of hubs and the high price tag on early equipment still make "b" - and ostensibly "g" - more affordable choices for the majority of office networking and public space use.
802.11a faces a separate challenge, too: provincialism. The Europeans have been working on a spec called Hiperlan in various forms for nearly a decade without pushing out working equipment. The Hiperlan2 Forum has a press release from early 2001 that suggests equipment will ship in volume by mid-2002.
The Europeans also haven't harmonized with the U.S. on frequency choices, making it even more difficult to meet in the middle. 802.11a may be relegated to the hinterlands -- the entire United States -- rather than make a worldwide splash. This might keep prices high for the long term.
This summer is crunch time -- when you find yourself with an abundance of choices. But the ever-dropping price of 802.11b cards and hubs, along with ever-increasing options for equipping things other than computers (like Palm OS devices and printers) may continue to make 802.11b ideal even when faster options start shipping.
Bluetooth and 802.11b's détente is a great development, as Bluetooth plans to ship regardless of whether anyone is ready to buy. But HomeRF's future is less certain. And 802.11a throws a wrench in the work for enterprise planners trying to figure out how to spend their money on this new, useful technology.
My advice and strategy: buy into 802.11b. It's robust, tested, and cheap, and will be around for the long haul because of the massive current installed base that will grow by millions of users and hundreds of thousands of locations in 2001. Whatever comes after it will be faster, but it has to be compatible with the gold standard of "b" to make inroads.
Glenn Fleishman is a freelance technology journalist contributing regularly to The New York Times, The Seattle Times, Macworld magazine, and InfoWorld. He maintains a wireless weblog at wifinetnews.com.
Return to the Wireless DevCenter.
Copyright © 2009 O'Reilly Media, Inc.