Printer Friendly

Time to say "bye-bye" to WiFi? Faster competition comes faster than expected.

In a fractious wireless market, where multiple standards are constantly competing for spectrum space and consumer attention, is the sector now ready for 24Mbps products? Will it be ready for 54Mbps sometime late this year?

Ready or not, super-fast wireless is on the way, and computer professionals should be aware of the complexities and competing standards in this market, where 11Mbps (actually 7Mbps in the real world) is now considered too slow and a number of companies and products are currently vying for dominance in the fast lane.

Proprietary products incorporating parts of 802.11g--the new, faster WLAN specification--are expected early this year. But, confusingly, they have been beaten to the punch by products based on 802.11a (a standard approved with WiFi in 1999) which offers a maximum throughput of 54Mbps operating in the 5GHz range. But before you throw away those WiFi base stations, it's important to understand why both standards are being touted, and the technology observers expect will make them viable.

WiFi Wherefores

Most readers will be familiar with IEEE 802.11b, now commonly referred to as WiFi. In its current incarnation, WiFi operates at about 7Mbps in the 2.4GHz radio spectrum. Recent security concerns notwithstanding, WiFi is finding good acceptance among consumers and businesses alike, even as shorter-range technologies like Bluetooth continue to come to market. But the specification's speed limitations mean that it is unsuitable for bandwidth-hungry applications like high-quality video. Further, 2.4GHz devices found in the home use different protocols and compete for a relatively meager portion of the radio spectrum--about 84MHz--and tend not to cooperate, resulting in interference. Microwave ovens are one common culprit, but cordless phones can also wreak havoc, even to the point of disrupting WiFi LANs and DSL service. More recently, Bluetooth has been shown to interfere with WiFi networks, another issue to consider when making deployment choices.

At the end of last year, the IEEE 802.11 Task Group G approved its first draft of the 802.11g standard, which supports data rates up to 54Mbps in the 2.4GHz band. The group was meeting at press time and general industry speculation was that it would be approved by January. However, CTR obtained an email to Task Group G members from the group's leadership indicating that "the earliest we can expect to see an 'approved' 802.11g standard is late 2002 early 2003." (Requests for comment from Group G officials went unanswered.) The first interoperable products based on the specification will likely operate in the neighborhood of 24Mbps, with faster products in the second generation.

The current 802.11b spec, which was ratified in 1999, will serve as the basic reference design for future 2.4GHz devices that conform to .11g (see Fig). However, sometime later this year 2.4GHz WLAN products may be supplanted by those operating in the 5GHz spectrum, which has fewer technical problems than the lower frequency. Some products based on this technology are already on the market, and several new ones were announced at Comdex.

While 802.11b and .11g are in the same family of specification and both operate in the 2.4GHz band, operationally they are different. WiFi is based on direct-sequence spread-spectrum technology (DSSS), while the newer technology is primarily based on Orthogonal Frequency Division Multiplexing (OFDM). OFDM is a physical layer encoding scheme which divides a high-speed data stream into multiple, slower sub-streams that are transmitted over different frequencies in parallel and then re-assembled (multiplexed) by the receiver. Because of its orthogonal nature, OFDM is considered highly efficient: sub-channels can overlap, which means it uses less of the available radio spectrum (although the FCC has expressed some reservations that it still uses too much).

Intersil versos TI

Predictably, 802.11g incorporates parts of both the existing WiFi standard and its faster .11a cousin. The expected .11g specification will be frequency-compatible with the WiFi standard while allowing speeds to scale to 54Mbps in the 2.4GHz band. (802.11a and .11b use different frequencies and hence can operate side-by-side but currently cannot interoperate.)

802.11g has two "mandatory" modes of operation: OFDM (offering 802.11 a data rates in the 2.4GHz band) and implementation of 802.11b Complementary Code Keying (CCK) for backward compatibility with WiFi. Much of the debate surrounding the new specification was centered oil whether it would incorporate OFDM technology from Intersil Corp. or PBCC (Packet Binary Convolution Coding) from Texas Instruments. TI has spent tens (some say hundreds) of millions of dollars developing PBCC technology or acquiring companies who had it.

While TI's technology will in fact be in the 802.11g specification, it is relegated to an optional mode, of which there are two: PBCC-22 (for 22Mbps operation) and CCK-OFDM, to support rates up to 54Mbps. Because the company's existing ACX 100 wireless products already use TI's proprietary PBCC technology at 22Mbps, the company had hoped that its customers' equipment would be able to interoperate at high data rates with future devices from other manufacturers. This now seems less likely.

For its part, Intersil has indicated that it will not support TI's technology. According to the company, by the second quarter of this year Intersil will develop and market a chip set that meets the proposed 802.11g standard. The new chip set will implement the proposal's mandatory CCK and OFDM modulation schemes, supporting data rates up to 54Mbps. The company will not develop chip sets with the optional CCK-PBCC modulation. "We feel that the mandatory elements of the proposed standard meet all the needs of the market," Intersil president and CEO Gregory Williams has said.

TI had claimed that its specification offered better control of interference and better backward compatibility with existing equipment. There was also some concern that the FCC would not support OFDM because it uses a broader portion of the radio spectrum than DSSS does. However, it appears that OFDM is the technology of choice for wireless networking, at least for the foreseeable future.

Unfortunately, there is already controversy surrounding the new version of WiFi. Some observers warn that 802.11g, while offering speed increases over existing WiFi hardware, does little to address issues of noise and interference, problems the 5GHz technology eliminates. "The 5GHz technology has less interference, the silicon is here already, and the products are coming," says Joel Conover, senior analyst with industry watcher Current Analysis and a wireless market expert. "I really don't think we'll see .11g go anywhere." Silicon and equipment makers will soon be faced with a decision: Spend the money on 802.11g chipsets and 2.4GHz technology, or bet that future WLANs will adopt 5GHz as the spectrum frequency of choice. To know which way the market is heading, it's helpful to understand what 802.11a will bring to the party--and what .11g will not.

How It Works

Under the existing 802.11a standard, the OFDM physical layer splits a 20MHz-wide data stream across 52 300KHz-wide sub-channels to provide data rates of 6, 9, 12, 18, 24, 36, 48, or 54Mbps. All products which conform to the specification must support 6, 12, and 24 Mbps, while the other speeds are optional. Some proprietary multi-channel products have reached throughput speeds as high as 72Mbps, but 54Mbps maximum appears to be a reasonable expectation for new products that can interoperate. As with all models of theoretical throughput, actual performance will vary depending on numerous intangibles. A safe bet is that 802.11a-based networks will max out at 30 or 32Mbps.

OFDM is not new. In fact, it has been in use in the ADSL standard for several years, as well as in Europe's somewhat controversial HiperLAN2 specification, which is in development (more on this below). In the wireless networking space, the market leaders for ODFM-based, high-speed products have been Atheros Communications and Radiata Corp.; Cisco acquired Radiata in late 2000. Accton, Intel, TDK, and Proxim introduced 802.11a-based access products at Comdex in November.

Atheros, who builds CMOS chip set solutions for use in wireless networking devices, was one of the earliest companies to market the faster WLAN technology. It is also behind the push to create a single, very high speed wireless standard that will support computing devices, networking equipment, and consumer gear like phones without the interference problems and limited spectrum availability that continue to plague 2.4GHz-based solutions. The proposed standard, known as 5-UP (5GHz Unified Protocol) has been presented in technical papers by the company but is currently in limbo.

Migration Headaches?

It's important to note that, in addition to the interference problems that plague 2.4GHz implementations, upgrading to 802.11g is not necessarily a simple migration. It may be backward compatible in terms of spectrum frequency, but 802.11 g is not a firmware upgrade: the silicon for it has not been developed yet, says Current Analysis's Conover. "This [802.11b to .11g] is not an in-place upgrade," Conover notes. "The current access points on the market were not engineered for this specification."

In addition, some .11a supporters even take issue with the characterization of .11g as "backward compatible." "The 'g' access points can see 'b' radios, but the 'b' radio cannot talk to the 'g' access point at higher data rates," says Atheros VP of marketing Mark Bercow. "So sure, .11g does provide 'backward compatibility' to .11b, but to what end? There's no speed increase if WiFi devices are on the network."

Bercow agrees that .11g silicon is not available yet, and doubts that 'g' will even reach recommendation status before the second half of the year. "Basically, what happened is that when .11a was approved [back in 1999], IEEE thought it would take a number of years for 5GHz products to be developed," Bercow points out. "Well, it only took two, and now these products are available and there's this interim standard that uses the older frequency."

Atheros's solution to the backward compatibility issue is the use of dual-mode access points (which may be available by the time you read this), which will allow WiFi and 'a' devices to operate side by side, with the newer network operating at a higher data rate and the WiFi network operating at 11Mbps (or thereabouts).

Atheros's technology, and more generally all forthcoming 802.11a solutions in the 5GHz spectrum, are trying to avoid the mistakes of the past that plague 2.4GHz devices. Unlike devices that operate at 2.4GHz, the FCC has mandated that 5GHz devices have power spectral density controls. This means that a device must use a power level that is equivalent to the amount of the spectrum it occupies. This differs from 2.4GHz devices, where high power output causes interference as the "loud" data stream constantly hops frequencies. Because of such controls, 5GHz probably has a rosy future in the United States, where the 5GHz range is generally wide open (unlicensed) and frequency competition is non-existent. But it's a different story in Europe and Japan, where not all of the frequency spectrum is available.

"This is the biggest hurdle to 5GHz WLANs," says Conover. "You can't easily internationalize this technology because you can't take advantage of the full spectrum everywhere. For example, in Japan only the lower 100MHz of the spectrum is available." Conover adds that 802.11a needs 20MHz of spectrum space to operate at high speeds (54Mbps). So, while users in the United States and Europe will have up to 12 channels available, in Japan users will be restricted to five.

Things in Europe are not so simple, either. In the E.U., HiperLAN2 also proposes to use the 5GHz spectrum, although this technology has been in development for more than two years and products are not expected to reach the market until mid-2002. According to the HiperLAN2 Global Forum, an industry trade group, while the two specifications share the 5GHz band, their primary difference lies in their media access control (MAC) layers: 802.11a's is Ethernet-based, while HiperLAN2's is more akin to ATM.

Still, HiperLAN2 has not yet won European regulatory approval (although, admittedly, neither has .11a) and both sides are working on adding physical layer enhancements which will prevent their 5GHz technologies from interfering with satellites and ground radar units which also share this frequency. Notably, there is an IEEE Working Group (Group H) which is tasked with making these upgrades part of the .11a standard, and their work may be completed in the early part of this year.

Which technology is likely to emerge the winner? In fact, both 'g' and 'b' radios will probably find audiences, depending on current configurations and future needs. But the fact remains that, as technologies, WiFi and its 2.4GHz operation spectrum are now almost three years old. They are yesterday's solution, and may not be able to address tomorrow's needs. In 1999, 5GHz was expected to be the networking technology of the future. Supporters contend that the future has arrived.
Figure 1

 802.11a 802.11b

Standard Available September 1999 September 1999

Available Bandwidth 300MHz 83.5MHz

Unlicensed 5.15-5.35GHz,
Frequencies 5.725-5.825GHz 2.4-2.4835GHz
of Operation

Number 4 (Indoor)
of Non-Overlapping 4 (Indoor/Outdoor) 3 (Indoor/Outdoor)
Channels 4 (Indoor/Outdoor)

Data Rate 6, 9,12, 18, 24,36, 48 1, 2, 5.5, 11 Mbps
per Channel 54 Mbps

Modulation Type OFDM DSSS


Standard Available July 1997

Available Bandwidth 83.5MHz

Frequencies 2.4-2.4835GHz
of Operation

of Non-Overlapping 3 (Indoor/Outdoor)

Data Rate 1.2 Mbps
per Channel

Modulation Type FHSS, DSSS

Source: Atheros
COPYRIGHT 2002 West World Productions, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2002, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Author:Piven, Joshua
Publication:Computer Technology Review
Date:Jan 1, 2002
Previous Article:Increasing network performance using molecular sequence reduction technology.
Next Article:NAS looking more like SAN every day: the best of both systems? Could be.

Related Articles
Big boss bonus babies.
Bye lets UO get jump on Arizona.
Beavers grudgingly get second bye week.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |