September 5, 2000
Work has been underway in
the U.S. toward development of a new, higher-speed wireless networking standard.
Early this year, that work bore fruit in the IEEE's 802.11a draft standard. The
European Telecommunication Standards Institute's (ETSI) Broadband Radio Access
Network (BRAN) working group is putting the finishing touches on an equivalent
standard for Europe, known as the High-performance local area network
(HiperLAN), type 2. The two standards are not identical, but they do have
something in common. They both operate in the 5 GHz frequency band -- new
territory for wireless LANs.
Why 5 GHz?
Most wireless networking
products in the market operate in the 2.4 GHz band. This band has served
wireless networking well, but there isn't a generous amount of spectrum
available there (the FCC has allocated 83 MHz), which put limits on the data
rates that can be achieved. At 5 GHz, much more spectrum is available -- which
explains why wireless developers see big possibilities for this band. Moreover,
promoters of 5-GHz solutions say that the popularity of the 2.4 GHz band will
soon result in intolerable levels of interference for users, dramatically
reducing performance.
This progression to 5 GHz
is really part of a broader trend in wireless networking. The earliest wireless
networking products, which came to market about a decade ago, operated in the
900 MHz band. Because these were proprietary designs, an effort soon ensued to
pursue a vendor-independent standard, to promote interoperability. This resulted
in the formation of the IEEE802.11 committee, which quickly began to focus on
the 2.4 GHz band. All radio-frequency (RF)-based products built to the
IEEE802.11 standard (and the later, higher-speed 802.11b standard) operate in
the 2.4 Ghz band. (I say "RF-based products" because the 802.11
standard also covers infrared-based products, but since [to my knowledge] none
of these have ever made it to market, we can safely ignore them for the moment.)
For many of us who monitor
developments in wireless communications, it's been assumed that the migration to
5 GHz was more or less inevitable, and that it would happen someday -- with
"someday" being quite a distance over the horizon. In fact, from one
perspective, we've only begun to utilize the 2.4 GHz band. Products based on the
original IEEE802.11 specification have been shipping for about three years, but
that specification offered data rates that topped out at 2 Mbps, and market
acceptance was lackluster. Extremely high prices for those products (compared to
comparable equipment for wired networks) didn't help.
It wasn't until the
802.11b specification was finished last fall, that high data rates (11 Mbps)
could be achieved. The combination of higher speeds and much more aggressive
pricing has finally turned the corner for wireless networking in the
marketplace, and vendors have seen much stronger demand for the 802.11b-based
product. But for the most part, these products didn't begin reaching customers
in quantity until the first of this year. Hence, most of the success we've seen
for wireless products in the 2.4 GHz band is not yet even a year-old phenomenon.
Be that as it may, the
completion of the IEEE's 802.11a standard has cleared one hurdle for vendors who
have their product plans zeroed-in on 5 GHz. Three vendors have already
announced products (but not shipment), serving notice that "someday"
may not be so far away. Atheros Communications, Radiata and Systemonic all have
made announcements in the last few months. But more about the specific
announcements in a moment. Let's take a look at the standards and technology
behind them.
A Closer Look at the
Technology
The IEEE802.11a standard
is based on the use of orthogonal frequency division multiplexing (OFDM), which
offers high performance (in terms of how many bits per second can be squeezed
through a given RF bandwidth), and reportedly has high immunity to multi-path
interference.
IEEE802.11a breaks the
frequency band of 5150 MHz (5.15 GHz) to 5350 MHz (5.35 GHz) into eight 20-MHz
channels. Each of these 20-MHz channels is composed of 52 narrow-band carriers
of 300-kHz bandwidth. OFDM sends data in parallel across all of these carriers
and aggregates the throughput. Each carrier can transport 125 kbps of data. Of
the 52 carriers, four are designated as "pilot" (synchronization and
control) carriers, and 48 are used for data transport. Thus, each 20-MHz channel
yields an aggregate data rate of 6 Mbps.
Like other members of the
IEEE802.11 family of standards, IEEE802.11a uses the Ethernet-like
multiple-access with collision-avoidance (CSMA/CA) method. (Ethernet proper is
based on CSMA/CD, where CD stands for "collision-detection." Wireless
LANs can't always detect a collision between two transmitting devices, so
instead, it tries for collision-avoidance [CA].)
The European HiperLAN 2
standard also uses OFDM, and it, too, uses 20-MHz channels, with each channel
being composed of 52 carriers, of which, 48 are "transport" carriers.
The nominal overall data rate is the same- 54 Mbps.
The two standards diverge
in access method used. As mentioned, IEEE802.11a uses the contention-based
CSMA/CA method. HiperLAN 2 uses time-division multiple-access (TDMA). TDMA
ensures that each transmitting node gets its fair share of access to the
network, offering quality of service (QoS) that's critical for voice and video.
Chips on the Way
The product announcements
made so far are for chipsets, and these will pave the way for complete
"end-customer" products. In July, Dresden-based Systemonic announced
development of its HiperSonic chip, a DSP-based product that can handle the
physical layer (PHY) for either HiperLAN 2 or IEEE802.11a. Volume shipments are
scheduled for 1Q01.
Two other developers
announced CMOS-based chipsets claiming very low power consumption for 5-GHz
devices. The Australian firm Radiata (with US headquarters in San Jose, CA)
announced a two-chip set that the firm claims operates at power levels
comparable to today's 11-Mbps chipsets. Pricing to OEMs in 100,000-piece
quantities is set at $35, with volume deliveries scheduled to begin in January.
The company says engineering samples will be available in October.
Sunnyvale-based Atheros
Communications last week announced its AR5000 two-chip set that the company says
will have power consumption comparable to today's 11-Mbps chipsets. It should
sell at under $35 in 100,000-piece quantities. (Atheros says this is comparable
to what 11-Mbps 2.4-GHz chipsets are selling for today.) The company plans
volume production of the chipset to begin in 2Q01.
Both the Radiata and
Atheros chipsets target the IEEE802.11a and HiperLAN 2 standards, and both claim
standards-compliance at data rates up to 54 Mbps. In addition, the Atheros
chipset has proprietary enhanced operation modes that work at extended ranges
and at data rates up to 72 Mbps.
Both of these vendors put
a heavy emphasis on their use of standard CMOS process technology. It's
difficult to achieve RF designs that can operate at 5 GHz without resorting to
such sophisticated and expensive process technologies as gallium arsenide or
silicon germanium. In addition to the cost issue, CMOS really shines at
low-power operation.
Atheros' 5-UP Proposal
With the ink only
beginning to dry on the draft standards for IEEE802.11a and HiperLAN 2, Atheros
Communications is already floating a proposal for enhancements. This Spring,
Atheros announced what it's calling a "unified protocol" for 5 GHz
(5-UP, for short).
With 802.11a and HiperLAN,
the 20 MHz channel is, at any moment in time, controlled by one node (the active
node). 5-UP would permit much finer "granularity," by allowing the 52
carriers that make up a 20-MHz channel to be dynamically and individually
assigned. A cordless phone, for example, only needs to use one carrier; that's
more than enough throughput for voice. Using just one carrier also greatly
reduces the power requirements for the cordless phone.
5-UP would also allow the
carriers to be reserved guaranteeing bandwidth for multimedia. If you're sending
CD-quality audio (which requires 1.5 Mbps) in streaming mode, a sufficient
number of carriers could be reserved to handle the required throughput.
5-UP would allow for
40-MHz channels in situations where eight separate channels aren't needed. In a
home environment, there isn't a need for multiple overlapping cells. A user
could opt for the 40-MHz channels, and get 108 Mbps throughput, as opposed to
the 54 Mbps that can be obtained with 20-MHz channels.
802.11a scales throughput
in 6-Mbps increments, from 6 Mbps to 54 Mbps. 5-UP could scale from 125 kbps to
108 Mbps, in fine increments.
Atheros maintains that
reserving carriers for certain nodes and applications means there will be far
fewer collisions, resulting in more efficient usage of the available throughput.
5-UP is at this point only
a proposal. Atheros president and CEO Rich Redelfs says that the company wanted
to get this proposal out early, before there's a significant installed base of
5-GHz products.
The proposal does hold
great promise as a potentially universal wireless solution. Right now, the
wireless space is fragmented. IEEE802.11b is garnering the market share; but
it's a data-oriented solution that's trying hard to transform itself into a
voice-and-data solution. HomeRF is a voice-and-data solution, but it's
throughput-challenged and market-share-challenged. Bluetooth is a very
short-range (10-meter) RF link that isn't here yet (but will be very soon). All
of these technologies have the potential to interfere with each other, so you
can't necessarily overlay several of them, to get the best of all possible
worlds.
Against this mélange of
wireless offerings, 5-UP offers the high top-end data rates of the 5-GHz band,
combined with the QoS features needed for great support of voice and video
alongside data. The ability to assign single carriers to devices like handheld
cordless phones gives it the potential for low-power operation that only
Bluetooth can match.
The real question hanging
over 5-UP is the political one. Can Atheros rally enough support for this
proposal to turn it into a standards initiative? And could such an effort make
its way expeditiously through the standards process, or will it bog down with
in-fighting?
Which Way Wireless?
The completion of the
draft standards for IEEE802.11a and HiperLAN 2, and the announcements of the
imminent availability of chipsets mean that development of the 5-GHz band is
about to begin. But how successful will it be? Won't the popularity of the
2.4-GHz band slow development at 5-GHz?
There's already a proposal
for new technology (championed by Santa Rosa, CA-based Alantro Communications)
that could double the data rate for the IEEE802.11b, to 22 Mbps. But the
proponents of 5-GHz solutions are asking "why double the data rate, when
you can multiply it by 5x, or more?" The availability of a significantly
broader radio spectrum at 5-GHz means that 5-GHz product will always beat their
2.4-GHz brethren.
Some argue that the
2.4-GHz band is falling prey to its own popularity. It's getting crowded, the
argument goes, and the potential for interference between the myriad devices
operating in this band is high. It remains to be seen whether interference will
be a significant problem, but devices using Bluetooth are only now getting ready
to enter the market. There's huge vendor support and interest in Bluetooth, and
if the technology is only half as successful in the marketplace as its promising
to be, there's at least the potential for interference between Bluetooth-enabled
devices and IEEE802.11 wireless LANs.
If IEEE802.11a and
HiperLAN 2-based products proved to be significantly more expensive than
IEEE802.11b solutions, and consumed substantially more power, there'd be a
strong argument for extending the data rates for IEEE802.11b. In that scenario,
the 5-GHz solutions might be reserved for the most demanding applications with
2.4 GHz remaining the band of choice for most uses. But with at least two
vendors quoting pricing and power consumption levels comparable to 2.4-GHz
solutions, I'm skeptical of such a scenario. But don't plan on consigning your
802.11b LANs to eBay just yet. Although "someday" might be arriving
sooner than we thought, it's not here yet.
For More Information
More information on the
family of IEEE802.11 standards can be found at
http://grouper.ieee.org/groups/802/11/main.html. More on HiperLAN 2 is available
at www.etsi.org. The HiperLAN 2 Global Forum (www.hiperlan2.com) works to
promote HiperLAN 2 world-wide.
For more information on
the chipset announcements, see the respective vendors' web sites:
www.atheros.com, www.radiata.com and www.systemonic.com.
