By Matthew Broersma
Tuesday, October 18 2005 01:43 PM
URL: http://www.zdnetasia.com/insight/communications/0,39044835,39280388,00.htm

New technology promises to increase the speed of wireless networks by a factor of 20, but the emerging standard is being delayed by squabbling vendors.

A major upgrade has arrived for Wi-Fi that massively boosts speed and range while all but doing away with interference problems. The technology is called MIMO, and is already making its way into homes and small businesses.

The technology, which uses multiple antennas to achieve dramatic performance gains, is already available. Chipmaker Airgo, for one, is currently working on its fourth-generation products with Cisco's Linksys subsidiary; Belkin and Buffalo all use Airgo's technology. Samsung is also planning to use Airgo's chips in its laptops.

Many experts believe that MIMO is the future of the wireless LAN. "The first wave, 802.11b, 11a and 11g, improved modulation," explains Leif-Olaf Wallin, an analyst with research firm Gartner. "The next step is to make the antenna smarter."

But while MIMO has the potential to become move Wi-Fi forward it is not yet standardised, with several different, incompatible implementations of it being shipped by rival chipmakers including Intel, which is used to getting its own way. The issues facing MIMO are very similar to the situation with the 802.11g standard in the last few months before the standard was finalised, with vendors shipping gear promised to be compatible with approved kit. But this time around, instead of being a few months off, the standard in question--802.11n--is about two years away, and a draft hasn't even been agreed on yet.

"People have a tendency to buy whatever's the latest thing."

--Ken Dulaney, Gartner analyst

The standards issue surrounding MIMO has led some industry analysts to issue a warning to enterprises and even the general public to steer clear of the technology for the time being. But so far, consumers and small businesses, at least, don't seem to be listening. Blistering performance gains are one reason--Wi-Fi gear with MIMO added on can make networks run at four times the speed of standard 802.11a/g networks, and 20 times that of older 802.11b networks.

But despite the standards problems, a lot of companies are choosing to invest in proprietary MIMO WLANs for the very simple reason that the technology is available now. "People have a tendency to buy whatever's the latest thing," says Gartner analyst Ken Dulaney.

Manufacturers say large enterprises are also getting on board. Airgo is targeting big business and promising to reveal an enterprise-grade licensee this year. The company says larger companies are "very open" to the technology, now that a number of competing chipmakers have jumped aboard the MIMO bandwagon. Trapeze Networks says it will begin selling MIMO Wi-Fi products to enterprises for specialised applications such as wireless videoconferencing.

How it works
MIMO's main benefits are all to do with using its multiple antennas--which can be internal or external--to process signals.

One of these techniques, used by most MIMO chips, is the ability to resolve multiple signal paths, known more technically as multipath signals. These are the echoes and fragments of signals that arrive after the main line-of-sight signal, such as the reflections off of buildings in a built-up environment. Traditional 802.11 gear sees these signals as distortion, but MIMO is able to use them to reinforce the main signal. That means clearer signals, longer-range signals or a bit of both.

A related feature, pioneered by Airgo, is spatial division multiplexing (SDM), which transmits multiple independent data streams within a single channel of bandwidth. This can increase throughput as the number of data streams is increased. Multipath processing can work with a conventional transmitter at the other end, but spatial multiplexing requires an antenna pair at each end of the transmission for each data stream--in other words, it won't give any benefits unless MIMO hardware is in use on both ends of the signal.

MIMO antennas each need dedicated processing hardware, which means manufacturing costs are unavoidably higher than current standard Wi-Fi kit.

Routers from Linksys and Belkin are currently using Airgo's True MIMO technology, with multipath and spatial multiplexing. The RangeMax Wireless Router from Netgear uses BeamFlex from Ruckus Wireless, formerly known as Video54. This MIMO variant has seven antennas and uses a technique called beamforming instead of spatial multiplexing.

Beamforming transmits multiple identical data streams, instead of independent streams, but the results are similar. D-Link's Super-G with MIMO wireless router uses a four-antenna

Atheros Communications chipset, which also uses beamforming instead of spatial multiplexing.

MIMO hardware generally shows only slight performance or range gains if the MIMO router is used with the conventional Wi-Fi hardware found in most laptops. This is one of the points against it with enterprises--to take full advantage of MIMO, companies would have to write off their built-in 802.11a/g client devices and buy MIMO-enabled laptop cards.

In return they should get significantly better range and throughput: the upcoming MIMO-based 802.11n standard is designed for theoretical throughput of over 200Mbps and a real-world throughput of at least 100Mbps. 802.11a/g theoretically runs at 54Mbps with real throughput at 25Mbps; 802.11b only realistically handles 5Mbps.

In September, Airgo began sampling a third-generation MIMO chip it says already meets these speed goals. The True MIMO Gen3 chip, which will be shipping in routers, laptops and other devices early next year, has a theoretical data rate of up to 240Mbps and actual throughput of over 120Mbps, says Airgo--faster than most wired Ethernet networks.

Such speeds might sound excessive, but they're needed for particular applications--particularly in the home, to start off with at least. "This is critical for video and consumer electronics, which need the higher throughput for HDTV or high-definition video," says Philip Solis, analyst with ABI Research. "This can also help link up the whole home, where signals typically have to go through multiple walls."

The equipment will also make it easier for users to run multiple services like Internet telephony and video transfer all at the same time. Industry observers believe this level of wireless connectivity will pave the way for new types of enterprise applications as well.

The lack of a standard doesn't worry manufacturers. "There's nothing to lose by shipping the MIMO products before 802.11n," said Airgo chief execuitve Greg Raleigh in a recent report, adding enterprises are "very open" to MIMO today.

Gateway has begun redesigning its notebooks to accommodate the additional antennas required by MIMO chipsets. Samsung is further along--in June the company said it would use Airgo's chipsets in upcoming versions of its X20 and X25 notebooks, instead of the Centrino chip bundle from Intel. Intel, for its part, recently outlined plans for incorporating MIMO into a future Centrino version.

Analysts say the zeal of hardware vendors is understandable, with such a huge market at stake. "When 802.11g products came along, market share shifted among Wi-Fi semiconductor companies. Those who came in faster with products got an early lead in the market," says Philip Solis, analyst with ABI Research. "802.11n will be almost like a new starting point. New players have everything to gain, and companies with the most market share now have everything to lose."

Political skullduggery
Nevertheless, many in the industry see proprietary MIMO add-ons as little more than a con. "We are seeing vendors try to cheat the public by releasing products early," says Gartner's Dulaney. "The specification needs to be approved, and the products need to go through the testing process before it's delivered to the market. People really want interoperability."

Many people will find themselves paying extra for technology they don't need and in many cases won't work properly, Dulaney says. "Few people even use 802.11b to its full extent. A 10Mbpswireless LAN is fine," he says. "But if there's an 802.11a client around, it will drop back to a, and they won't even know it. People don't have spectrum analysers. They'll have the wool pulled over their eyes."

Something similar happened with 802.11g equipment in the home, Dulaney notes--the equipment could be dragged down to a fraction of its intended speed by a neighbours' 802.11b equipment. "In the long term, sure, we always need more bandwidth, but we're telling people not to worry about it for four or five years," Dulaney says.

Complicating matters further is that while the manufacturers are aggressively pushing their proprietary hardware, the standard is looking further and further away.

The IEEE formed the 802.11n Task Group (TGn) in January of 2004, and initially it didn't look as though finding common ground amongst the different proposals would be difficult, since they were all relatively similar. At that stage, the process was expected to be completed in late 2006.

A deadlock developed between the WWiSE (World-Wide Spectrum Efficiency) group, backed by Broadcom and Airgo, and TGn Sync, backed by Intel and Philips. In July the two groups said they'd work together with the third major group, MITMOT, on a joint proposal. The compromise raised hopes that a joint draft could be presented in November, with a final draft set for January.

The skulduggery hasn't stopped there, however, with a group of the biggest WLAN chipmakers--Intel, Broadcom, Atheros and Marvell--reportedly collaborating outside TGn on key interoperability aspects of the specification, which they plan to present to the IEEE by November. Marvell went on record defending the meetings, which have riled some on the official task force--notably Airgo--and even raised antitrust concerns.

Intel and friends upped the ante against Airgo last week when they formed a group called the Enhanced Wireless Consortium (EWC). Led by Intel, Broadcom, Marvell and Atheros, the group is the latest example of a long tradition in tech: The big guys, one way or another, usually end up calling the shots on standards.

The latest developments may be an attempt by the big, established chipmakers to keep Airgo-style spatial multiplexing from being an important part of the final standard, says analyst Solis. The turmoil may mean that a joint proposal won't arrive until January or later, which could push back the finished standard by months, well into 2007.

"It is a highly political game, it is about the positioning of the various vendors," says Gartner analyst Wollen. "Not all the vendors in the standardisation process have an interest in it going too quickly, since it could eat away at their current market."

If the standard arrives in 2007, big enterprise players such as Cisco won't get on board until around 2008, Wollen estimates: "We don't foresee 11n penetrating into the enterprise before 2010."