Manhasset, N.Y. -- Vendors targeting the next-generation wireless-LAN standard went on the defensive last week after the first tests of draft IEEE 802.11n clients and routers indicated poorer-than-expected performance, a lack of interoperability and, in one case, inadequate security. The chip and equipment companies questioned the testing methods, saying they suspected preproduction systems and software versions had been used, and quoted higher performance numbers based on their own tests.

The response even included a stated suspicion of collusion between Airgo Networks Inc.--the market leader in multiple-input, multiple-output (MIMO) chip sets, on which .11n systems will be based--and analyst Craig Mathias, principal at Farpoint Group (Ashland, Mass.). Mathias performed the initial tests at EE Times' request on equipment from Netgear Inc. and Buffalo Technology.

But the countermoves lost their punch when subsequent tests by other outlets of a Draft N (prestandard) system from Cisco Systems division Linksys also showed underperforming implementations and exposed interference problems with legacy 802.11g networks.

Farpoint Group's independent tests of off-the-shelf draft-compliant wireless networking equipment were performed in a home environment and analyzed the capabilities of Buffalo's AirStation Nfiniti router and client, which use Broadcom Corp.'s Intensi-fi chip set, and both versions of Netgear's RangeMax Next client and router. One of the Netgear versions is based on Broadcom's Intensi-fi chip set and the other on Marvell's TopDog.

The draft-compliant equipment was then compared with two established products from Linksys: the SRX400, based on Airgo's third-generation proprietary MIMO chip set; and the IEEE 802.11g-compliant Wireless G line, based on Broadcom's 802.11g chips.

The performance comparisons had been much anticipated in the competitive wireless-LAN market. With the introduction of a draft standard in January, the stage was set for the introduction of so-called prestandard equipment. IEEE dislikes the prestandard designation and, apparently to little avail, had asked developers to avoid it. Farpoint's test results suggest IEEE's instincts may have been on target.

"Draft N is misleading," Mathias said last week, noting that during the tests "we couldn't get the [Netgear and Buffalo] equipment to talk to each other. We couldn't even get the two Netgear systems to talk." He called the finding "surprising."

Mathias also noted that the final 802.11n standard will differ considerably from the draft that vendors are using for prestandard gear. "Buyer, beware," he said, warning that the transition may involve more than a simple software or firmware upgrade.

"The companies are looking for a short-term market advantage, but consumers are looking for performance, not a specification," Mathias added. "The more-established products just blew them away. Even the G stuff [Linksys Wireless G] performed better. It's like we're going backward."

Along with interoperability and range, security implementations were inadequate, Mathias found. "We suspect these products were rushed to market to capitalize on 802.11n draft hysteria, and grant that they could [be improved] via firmware and software upgrades over the next few weeks or months," he wrote. But advances in MIMO radio design over the next year will obviate such upgrades, he added.

Point and counterpoint
The disclosure of the test results led to a rapid-fire defense on the part of semiconductor and equipment vendors alike. Bill Bunch, product manager for wireless-LAN products at Broadcom (Irvine, Calif.), said he doubted Mathias had used production-ready equipment with the latest software. Mathias countered that he had used off-the-shelf equipment bought from a store.

Bunch also claimed in-house tests had shown higher performance for Broadcom MIMO chips than for Airgo's, with Broadcom's chips coming in at up to 180 Mbits/second of throughput. Airgo has shown rates of up to 104 Mbits/s. "On speed we beat Airgo hands down, and there's a heck of a lot of performance yet to be squeezed out," he said.

But Airgo CEO Greg Raleigh said tests his company had done on Netgear equipment with Broadcom chips had shown the gear could barely get past 64 Mbits/s of true payload throughput.

As for the tests' dismal interoperability results, Bunch acknowledged that interoperability is a goal still to be achieved. "There's a very clear road map to interoperability, and any technical issues will get ironed out over time," he said.

Bunch also slammed Airgo's chips as nonstandard, calling the technology "a proprietary walled garden."

Raleigh bristled that Broadcom is "claiming compliance" when "there's nothing to be compliant with. The final standard will be a lot different from the draft."

Broadcom, Marvell and others "have been making claims for a long time [for their prestandard chip sets]," Raleigh said. "At some point you have to be held accountable. At what point do you make evaluations? [Do you do them] based on what's coming in the future? They have to realize they can't just wish the performance into being."

Raleigh expects an 802.11 standard by mid- to late summer and said Airgo's next generation of chips would be standards-compliant. He would not say when those chips would emerge.

Netgear marketing vice president Vivek Pathela also questioned the technical accuracy of the tests, claiming performance figures for Netgear's equipment of up to 140 Mbits/s and 400 feet. Both Pathela and Bunch suggested the tests had favored Airgo, with Bunch saying Mathias had gone "to great lengths to compare Airgo to Draft N."

Pathela went so far as to suggest that Mathias and Airgo were in collusion. He claimed that Airgo's Raleigh had sent him similar figures before Mathias released his report, and that Mathias had performed the tests in a house leased by Airgo.

Raleigh said that he had run his own tests and that those were the numbers he had sent to Pathela. And Mathias dismissed Pathela's claims as an attempt to "shoot the messenger," saying that he would be paying rent for the use of the house and that the testing had been "as thorough and impartial as it could be. How many people do you know who use turntables and a spectrum analyzer in tests like this? Farpoint Group's reputation for thoroughness and impartiality speak for themselves. I resent any insinuation that our opinion is for sale. It is not. Ever."

Tests by eWeek (www.eweek.com) and CNET (reviews.cnet.com) bore out the performance findings. The eWeek tests on Linksys equipment also indicated that Draft N systems can cause legacy 802.11g devices operating in the vicinity to drop their connections.

Broadcom's Bunch acknowledged that in the 40-MHz channel-width mode, Draft N devices can overlap with one of the three 20-MHz channels in the 2.45-GHz band. He said Broadcom uses a "good-neighbor mode" that avoids channel collisions.