Remember a while back when chipmakers Broadcom and Atheros
had a little tiff about some interference issues, allegedly caused by proprietary speed boost options that aren’t part of the 802.11 specifications?
Well the Wi-Fi Alliance apparently remembers. Back in February, it said that it would investigate the claims, and today it had an answer for the problem. In the future, if it sees any such problems during certification testing for Wi-Fi interoperability, it may withhold or even revoke Wi-Fi Certification stickers on certain products. The policy goes into effect today.
Broadcom said in November 2003 that Atheros’ channel-bonding feature, part of the proprietary Super G speed boost, was causing an “enormous degradation” of nearby 802.11b and 11g connections. This was confirmed in testing by third parties like Tom’s Networking, though the effect seemed limited to Broadcom-based networks, and only those that were very close in proximity. In March of this year, Atheros said it had changed channel bonding to be dynamic, a move which would eliminate the issue. However, the company also said at the time that the change was not to fix anything, since no end user customers had yet complained about Super G interference, and none of their testing showed the alleged problem.
“We fully support the Alliance,” says Dave Borison, product line manager at Atheros. “We think this makes a lot of sense… there may be third party extensions that cause performance degradation.”
Borison says today’s announcement by the Wi-Fi Alliance validates that Atheros Super G isn’t interfering with third parties, since its chips are found in the Alliance product test bed. Those products are usually only run in standard mode, however, not with any proprietary extensions turned on.
Bob Wheeler, senior analyst following the WLAN chip vendors at the Linley Group, says that while Super G’s potential interference might already be addressed, “we’ll continue to see turbo modes introduced, and it makes sense for the Wi-Fi Alliance to make sure there aren’t products [that interfere] on the market.”
He cites the struggle taking shape in the IEEE 802.11n Task Group, which will be defining new throughput speeds of over 100Mbps for future WLANs, as what the Alliance is likely preparing for. Since everyone in the market was blindsided by early 802.11g products from Broadcom, the expectation is that many vendors will try selling early 802.11n products to get a drop on the customer base.
“Vendors will sell these pre-11n products, and that will cause a lot of confusion in the market,” says Wheeler.
Atheros couldn’t yet comment on 11n plans, but Borison said “it’s something we’re looking into.”
Phil Solis, analyst at ABI Research, says this move by the Alliance is intended to “shore up potential problems with Wi-Fi branding. They [the Alliance] want to assure companies that there won’t be a problem going forward.”
This raises the issue of whether the Alliance branding is even necessary for products.
“I’m not sure to what extent people are looking for the ‘Wi-Fi Certified’ label,” says Julie Ask, senior analyst at JupiterResearch. “There are a lot of brands/branding of both hotspots (Wi-Fi ZONE, T-Mobile, etc.) and APs/NICs (Centrino, 54G, etc.) that are designed to help the consumer with interoperability — I’d be surprised, with all of the attempts by companies to put their own brand above that of the generic or standard service, that it doesn’t create confusion rather than helping.”
The Alliance, which wasn’t available for comment, did not specify how it would deal with testing for proprietary extensions (AKA vendor-specific high-speed options). The Alliance has a pre-set product test bed which it might have to change based on specific complaints (such as Broadcom vs. Atheros). It’s unlikely that it would be testing all such speed-boosts as a matter of course. The latter option would be too time consuming, but then again, maybe not: the Alliance also announced today that it’s opening up additional certification testing labs in Germany, Spain, and Korea, bringing the total number of labs up to 14.
