Technology designed by chip provider Atheros Communications Inc. has come under fire recently by rival company Broadcom Corp. in a game of what’s better: proprietary or standards-based solutions.
The technology in question is Atheros’ proprietary Super G four-piece technology, which it introduced to try to increase efficiency and throughput on a wireless network. Only one of these pieces is under fire by Broadcom, which said that equipment made by Atheros causes extreme degradation in the speed of nearby 802.11b and 802.11g networks.
The first element in the company’s Super G technology is what Atheros calls “bursting,” according to Dave Borison, the company’s product line manager based in Sunnyvale, Calif. Basically, bursting is used to increase throughput and minimize overhead.
“What I mean by overhead….You know how people talk about a 54 megabit link, when in reality you are only getting about 22Mbps or 23Mbps of throughput? That delta between 54 and 23 is all the overhead,” Borison explained.
The second piece of the technology is called fast frames. Borison said that although the IEEE 802.11 standard allows for up to 4,000 bytes of data per frame to be sent across the wireless network, most products on the market only allow for 1,500 bytes.
“So we do something called fast frames where we send 3,000-byte frames, so we basically double the number of data per frame,” Borison added. “Again sort of minimizing that overhead.”
For the third element — compression — Borison said Atheros used the same compression engine used in Microsoft Corp.’s WinZip application, which allows the company to “cram” much more data per frame maximizing the efficiency of the network and increasing throughput.
Dynamic Turbo is the last element and the one that is receiving all the attention from Broadcom. This is done through channel bonding — bonding multiple channels together — a practice that Borison said is “very common in the industry.”
Borison said that although Atheros agrees with Broadcom’s claim that its equipment does cause degradation in the speed of nearby 802.11b and 802.11g networks, this would happen with all 11g networks working in close proximity to each other. He added that Broadcom’s stunt at the Comdex trade show in Las Vegas in November was nothing more than a marketing gimmick. The “stunt”: taking an 11g network, running a video, and turning on a Super G network next to it to demonstrate that the Super G network would impact the performance of that video.
“And the truth is, we agree with that, that makes perfect sense. That’s the laws of physics,” Borison explained. “That’s the way radio networks work. If you operate any network in close proximity that’s what will happen.”
He added that Atheros’ key message to Broadcom is that its exhibit at Comdex does not demonstrate real-world usage because, for the most part, two networks would not be set up within 10 or 20 feet of each other.
In an interview with ITworldcanada.com, David Cohen, senior product marketing manager at Broadcom based in Sunnyvale, Calif., stated that Broadcom is not “anti-Atheros,” calling it an “excellent technology company.” However, he noted that Broadcom does have significant industry-related concerns with the company’s Super G technology.
Cohen said Atheros’ technology is undermining what the IEEE’s 802.11 standards were put in place to accomplish by the Wi-Fi Alliance by breaking interoperability and compatibility. When you have a technology like Super G that is just not going to be compatible with other technologies, it all just goes out the window, Cohen noted.
“This can be pretty significant because you are making an assumption that you are only going to have Super G devices in your home or your office, and that is not a very good assumption given that you may want to buy products from different vendors,” Cohen explained.
On top of the compatibility issues, Broadcom is also very concerned about the “harmful interference” that it said Atheros’ Super G technology causes. Although Atheros said that Broadcom didn’t demonstrate real-world situations at the COMDEX show, Broadcom stated that in tests done at the company, the Super G network caused interference in distances of up to 15 to 30 feet of a nearby network, not just up to the 20 feet that Atheros mentioned.
Cohen said the instigator behind Super G’s problems is the fact that Atheros is using its channel bonding technique to take up the width of two channels.
“You can have one network on channel one, a nearby network on channel six and another network on channel 11 and everything is hunky-dory, they are literally using different parts of the spectrum,” Cohen said. “And along comes Super G…instead of using up one channel like you are supposed to for the standard you are using up the equivalent width of two channels, and so if anybody was in that other channel, you could be causing interference.”
Cohen said that it doesn’t stop here and it “actually gets worse.” Instead of using channels one, six or 11, Atheros has created a signal that is centered on channel six but is doublewide. This means that Super G takes up all of channel six, but also bleeds significantly into channels one and 11. So, in effect, Super G actually causes interference no matter what channel any other technology is using, according to Cohen.
“If you had one of these networks nearby, you would find that you would have interference caused by that Super G network if you are on channel six by bad luck, you would certainly have interference caused, but you would also have nowhere to escape to if you were a neighbor to someone using this Super G technology,” Cohen said. “You would try to switch your network to channel 11 — still no good, you get interference. You try to switch to channel one — still no good.”
Jeff Duntemann, a Colorado Springs, Colo.-based industry analyst and the author of “Jeff Duntemann’s Wi-Fi Guide” said that what is happening here isn’t a conspiracy against Atheros, but is rather “a case of all of us against the laws of physics.”
Duntemann said that the real problem stems from Atheros trying to increase the bandwidth of its wireless networking technology and how many bits it can put through this wireless pipe per second.
“The problem with that is the more bits you put through any kind of connection per second the more bandwidth in the ultra magnetic spectrum you need. What [Atheros] has done to blow up their bandwidth to this very high value is…they are basically hogging the entire band. You can’t put that much throughput on one Wi-Fi channel, it’s physically impossible,” Duntemann explained.
He added that Atheros has used its channel bonding technique with the attitude that because one channel isn’t enough space for its product, “we are just going to have to use two.”
Most Wi-Fi gear only uses one channel at a time, but the laws of physics say you can’t push that many bits through one channel that quickly, Duntemann noted. So, they have had had to use two channels and the real problem with that is now Wi-Fi channels are overlapping.
Duntemann said one of the key problems is that the regulatory people and engineers that designed the system probably never imagined that wireless networks would extend past the corporate walls into such a large number of residential users all using the same bandwidth.
“So, the channels overlap and there are really only three channels on the whole Wi-Fi band…only channels one, six and 11 are completely non-overlapping and if you go on onto five and six together and if you bleed a little on each side, and it’s looking like that’s what [Atheros] does, you basically have taken out the whole middle of the band and anyone else who wants to use the band, if they are lucky, has to use channel 11,” Duntemann explained.
So, realistically speaking, Duntemann said Atheros is moving its technology beyond what Wi-Fi bands can hold.
Duntemann said that the only possible answer to this situation would be for the various regulatory commissions and the various governing nations to put more bandwidth at the disposal of the wireless networking industry. Although this solution would remedy the problem, Duntemann isn’t holding his breath.
“There has been some talk about that but the government moves with this glacial deliberation and it could be a couple of years before we get there. They are going to have to do something soon, [however] because there is a standard on the horizon for 100Mbps throughput and up that the IEEE is working on and it’s going to be a couple of years before it is settled but, as usual, the technology leaders in the industry are pushing the envelope ahead of the standardization process,” Duntemann said.
He added that it is probably a key motivation for Atheros to pioneer a technology that the IEEE will then adopt for this higher speed standard, but Duntemann warns that this process does sometimes backfire.
When asked why Atheros hasn’t defended itself against Broadcom’s allegations more vocally, Borison said that fear in the marketplace is the main reason the company is staying low key.
“We are concerned if we take that message to the market, because most people aren’t experts, we are concerned about creating fear in the market that Wi-Fi isn’t working…and this is not the case,” Borison said. “This is the way radio networks work and we didn’t want to stymie the market and impede its growth.”
Borison went on to say that Broadcom is creating fear, uncertainty and doubt in the market, but “the truth is, it is a great marketing ploy.”
“If you are losing market share, a best defence is a good offense and that’s what they did…but I think in the long run it will fail.”
Broadcom reaffirmed that its only agenda is what it calls “a major industry problem” with the Super G technology and added that Atheros’ motivation for keeping the technology around is very clear.
“You can put a big number like 108 on the box and try to sell products at retail to consumers that may not know better even business folks that may not know better, but again you are going to have a lack of compatibility and you are going to be causing very harmful interference to your neighbors and you may not even be aware of that,” Cohen added.
