By Eric Griffith
March 25, 2004
UPDATE: A reminder to all: when a company says it delivers wireless performance at a certain amount of throughput, keep in mind that the number you see is only theoretical.
The last couple of weeks have seen another avalanche of announcements about higher-speed Wi-Fi.
Chipmakers are usually the impetus for making such claims. Agere said last week that it could achieve speeds of 150 megabits per second (Mbps) with a new chipset. Today, Broadcom touted its Afterburner technology, saying it performs with 40 percent greater throughput than typical 802.11g. All the major Wi-Fi silicon providers (perhaps with the exclusion of Intel), including Atheros , Conexant and Texas Instruments , offer technology to boost the speed of 802.11g products.
Yet almost all of the numbers vendors use — 90, 108, 125, 150 Mbps — are just theoretical bandwidth maximums. They’re what the technology could deliver if the radio frequency environment is clear of any interference, if all the products involved use the same chips and settings, and if there were no overhead at all on the network (which there’s always plenty of, even with slower 802.11b).
Some companies say they’re getting away from using the theoretical maximum numbers; Atheros marketing director Colin Macnab told Wi-Fi Planet recently that he’s dead set against them. But the numbers look great on a package in Best Buy, and sometimes that’s a competitive advantage.
The marketing allure of the numbers is made worse for the companies when they come out so high in tests. In the case of Agere and Atheros, both include compression algorithms in their bag of speed-boosting tricks, which can lead to some heavily inflated throughput scores.
Using compression with the right file can provide amazing-looking results. US Robotics’ senior product manager Jim Thomsen says a highly compressible file with no real data in it — all binary 0s, for example — can look great on a chart, but no one will ever transfer such a data file in the real world. USR recently offered upgrades to existing products, saying it can now deliver 802.11g at 125 Mbps.
USR is using TI’s chips in its products, which don’t use compression. However USR still calls its new downloadable firmware upgrade a 125Mbps speed boost. It uses only frame bursting (adapted from the future 802.11e specification) and a feature available out of the box called packet aggregation, which can push through larger network packets in the same amount of time as a standard 802.11g product. Thomsen admits that 125Mbps is a theoretical max; in the real world they expect around 35Mbps.
- TI Prepares for Wi-Fi Phones
- Linksys Gives 802.11g a Boost
- Channel Bonding Gets Automatic
- Agere Claims Boost to 150Mbps
- Intel Shows Off Converged PDA
- PoE Powering WLAN Growth
Atheros, with its Super G technology, has customers like D-link and Netgear marketing products at 108Mbps buts claims it can deliver a real-world 60Mbps throughput. Again, that is using frame bursting, compression, and the controversial feature called channel bonding, which takes two of the three usable channels in 2.4GHz 802.11b/g and uses them to double the speed. Atheros recently said it has fixed channel bonding by making the process dynamic. Products won’t use adjacent channels that have traffic on them.
Broadcom’s 54g Afterburner is now available in products from Linksys, which calls the line SpeedBooster, but Linksys admits real-world throughput is 30 to 35Mbps. Linksys pokesperson Karen Sohl says the company goes out of its way to not use the term 125Mbps in its marketing — they only claim a performance boost of 35%.
Broadcom’s Afterburner will also be in future products from Buffalo Technology, Belkin and others — most will likely be marketed as “125 High Speed Mode” though Broadcom itself seems to eschew the direct use of “125” as well. As they were the company that first brought channel bonding’s potential issues to light, Broadcom is labeling Afterburner as a “good neighbor” enhancement; it says channel bonding creates severe interference to neighboring Wi-Fi devices.
That’s an accusation to which Macnab at Atheros has previously responded to by saying that to his knowledge no end user has ever returned a product based on Super G due to channel bonding.
That brings up two questions: are people even utilizing the speed boosts? If so, are they savvy enough to notice such issues?
A major downside with all of this throughput number jockeying is that interoperability at the high speeds is out the door. Even though almost all of the above use some kind of frame bursting from 802.11e to improve speed, you’re unlikely to get a Broadcom Wi-Fi card to talk to an Atheros-based access point at any speed higher than the normal 802.11g. Luckily all these products can usually “dummy back” to supporting plain vanilla 802.11g, which is still pretty impressive at real-world speeds of 22 to 25 Mbps.
Home users, it’s argued, tend to buy all their products from the same vendors at the same time, so maybe that’s why speed boosts in Wi-Fi products tend to be targeted toward the consumer and small business end of the market.
Enterprises buy equipment in bulk for a major WLAN deployment, and do so over the course of months or years, so they’re not so lucky. Philip Solis, analyst at Allied Business Intelligence (ABI) says most enterprises wouldn’t use them anyway: “[Enterprises] want standard equipment.”
Until a major enterprise vendor like Cisco or Proxim starts making products supporting the Afterburners and Super Gs, that’s unlikely to change.