Think about a wireless LAN (WLAN) with enough throughput to match your switched Ethernet infrastructure.
That’s what the Institute of Electrical and Electronics Engineers (IEEE) is thinking about.
Last week, the international standards group launched a working group charged with crafting changes to the 802.11 WLAN standard so that these networks would deliver at least 100M bit/sec. That number is throughput, what users see when they transfer a file, for example, as distinct from the data rate, which is the raw speed before you subtract the overhead associated with the protocol.
In the case of 802.11, the overhead adds up to a whole lot, typically more than half of the data rate. An 802.11b access point, rated at 11M bit/sec, typically gives a throughput of less than 6M bit/sec, often far less. The 802.11a and 802.11g hardware can give users about 18M to 22M bit/sec. The data rate for both is 54M bit/sec.
Silicon makers have been boosting WLAN throughput to around 100M bit/sec for some time. The catch is: You have to have the same chips in both the client and the access point, and high throughput sacrifices conformity to the 802.11 specification. Atheros, the first vendor with a 54M bit/sec 11a chipset, markets complementary metal oxide semiconductor (CMOS) chips that support what it calls “Super G” and “Super A/G” — proprietary boosts of up to 100M bit/sec throughput.
Atheros plans to contribute these and other technologies to the 802.11n task group, as it’s called in IEEE terms. “The greatest challenge will be to deliver higher performance while simultaneously reducing power and cost,” says Craig Barratt, Atheros president and chief executive officer.
Barratt says 802.11n will “promote the idea that wired networks can be replaced with wireless technologies.” One reason to embrace them is that high-throughput WLANs will eliminate cabling costs. But that’s only true of the wires needed to connect clients to the wiring closets. WLAN access points still need to link via Ethernet cable to wiring closet switches.
Network executives already seem to be discounting high-throughput claims that are based on their WLAN experience. “Unless you are sitting right under the access point, you just don’t get the maximum throughput,” says Dewitt Latimer, deputy chief information officer and chief technology officer at University of Notre Dame in South Bend, Indiana.
WLAN throughput falls off more or less rapidly the farther a client device moves from an access point. The drop depends on how much metal, wood, concrete and other construction materials is between the two devices. In addition, in almost every case today, an access point is a shared medium: whatever throughput it can deliver is divvied up among however many users connect to that one access point. “Most practical applications, such as three students sitting under a tree working on a paper (with wireless notebooks), tend to be insensitive to bandwidth. I don’t think high throughput WLANs will be a big driver until we see things like streaming media applications being untethered.”
The 802.11n task group’s first order of business will be to define a group of application scenarios, describing how the high-throughput technology will be used. In turn, these become the basis for evaluating and comparing what’s expected to be several technologies contributed by different vendors, according to Brian Mathews, publicity chair for the IEEE 802.11 Work Group that oversees this standards work.