News

Buffalo exposes "spectrum hogging" and illegality in new WiFi standard

by Staff Writer | posted on 18 April 2006


There's nothing particularly clever about Buffalo's announcement of an 802.11n-compliant range of WiFi gear: it's just taking the standard Broadcom silicon and stamping it with the brand. But it is refreshingly honest to see the company attacking the standard itself.

While most commentators have merely reported Buffalo's announcement, with more or less enthusiasm, the real story was exposed by PCW reporter, Clive Akass, who got Masato Kato, Buffalo engineering manager, to warn about over-hype of network speed claims.

In a market where some people are quite literally claiming that their versions of 802.11n will match 100 megabit wired Ethernet LANs (a very dodgy claim indeed!) Kato told Akass that real world speeds won't look anything like that good.

Worse still, said Kato, the top speeds "will be achievable only by hogging bandwidth in channels already congested with 11b and 11g links." And, he added, channel bonding of that sort is actually illegal in Japan.

Broadcom technical supremo, Henry Samueli, recently told Newswireless that the problem wasn't a serious one. "Yes, there aren't enough channels at 2.4 GHz," he agreed, "but 802.11n will work at 5 GHz too, where there are a lot more channels; and the standard means that the transmitter will look for a free channel."

/Guy/ /Kewney/ /writes/ The real hype comes from people who quote the raw data modulation rates of WiFi as if it were the equivalent of the bit-per-second data rates of wired Ethernet.

In a situation where there is only one WiFi transmitter and only one client, and where they are really very close together, it is possible to see the theoretical maximum data rate. But signalling and other considerations swallow a lot of those bits, distance cuts down the speed, and most WiFi access points are shared.

In a typical WiFi hotspot, transfer speeds from the coffee-drinking laptop user to the Web are going to be far, far slower still, because coffee shops don't invest in megabit-speeds Internet connections. You're likely to see no more than your own share of an ADSL upload link - never faster than 256 kilobits.

And at home, where brick walls cut down the signal and neighbours share channels, most users find that max data transfer rates from their notebook to the desktop machine in the den will be closer to five megabits per second than the hundred megabits trumpeted in the brochures.

A recent "real world" test in an urban office carried out by NewsWireless did a transfer of the equivalent of two DVDs from a standard Centrino laptop to a server. The data can be shifted in less than an hour on Ethernet; but despite having only a two-metre distance between laptop and Access Point, it took over 24 hours over the WLAN.

And that was over a MIMO link, advertised as offering 240 megabits per second.

If the notebook had been equipped with a MIMO pre-N card, instead of standard Centrino, the data would have been shifted in less than 12 hours - perhaps. But that would only be achieved if the wireless link was not interrupted, because standard Windows folder copy operations will time out if the line goes down. And standard Windows copy operations have to restart from the beginning; they don't resume where they left off.

The moral is: make sure backup operations are done via USB or Ethernet cable. Wireless, however useful it is, can't match wired Ethernet for speed or reliability.

 Tags: , ,  and


the 'illegal' standard... - You can discuss this article on our discussion board.