SuitWatch/Monday, February 11, 2002
Weaving the Wireless Web
Wireless is about the hottest word on the Web. Also one of the biggest, since Google finds 10,300,000 documents that mention it, which about doubles the number that mention "lunch".
It's also one of the oldest. In December we passed the 100th anniversary of Marconi's first wireless transmission across the Atlantic. <http://news.bbc.co.uk/hi/english/sci/tech/newsid_1705000/1705485.stm>
For most of the intervening years, wireless has roughly been the British synonym for radio, a term rarely used west of the pond, where the closest relative was "cordless".
Then Wi-Fi -- 802.11b -- came along. The standards have been around for quite a while <http://www.wirelessethernet.org/>, but 2001 was the year Wi-Fi's adoption curve started looking like a hockey stick. Suddenly Wi-Fi -- a form of wireless Ethernet -- became as infrastructural as Ethernet itself. Now, thanks to DHCP (Dynamic Host Configuration Protocol) <http://www.dhcp.org/>, anybody with a Wi-Fi card in their laptop can hop onto the Net over any local access point (aka "hot spot") configured to allow it.
MobileStar operates almost countless access points in Starbucks cafes. WayPort has them in many airports and hotels. There are noncommercial cooperatives like NoCatNet in Sonoma County, CA, Consume.net in London, SeattleWireless.net in Seattle and SFLan in San Francisco. Not to mention countless unaffiliated open hot spots everywhere somebody has a base station that doesn't ask for a password. And there are plenty of those.
As a result, "wireless" is rapidly becoming synonymous with Wi-Fi. This is both a good and a bad thing. It's a good thing because we need wireless infrastructure, and that has to start somewhere. It's a bad thing because Wi-Fi is profoundly limited as a building block. Also because Wi-Fi commercialization risks limiting its practical scope to Windows users, which violates the nature of the Net as a commons.
Let's look at the building-block issue first. The 802.11 standard limits Wi-Fi to just 14 or fewer channels out of the 22 assigned in a narrow wedge of the 2.4 GHz spectrum. <http://www.btinternet.com/~duncan.jauncey/consume/wavelength.html> By contrast, Bluetooth uses 79 nearby channels in the same band. What's more, Bluetooth hops channels 1600 times per second, while Wi-Fi hops only twice per second. Even ignoring serious concerns about Bluetooth interference <http://www.e-magazineonline.com/FEA/FEA012505.htm>, we're talking about a spectrally limited resource here.
There is also the matter of range. There are Wi-Fi jocks who boast up to several miles of range per base station, using highly directional antennas. Parts of Aspen, Seattle, San Francisco and other towns are thick with free Wi-Fi access provided by rooftop base station antennae. But a Wi-Fi base station in a house generally has less range than a cordless phone. Our 1900-square foot house needs two base stations to give us full coverage for laptops equipped with off-the-shelf Wi-Fi cards.
Short range has certain almost cellular advantages, however, since more stations can fit in a given geographic area. And a number of familiar infrastructure builders are stepping forward to make the most of the access that's out there. Closest to home for the Linux community is Sputnik <http://www.sputnik.com>, a new affiliate organization that allows members to share Wi-Fi bandwidth with other members, no matter where they are. Its software runs on any Intel-based PC hardware. Still in its early stages, it was founded by high-profile founders of a well-known Linux dot-com who don't want to call undue attention to themselves while they work the kinks out of the brand new service (the site only went live a few days ago -- stay tuned).
Less modest is Boingo Wireless <http://www.boingo.com/>, which was launched to great fanfare in December by Earthlink founder Sky Dayton. Much as Earthlink aggregated the services of many widely distributed ISPs, Boingo aims to do the same for the spreading population of Wi-Fi hot spots in the world. The big difference is that Boingo at this point only has public plans for Windows clients. So, if Boingo becomes as ubiquitous it hopes, it will serve to exclude non-Windows clients, which include a small but significant population of Linux and Macintosh laptops.
In a more purely infrastructural direction, there is the Open Spectrum effort <http://www.edventure.com/conversation/article.cfm?Counter=756613> led by Kevin Werbach, an analyst with EDventure Holdings <http://www.edventure.com/conversation/articles.cfm?Page=kevin> who has a significant background on the regulatory side of the wireless tracks.
A Harvard-educated attorney (he was publishing editor the Law Review there), Kevin worked for the FCC (as Counsel for New Technology Policy) before moving on to EDventure a few years ago. He wants Open Spectrum to take advantage of both spectral regulatory relief (more frequencies, unlicensed operation) and spread spectrum technology <http://www.kmj.com/proxim/pxhist.html>. This calls for some resourcefulness -- the kind Linux technologists live for -- in new product designs and deployments. In "A Cure for Bandwidth Blues" <http://zdnet.com.com/2100-1107-531165.html>, he writes,
Using smart "software-defined" radios, nodes in unlicensed wireless networks could cooperate actively. Devices can act as repeaters for traffic between nodes. They could dynamically select power levels and coding schemes based on the behavior of other nodes. And they could cooperatively sample and adapt to background noise. Through these mechanisms, which could be encouraged through equipment certification rules, the spectrum would go from a fixed resource to one that expands with additional computing power and technical innovation. New users could actually increase the bandwidth available by contributing to the cooperative intelligence of the network.
To hasten the regulatory process, Kevin has written Open Letter to the FCC on Spectrum Policy.
The second big issue is range. Let's face it: Wi-Fi is to the Wireless Internet what walkie-talkies are to cell phones. In fact, cell phones give us a very useful model of what we're looking for from the Wireless Internet.
Basically, the extent of the Wired Internet is roughly the same as that of the worldwide telephone system, augmented here and there by cable and satellite. My point: you can expect to connect to the Net everywhere you find hard-wired phone service. If there's an open jack, that's your port. This is extended slightly by Wi-Fi, which essentially provides cordless Ethernet service.
Here's my next point: you should also expect to connect to the Net everywhere there is cell phone service. There is no reason, beyond politics and telco company inertia (both very significant factors) that the Internet shouldn't be available, even at broadband speeds, everywhere you find cell phone service.
That's roughly the point Marty Cooper, who invented the cell phone at Motorola, has been trying to make for years. His company, ArrayComm <http://www.arraycomm.com> (disclaimer: a past consulting client of mine), sells one means by which two or more digital cell site antennas can combine signals to constantly maximize bandwidth at any point in three-dimensional space inside a cell's coverage area. And it doesn't require new equipment outlays (though some would certainly help).
One way or another, that's the next generational stage we should be looking to complete. However we get there, you can expect the most resourceful operators to employ the most useful building components. Including operating systems.
Additional link:
Compaq Networking & Wireless <http://www.compaq.com/showroom/networking.html>