News

This 3G data card is running at three megabits, in the Netherlands ...

If this report reaches you, you have proof that the Flarion alternative to WCDMA, for 3G wireless data, works. I'm writing this (at some risk to my queasiness) in a car, being driven around The Hague where T-Mobile is just finishing its testing of Flash-OFDM as an alternative for wireless data, comparing it to W-CDMA. And it's fast.

The small administrative capital city of the Netherlands is live to wireless broadband. I was streaming video over wireless as we tested this - not just out of doors miles from a hotspot, but mobile.

I counted maybe half a dozen wireless masts around the city, utterly indistinguishable from ordinary 3G W-CDMA antennae, and I'd guess there are twice as many in the test zone, though it's impossible to tell. We drove around the network of back streets and freeways at the legal speed limit (max 120 km/hr, or 72 mph) and ran into only one dead spot. That, pretty much, is what you'd get from an ordinary 2.5G data feed - but you wouldn't be getting three megabits per second.

Here in my hotel room, I'm quite close to one of the feeds, which is less than a kilometer away. Download speed from a video server on the same subnet is currently averaging a slow 500-700 kilobits per second, peaking at a megabit ... but the mast is the other side of the hotel. It's easily superior to any hotel wireless broadband I've ever used.

If I were on a GPRS link, I'd be getting around 36 kilobits - not enough for audio streaming - and it would be intermittent. In the car, however, I was simultaneously picking up iTunes Internet radio at 128 kilobits, and the video stream, peaking around 2.4 megabits. Handover from cell to cell was as near seamless as it gets.

If I were on 3G W-CDMA (or any other form of 3G) I'd be getting a tenth of that. It would be over 100 kilobits, occasionally over 200 kilobits - but the kicker is the latency. I'd be getting a second or more delay around trip. Here, the maximum latency is around 53 milliseconds.

And here's the bit that will amaze wireless 3G data users: an extract from the upload log of my FTP program, sending a large file to a remote server:

PASV
227 Entering Passive Mode (172,16,1,16,157,246)
connecting data channel to 172.16.1.16:40438
data channel connected to 172.16.1.16:40438
STOR SQLEXPR.EXE
150 Opening BINARY mode data connection for 'SQLEXPR.EXE'.
transferred 36665064 bytes in 899.323 seconds, 318.513 Kbps ( 39.814 KBps).
226 Transfer complete.
MDTM 20040823145043 /guykewney/SQLEXPR.EXE

The illustration shows this graphically. It's the download-upload meter, showing how fast data is flowing through the network interface. Red bars show download speeds, green bars show upload; and the yellow is where they overlap. As you can see, in my hotel room I didn't get faster than 1.4 megabits per second, while the snap shot shows downloads running at 640 kilobits, and uploads at 494 kilobits. Averages were rather better for downloads, rather worse for uploads.

The bottom line is: I transferred a 36 megabyte binary file, upstream in 900 seconds - 15 minutes.

Yes, that's wireless going faster than standard (wired) ADSL. My home cable modem restricts uploads to 128 kilobits. Normal ADSL peaks at 256 kilobits upload. It's absolutely impossible to get a 3G upload to reach its theoretical 64 kilobits. But this, using ordinary 3G frequencies and masts, was averaging 318 kilobits per second, upload.

COMMENT

I could run a mobile Web server with this.

Would I want to? Well, heck, yes! - it's not yet clear what the costs will be, but from the trial already under way in Raleigh-Durham in a 12,000 square mile area, people are paying for their service, and the top level cost is around $80 a month. Average may be half that, or even less.

Now, do we suddenly understand why DSL prices are dropping?

Well, not quite. DSL prices are dropping, mainly, because the world's big telcos are anxious to postpone the Big Dip In The Pocket. They are desperate to avoid finding the capital to convert all their copper to optical fibre. And maybe they won't have to, if they can keep the copper profitable while people find ways of sending a 100 megabit signal over a ten mile copper link for fifty pence a month. So what they will do is fiercely extend the speed of ADSL to four and then six megabits per second, and the range of ADSL from five or so miles to ten, and on. And enough of us will get excited, perhaps, to make politicians feel that there's no mileage in pressing these people to upgrade the network, not just yet.

Except, the world is going mobile. We don't want to be restricted to the TV room, or the home office. We want to work with all our data not just downstairs in our own sitting room, we want to watch TV streaming but not just in the entertainment room. We want to go further. We want to go down the pub, or into the local shops, and still have access to our own data, without having to log off, log on, and pay some money-grabbing hotspot provider.

And we want to send our data to other people. Webcams. Voice calls. Clips we took with our video hand-phones. The twenty 8 megapixel photographs we just took in the town centre or on the beach.

Never forget upload. The telcos don't yet understand how important it is, but it is, and they'll learn ...