Cellnet GPRS broadband offering sounds strangely narrow
27Kbps? Don't all broadband at once, obviously...
Signs are emerging that BT Cellnet is well and truly on the bleeding edge of high speed mobile data. The company trumpeted its first GPRS (General Packet Radio Service) call mightily last year, but what it thought it was going to be doing then, and what it's actually doing now, are two separate things.
On the way in the company was - in common with practically everybody else in the GSM world - babbling happily: "Initially GPRS handsets and data cards will be capable of operating at speeds five to ten times faster than on current GSM networks. Even faster devices are being developed for the future." Ahem. This from a current BT Cellnet document headed, appropriately enough, "GPRS - The Facts."
A couple of weeks ago Nokia was telling us that the speed you'd get would more likely be 43Kbps. BT Cellnet has been working with Motorola on GPRS, so its number are likely to be different, but it's now quoting an initial 27Kbps, ie. Just three times current GSM data speeds. That multiple, incidentally, sounds suspicious, as it indicates three 9.6Kbps timeslots, so the numbers seem to be stacking up with Nokia's 3x13Kbps (and no, we still don't know precisely where the extra 4Kbps comes from).
BT Cellnet is still talking about data speeds going up to 171Kbps, but isn't saying when. And we've got quite a bit more downscaling. BT initially anticipated having the service "fully rolled out in the first half of 2000," with a "full commercial launch" by the Summer.
Instead of this, it's now making the service available initially to businesses in the south of England, rolling this out to the rest of the country subsequently, and promising a consumer service by the end of the year. BT Cellnet last year claimed it was six months ahead, but as the planned deployment looks suspiciously like a beefing up of the trials it's been running for the first half of this year, that lead may turn out to be illusory.
The slower rollout will also have the effect (obviously) of reducing the potential number of users to start with, and maybe buying time to get the bandwidth up before everybody wants to pile in? One has one's suspicions. And maybe not just suspicions.
We've heard from Mike Ritter, CTO of Metricom, on the subject. Metricom is of course biased, because it's been in the business of wireless Web access in the US for some years now, and it's therefore Metricom's lunch the cellular outfits are proposing to steal. But as Metricom is actually delivering, Ritter's worth listening to: "...even with the secret sauce addition to bring the raw data up to 43Kbps, if you actually run TCP/IP protocol over it (that would be web browsing) you would be lucky to get half that data rate actually delivered to the user."
Given that the BT Cellnet and Nokia numbers are ballpark the same, that'd give you 14-21Kbps. But Ritter's even more pessimistic (or optimistic, considering where he's sitting): "There have been studies published in the open literature about the performance of the GPRS protocols that say, depending on the data model, it will actually only deliver 1/3 or 1/4 of the raw data rate to the user. Add al this together - the carrier can displace eight voice channels (not gonna happen for a long time - batter and DSP problems) and you still don't get much past 40Kbps to the user.
"One thing we learned very quickly was that if you don't optimize your network for TCP/IP traffic it won't work very well..." This may relate to Nokia's launch of its All-IP IPv6 core for 3G networks at the end of last month, but as implementation of that won't happen till 2002, we foresee something of a missile gap in the interim. ®