BT, broadband and electromagnetic noise

Just been to a BT Wholesale showcase down near Gatwick, where they've been showing off some of the new tools for their ISP clients that will come along with 21CN.

The basic gist of it is that it'll supposedly be much easier for ISPs to be able to sell differentiated services - i.e. the more you as a punter pay, the better the quality of service you'll get. This should be good news for SMEs in particular, as such differentiation usually only benefits the corporates.

They also gave a long spiel about the various things that will affect your broadband throughput, mostly because of electromagnetic noise. Faulty electronics are a big problem - faulty TVs, faulty set-top boxes, faulty electric fences, faulty street lighting, faulty dimmer switches, faulty power supplies - but also "some" Christmas lights and "some" fluorescent and (alarmingly) low-energy lighting. Not to worry, though, it's only "rogue pieces of equipment" we need to watch out for.

One interesting aside - apparently BT once found one of their end users had a faulty TV, which was playing havoc with his neighbours' broadband out to a 200m radius. BT bought him a new TV. Not sure they wanted me to pass that on, but there you go.

Anyway, BT is trialling a cheap piece of kit it says will mitigate the electromagnetic noise issue for many people - essentially a self-install interstitial plate called the i-Plate (they're confident they won't be sued by Apple) that can be bunged onto/into telephone sockets. BT says it will distribute this to ISPs at cost. For end-users who have old-fashioned bell wires going into their master socket, so to speak, this i-Plate will supposedly make as much as 1.5Mbps difference.

We were treated to a demo of the whole electromagnetic noise thing with a rather complex set-up, all being monitored to show us throughput changes and using a standard 8Mbps-max line. To start, we saw an unidentified third-party modem (poor modems are another problem, of course), which gave 2.9Mbps throughput. Switch that to a BT modem and up we go to 3.7Mbps. Then, on went a bog-standard fluorescent light nearby and down went the throughput to 700Kbps.

Now, they'd actually done some jiggery-pokery behind the demo board to "simulate" what would happen if the fluorescent light was running off a faulty power unit, which is why the throughput dropped so much. I'm no engineer, but it's awfully hard to see what sort of analysis we were supposed to glean from this complicated set-up - what's the biggest problem? The fluorescent light? The "faulty" power supply? What was behind the board?

Anyway, never mind. They then switched the line to a socket fitted with the i-Plate, and - because "unfiltered" bell wires pick up noise that standard filters don't remove - up went the throughput to 2.6Mbps. All very educational, if wildly difficult to interpret.

Of course, the biggest problem with the i-Plate will be convincing punters that it really is easy to self-install - it is, but that's not the point. People don't see electrics that way, and even if no rewiring is required they may think they have a real job on their hands. Also, as a colleague has just pointed out to me, how will they know in the first place whether or not they have old-fashioned bell wires going into their sockets?

Now, about those faulty streetlights...