xG, xMax, xetera...

I'm back from my holidays - hurrah! - to find not a lot of news, but a steady stream of interesting emails.

I'm back from my holidays - hurrah! - to find not a lot of news, but a steady stream of interesting emails.

It seems xG Technology is still exercising the imaginations of some readers. I have two types of feedback when I write about this Florida company claiming enormous breakthroughs with its xMax wireless data.

One is the "You're so wrong, Goodwins, you shouldn't be allowed to listen to wireless, let alone write about it. xG is the best thing ever, you'll see" (only much ruder, normally).

The other is more along the lines of "I'm a professor of electrical engineering, and this whole thing seems utterly unbelievable. Why are people taking it seriously?". Now, this second kind of correspondent tends to back up their logic with mathematics and useful information while the first doesn't - and don't think I haven't asked - but should I let that influence my opinion?

One of the most seductive arguments for xG being at best not much better than existing systems is tied to the Shannon capacity curve.

This is a graph illustrating a physical fact of life for wireless engineers, that the more noise and less signal in a channel, the less data you can send. It's not a straightforward relationship: the curve itself marks a boundary beyond which you cannot usefully send data no matter what you do. Decide on your power level and characterise the noise and interference on a channel, and the curve will tell you how fast you can go.

This limit isn't something that can be circumvented by cleverness, any more than the speed of light can be broken by a particularly ornate sort of rocket. If you're doing wireless communication, the Shannon curve is your speed of light. Most current state of the art modulation schemes get pretty close: xG says it is thousands of times better than the current state of the art, which puts it firmly over the other side of the curve and into Star Trek territory.

Confusingly, xG has also published a bit error rate to signal-to-noise graph -- the standard way of describing physical performance -- that shows (without explanation) that its technology is, at best, slightly better than most existing techniques. That graph shows that xG is within the Shannon limit (how could it not be?). Yet the company also says that the graph doesn't demonstrate the "significant advantages" of the "xMax system gain".

That's like saying you've built the world's most fuel efficient car, publishing standard test results that shows it guzzles as much petrol as a Bentley, then saying "But these results don't reflect the inherent advantages of our automobile, which really does go a thousand miles on a cup of four star".

One has to ask how a system that is tested to do ten miles to the gallon will achieve a thousand miles per cup.

Two further mysteries are connected. As I've said before, market researchers Frost and Sullivan have given xG their European Technology Innovation Award for 2007, and are bigging xG up even more - which must mean there's something they know that I don't. Most people do, of course, so I've emailed them to find out what it is.

And second, my lead on the Shannon curve came from a blog called Wireless Fact and Fiction, which had a short burst of very cogent explanations around the physics of wireless as directly related to xG's claims - and then closed down abruptly a couple of days ago, just as I got back from holiday, with all content deleted (Google cache still has the full story at the time of writing, if you fancy a modicum of fun rootling around). [UPDATE - Wireless Fact and Fiction was pulled, the author tells me, purely because it was taking up much more time than it deserved, and it may come back later]

It's all most odd. But do keep writing in.