News

Gigabit Ethernet To the Pocket - and other GSM dreams from Mobile World Congress

Here's a neat new cryptonym for you - GETTP. It stands for "Gigabit Ethernet To The Pocket, and it's the prediction by Kevin Sheehan, CEO of Hatteras Networks, for what we'll see at Mobile World Congress - not this week, but in 2018.

OK, nice game, let's play. What do we really think will be the big story of 2018 in mobile technology?

Looking through the list of "futuristic" ideas Sheehan offered in his keynote this weekend at NetEvents press summit pre-MWC, one thing stands out by its absence. See if you can spot it:

A nice little list, but I think it leaves opportunities for new ideas. My suggestion: think energy.

Nobody looking at IT in the second decade of the century need make a single prediction, if they don't start off with the energy factor - and when I suggested this to Sheehan, he agreed at once.

But it's an issue which the industry isn't facing up to, probably because they still regard mobile devices as using too little power to matter. In one sense, that's right, and is half the picture we have to paint; in another sense, it's entirely wrong.

I asked Nuance marketing vice-president, Robert Weideman
what he thought about the power implications of his plans for Tegic voice recognition in every phone. His first response fitted right into the picture I'm seeing: he thought it was irrelevant. Weideman knows his stuff - he's been working in speech and recognition tech from the days when Nuance was still called Scansoft - and so I asked why?

His answer, after a bit of careful thought, suggested that nobody is actually thinking of trying to use speech recognition for anything much beyond giving commands and control instructions to the phone: "dial this number" or "find this info" - useful! yes... but brief bursts of processing. But that's not all you can do, and people will think of applications.

Just to give a sort of Sci-Fi speculation: would it not be useful if your phone was actually listening to you all the time, and acting as a sort of super-secretary? How would that work?

Well, imagine a phone of the future. It doesn't just have voicemail. No! it has your diary inside it, and it has your voice, too. So if someone rings you, and you're busy, instead of "please leave a message after the tone; if you wish to re-record your message..." it actually answers the phone.

Your contact, Jim, says: "Guy! are you free for lunch Tuesday?"

The super-secretary understands this (naturally, it's a Nuance application here!) and has a problem. Today is Tuesday, and this could be the wrong day. Disambiguation is needed, as they say. So, using your voice, synthesised, it asks: "This Tuesday?" and Jim says: "No, next week" or "No, the 26th" - and your secretary checks that day. At that point, just like a real secretary, it pops up a message: "Would you like to have lunch with Jim next Tuesday? You have an appointment with the staff at 11, and an interview at 3 pm." You hit "yes" or "No" and carry on with your main conversation.

Even better, imagine you are chatting to Jim, and the conversation comes round to Betty, and after a bit, you both realise that you need to bring her into the conversation. The supersecretary has been listening, and has established which of your contacts you're referring to, and has pulled out her details. She's contacted Betty's super-secretary, to find out where Betty may be. So, when you say "Is Betty available to join the conversation?" the supersecretary knows "yes," or "No, but she'll call you in five minutes."

Now, it doesn't take much imagination to realise that the technology to do all this is very nearly ready. My phones already have my diary on board. They have the identity of most of my 2,000 most frequent business contacts attached to their phones, and they are quite capable of understanding common words and phrases. All we're talking about, is a moderate improvement in the whole chain, and it would be possible.

Except, of course, you can't build a battery big enough to keep that speech recognition constantly running in the background, and still be portable. Well, a car phone, maybe, but not something you can hold in your hand.

That's not the only problem facing the people who run the power budgets for mobile. The main one, of course, is the one which says that a PC should use no more power than a phone; and many people think this means that the personal computers of 2018 will actually be phones. So in that sense, it's good news for the mobile networks; but they won't see it that way. They make their money from call revenues. If your phone flattens its battery because it's playing PC (taking dictation, planning a route through the city, surfing the Internet, showing you video clips, calculating your expenses and so on) then bang goes their income.

I've heard some weird stories of Tesla coils being used to "transmit" electrical power to hand-held devices. Samsung, at least, is aware of the problem and working on seeking a solution! - but even if it actually works in a non-alarming way, I'm not sure that it's for everybody. Apart from suppressing the sparks, there's the issue of how much it costs.

In ten years' time, I suspect we'll probably be finding out that we generate our own power whenever we can. There are nice devices which will charge your phone battery from sunlight. But they won't be tiny, portable things, and they probably won't work well indoors. And we'll need all the power we can get for life-support issues like home heating.

It boils down to a world in which cpu-hungry applications have to be run off-device. That raises the question of how much data they have to swap back and forth, and how often; and how big the mobile data "pipes" are.

And that, in turn, brings us back to GETTP - GigE on mobile. Don't get too excited! - it may even be possible, but it's unlikely to make a terrible amount of difference. Oh, it will make a difference to network latency (if they update the protocols, that is) but I'm not convinced that you'll be able to run GigE for more than a few microseconds a dozen times a day. All that data has to go back to the wired network at some point, and it's expensive to maintain.

If you're designing a nice application for mobile which transfers a gigabyte a day, expect the mobile networks to get after you with a hit squad. They won't be able to provide carriage at a rate your customers can afford till well after 2018.

Maybe 2048, but I wouldn't bet on it...