Features

Intel moves to standardise phones? "No, it increases diversity"

Get your head around this one, if you can: Intel is building a reference platform for mobile phones with Symbian and with Nokia. Does this mean all phones will, really, be the same thing? Not at all! - Erik Anderson preaches a gospel of diversity.

The announcement was confusingly banal:

SYMBIAN EXPO, London, Oct. 5, 2004 - Intel Corporation, Nokia Corporation and Symbian Ltd. announced today a collaboration to bring smartphones based on the Nokia Series 60 Platform to market using Intel technology as part of Intel's recent membership in the Nokia Series 60 Product Creation Community. Also, Intel and Symbian have agreed to invest in the joint development of a reference platform to enable a new class of 3G devices based on Symbian OS and Intel XScale Technology.

So the phones: they'll all be the same, right? I ran into Eric Anderson, Director of the Communications Processors Biz Unit - which is in the cellular and handheld group in Intel - which is part of the comms group - at London's Excel centre, and he said: "This is a positive thing for the industry."

That's OK, then. "What everybody has been struggling with, as the boom in small technology expands - not just phones, but memory, small disks, batteries, cameras - is the tremendous amount of engineering work to be done."

The traditional way of differentiating one phone from another, he says, is to make them different. Well, yes ... but isn't the opposite "making them all the same?" "No. Today, you differentiate by bundling different features. You may say 'I want this to have this DRM package, those applications, these hardware abilities' and then take the same platform, but make something with a completely different profile for a different market."

Intel clearly feels this is an important opportunity. It rather blew the first attempt: a reference platform it created 18 months ago for the Microsoft Windows Mobile platform, which has not surfaced in production. Indeed, senior Intel people, asked about it, affect ignorance of its existence. The deal with Symbian, they imply, is the real one.

Money, they say, goes where your mouth is. Nobody is saying how much money went here, but you don't have to be psychic to see that Intel is sponsoring this development, and heavily. It wants into the phone business, and it wants to "platformise" that business.

Platforming is not making things dull and the same, however. "It's really a question of plug and play," Anderson says. "There are all sorts of hardware and software options, and they can all be included or excluded, first by the phone makers and then, later, by the mobile network operators."

"There's a tendency to confuse the PC market - with one standard device - with the CE market, where the capabilities are standard, but what the user sees is dramatically different," said Anderson. "Tyres, airbags, all sorts of standard components, go in different vehicles, but prices and profile vary enormously."

And so, he says, you'll see similar changes in both the CE and phone industry; different aspects will become "platformised," while individual product will be differentiated on much more specific criteria. "Gaming devices will be differentiated on selected criteria; enterprise messaging corporate internet device will have different criteria - and yet, those two devices will be able to exchange email, pictures, other content, freely."

The question, says Anderson, is: "How do you have a configurable software system to do all these different things, and how do you deal with the interoperability challenge?" The operator wants things to work the same on all devices. But they want the devices to be more different. "Every step of the value chain has upped the ante in terms of the value. Everybody wants to differentiate - but at the same time they want to interoperate. And these are all resource consuming requirements."

So the challenge is to reduce the resources consumed in developing and testing. Do it once. "How do you do this without doing the same work again and again and again, is what the hardware people and the network people want to know. This platform work is designed to reduce repetitive, non value-added work."

Where the reference platform comes into play: you bundle the standard things. "People are not competing on them. People just want them to work well, reliably. So, for example - DRM didn't exist in phones two years ago; in two more years, it will be standard, no effort. This is the sort of thing that drives us to work together; we see the common good of it."

The Intel-Symbian-Nokia partnership is, Anderson says, just the first - there will obviously be more. "For us at Intel, developing ref platforms is not a new thing; in wireless, it has been quite new. Application processors add power across many devices, not just wireless. And so far, we're not making, specifically, comms processors. In the case of comms processors, it's a newer area for us."

The platform has a limited life, obviously; the reference platform concept, however, is like Margaret Thatcher. It will go on, and on, and on - or at least, that's the plan. "The silicon cycle is longer than the software cycle. I think it's official that Symbian will upgrade its platform three times a year, and probably Series 60 will refresh often - but the silicon will be about 12-18 months before it uprates."

So, when will we see Symbian 3G phones all with the reference platform in? Hard to say, but again, probably around a year to 18 months, he thinks. "But it's a developing platform, and will continuously change. The important thing is that the upgrade shouldn't cause problems."

Intel's focus is on the actual silicon. That means the support for applications and communications: the core asics, as they say. "We don't include the things which the OEM brand manufacturer would put in, for example, choice of display, camera, keyboard. The S60 platform supports several sizes of displays; that's a Nokia software capability. We produce an engine capable of supporting a wide range of those things; easily - the idea is that it must be plug and play."

Part of upgrading the platform will be making sure they keep pace with the latest resolution of camera. "A hardware platform needs to be a bit ahead of what the phone makers are taking advantage of." How far? "Again, I'd say the life of a platform is 12-18 months, and you need to be that far ahead. Past that, it sells at a lower price. Too much more than that, and it starts to be more expensive than is appropriate at that time period."

And will others follow suit? "I wouldn't be surprised if others announced soon. There have been many announcements of collaboration in the industry - eg Vodafone and Nokia on Java work, a month ago.. that could be regarded as platform work. We expect it to be a competitive platform market. There are so many operating systems, so many protocol stacks that's so much part of the challenge of the industry - so many device types, and increasingly, different business models. Some will do VoIP, others will harness WiFi. The work for the platform provider increases if there is more diversity."

And will there really be more diversity? "Yes. There was, perhaps, a quiet period after the 3G auctions; a lot of technology was too big. The normal cycle of integration hadn't caught up with what end-users were expecting. But that's ending, and in 2005, people will be catching up; there's action to be had in the wireless space - that's making it dynamic, confusing - we need reference platforms to get through this."

What you can do depends on the device you are holding. Not the underlying "platform" - and the purpose of the platform is to make this all plug and play, easier to do. And more segmented devices means you have the need to achieve scale. "If every small segment has to have its own special components, then the costs are prohibitive. Platformisation allows component makers to sell to a broader market."