Intel's PANs, WLANs and soft radio
Our Developer Forum notebook
We used to come to Intel Developer Forums to sniff out speeds and feeds. These days, the most interesting stuff is tangential to Intel's traditional PC business, which has now become so competitive that Intel can't afford to leave an announcement on ice for a major keynote. If the two coincide, all well and good.
But with this PC business in its most serious recession ever, Intel also appreciates that helping infrastructure and markets grow is more important than tightening up the process and ratcheting up the clock frequencies.
So it makes some guesses - some perspicacious, some simply hopeful, and a few quite hilariously hope-less. We'll update this roundup throughout the remainder of the show.
Ultra Wide Band
Back in February, Intel gave the first public demonstration of UltraWideBand networking, which transmits data in short pulses over a wide range of the spectrum. GPS systems didn't stop working, and the world carried on much as before.
UWB performance drops off dramatically: the 400-500 mbps throughput at distances of 10 metres falls to about 25 mbps at 20 metres. So this clearly isn't a cellphone replacement technology, when rural cells span 20 miles. You can build more cells, but technology isn't the problem here, as much as politics: more cells means more planning permits. However, it's a libertarian's wet dream, because they can dream that only an artificial government-created scarcity stands between us and nerdvana, when in fact this artificially-created scarcity stands between us and a mess that looks even worse than the current cellphone free-for-all.
Intel's Carol Jacobsen described it as a promising technology for high speed, short range connectivity, and a bad choice for WLANs. Intel envisages it in a point to point and PAN (personal area networking) successor to Bluetooth. UWB is simply the physical layer of the stack, and a IEEE working group is expected to define the MAC interfaces by 2004, with the first standards-based products hitting the market the following year.
But not without a hitch, if we can take Sean Maloney at his word. The Brit runs Intel's network division and told us yesterday that every new network technology gets a good kicking before it's finally accepted by the market. Bluetooth had just received such a kicking, he said.
Maloney dissed suggestions that we have years and years of excess capacity in our communications infrastructure after the latest splurge of capital investment, and he expected to see growth in the network market. Where, I asked him? Most of the decisions for deploying 2.5G and 3G had already been taken. Maloney said he expected explosive growth from 802.11 networks. (Intel is building 802.11 support into its mobile chipsets).
That was the only mention of Bluetooth, which is odd, as figures here suggest Bluetooth will be shipping as many units in a week as 802.11 will in a year. (Note: this is a Bluetooth vendor writing, and can hardly be expected to be unbiased, but unless cellphones are banned overnight, the trend is clear).
Where Intel should step in, is to create a clean and free Bluetooth stack for every PC operating system under the sun. Microsoft has veered between dissing Bluetooth and hugging it to death, and while device to device connections are OK, device to PC connectivity is a mess.
Maloney's point, that process improvements (9 nanometers imminent, 6 next) permit much more integration, was amplified in a Labs Wireless session later yesterday, which reiterated the goal to put the analog parts of radio onto the chip. This "soft radio" project could take seven years.
This was one of many talks that grappled with the problem of intelligent roaming between network technologies. How could we take our notebook PC from a meeting at a wireless hotspot, and keep our connection on the packet data network provided by the cellphone carrier?
"If I could solve that problem I'd be the richest man in the world," he said.
"I think Nokia already has," whispered my friend from The Guardian, who pointed out that the Finns already have an 802.11 card with a GPRS SIM built-in.
And they're not alone, for the industry crossed this river about a year ago. Carriers were once terrified of 802.11, and wanted to squeeze every second of airtime out of users. Now they're broke, and 3G revenues are a long way away, and squabbling over pennies isn't so clever. Every carrier is embracing WLAN hotspots, and we'll see some interesting client devices as a result.
But perhaps not quite as Intel envisages. One of its concept designs, a "Digital Briefcase" was a lunchbox-sized device that included biometrics ... a video camera! .... with MPEG playback! ... wireless capabilities! ... and a color screen! In other words, it looked like one of these, which is somewhat smaller than a lunchbox, as you know.
Aaarggh. Sometimes the most instructive gift you can possibly give a labs-bound wireless engineer isn't an oscilloscope, but a return plane ticket to Finland or Sweden. ®
Sponsored: Optimizing the hybrid cloud