Tech Detective: Want A Personal Server? –

Intel’s Silicon Valley headquarters are a sharp blue, broken up by panes of mirrored glass. Inside, the colors are the same, only softer. A bank of steel elevators open onto a sea of gray cubes, the hallways are painted a light blue. And on a corner table of Roy Want’s fifth-floor lab is a silvery lava lamp, blowing bright blue bubbles. Aside from a yellow water bottle tucked under a desk, it is the only splash of color in a windowless room of computer terminals and dry erase boards.

Still, he pushes the bubbling blue lamp aside to make room for a PowerPoint display loaded inside his laptop.

Want is a principal engineer in Intel’s ubiquitous computing project. That means he’s trying to connect people to each other and to the Internet wherever they are and whenever they want. He’d doing it with a device no bigger than a deck of cards that he calls a Personal Server. It’s a circuit board with two processor chips, a memory card, a Bluetooth radio chip and all the software to make it work. Weighing no more than a cell phone, it’s a personal computer minus the keyboard and display that runs on a lithium ion battery for about five hours before it needs to be charged.

While the slide presentation boots up, he pushes his sandy blonde hair out of his eyes. He’s a geek, with a slight British accent, laughing at his own pocket protector jokes. There’s a sign reminding his engineers of the University of Washington’s computer science TV show, on every Wednesday at 10 a.m. Work-flow charts and things-to-do lists cover the walls. He’s approachable, in khakis and a red shirt, and serious about his work.

Though Want joined Intel Research two years ago and found internal funding for his project six months later, he began his research in ubiquitous computing in the late 1980s at Olivetti Research. After finishing his Ph.D. in computer science at Cambridge University, he joined Olivetti where, in an era before the cell phone he began developing smart phones. Olivetti’s smart phone system could locate a person within the building and then route his call to the nearest phone. Later, at Xerox PARC Want’s team created and built a computer that was not only aware of its location, but also of the people and devices sharing the same space. It could communicate wirelessly with the other workstations to, for example, print documents to the nearest printer.

Those were the days before PDAs hit the market. Sure, you can store phone numbers and a schedule on your Palm, you can trade business cards and jot down notes, you can read email and then zap everything back on to your hard drive at the end of the day. But that’s not the idea behind mobile computing. For Want and his team, it’s not about storing and transferring data. It’s about making a fully operational computer, one with all the capabilities of a PC but in a much smaller package. That’s why the Bluetooth radio inside is so important-it will transmit the server’s data to any workstation. Any computer becomes your computer.

“Never underestimate the bandwidth of a guy with a disk in his pocket,” he says. It’s unclear whether that doubles in Want’s circles as a pick-up line. Whatever its hidden meaning, the ultimate point he is making is that the wireless revolution is finally here.

Imagine, Want says: You’re making a presentation at a conference. Maybe your slides are stored online behind your company’s firewall. Maybe the server is down, maybe the connection is slow. Instead of trying to link up to the home office, the presentation is stored in a personal server hidden in your briefcase. When you get to the podium your server communicates with the laptop at the podium and the slides appear on the desktop. Or, you’re at the airport. Your flight information is loaded onto your personal server, but you can’t find the gate. Go to the nearest display and, while the two devices communicate, the display will show you where the flight is and then transmit and store the updated information back into the personal server.

“It’s a socially acceptable wearable computer,” he says.

It’s a vision that’s still several years out, but it’s one that relies on already existing technologies: High-density, small volume storage, low-power, high-performance processors and standardized, high-bandwidth radios.

Want pulls out a handful of prototypes. The oldest is the size of a sheet of paper, but the most recent iteration fits easily into the palm of a hand. He places it next to his laptop and switches it on. It whirrs and stirs, it flashes red then green then blue while it debugs and boots up. Once it quiets an icon appears on the upper right-hand corner of his laptop’s screen. Click twice on the icon and a new screen appears – it’s Want’s personal desktop. He clicks through some photos of his group’s summer picnic quickly because he wants to get to the MP3 files. First he plays a sonata. Maybe I’m not convinced? How about the “Men In Black” theme? The data is streaming at 600 Kbps from the personal server to the laptop. Want to watch a movie instead? How about the “Star Wars 3: Revenge of the Clones” trailer? With the 512 megabit Flash card he’s using, he can store 50 of these five-minute movie clips. Then he pulls up the proof – his published papers in the area of mobile computing. Any changes he makes to the documents will be stored on the personal server. And even after all that, the 206MHz processor is running at 130 MHz. The personal server has been programmed to use only enough power to get the work done.

Still, proof in the lab does not translate easily into technology on the street. Wireless networks are not ubiquitous, and even where they do exist they are rife with security problems. Consumers are not willing to pay extravagant sums for wireless service, and current power storage technology means that users would have to recharge their servers like they do cell phones. And the biggest hurdle of all: All devices everywhere would have to be compatible with the personal server. That’s a lot of programming and hardware retooling.

Want is undeterred by this seemingly mammoth hurdle. Maybe the first personal servers would exist in a closed wireless network, he says, like the floor of a manufacturing plant, where the personal server could track inventory and output. On a wider scale, the first personal servers might be embedded in cell phones. There are already 500 million cell phones out there, and each of those needs to be recharged at the end of the day.

Clearly, there are technical and market issues to be worked out. Right now Want’s research is just research. Intel has no immediate plans to manufacture a line of personal servers, although it has always been in the business of making chips and developing devices that use those chips. Want’s Personal Server uses at least three.

For its part, Intel is investing heavily in Want’s vision of ubiquitous computing, although it would not disclose his project’s price tag. It will begin installing WiFi chips into its processors beginning the first half of 2003, while Intel Capital has announced plans for a $150 million 802.11 investment fund. Intel Capital is scouting technologies that enable sensor networks, but hasn’t made its first investment in that area. It has seeded a number of Bluetooth companies including SyChip, a maker of integrated circuits for wireless Internet devices, and CSR, a company that designs and sells low-cost single-chip radio devices for cell phones and laptops. To fuel these devices, Intel has brought five fuel cell companies into its portfolio, including PolyFuel, a maker of micro fuel cell power systems, and cap-XX, a supercapacitor developer.

There’s a knock at the door, another Intel researcher, a clipboard tucked under his arm. Ant is preparing another demonstration of the Personal Server. This time, it’s already booted up, running, still sitting on the table. He’ll go through the same PowerPoint presentation. This time, however, it’ll be running directly off the Personal Server.