Right now you're using a 30-year old computing paradigm. It consists of a typewriter (and mouse) for input, a TV (and printer) for output, and a tape (or CD) for storage, all connected by wires. But a TV, tape recorder, and typewriter do not make up a computer. In Gordon Moore's famous 1965 Electronics magazine article, where he propounded what has become "Moore's Law" (chips get faster-and-faster-and-faster), there is, in fact, a drawing of a home computer. It is held by a salesman in a department store. It has no screen and no typewriter. It is, simply, a box.

Acomputer is intelligence mediated by software. Its input could be a sensor, and its output an alarm, while its storage could be a database updated locally. Why do we need wires between these devices? Why not just move the data wirelessly? When all this data is connected under the Internet protocol, it's also available to you from anywhere in the world, via any and every device you take with you on your way.

This is the world of "always-on" - applications depend on an always-on wireless network and the always-on connectivity of Internet broadband. The idea is that one-chip sensors pass data via radios to a server that can then calculate their meaning and take action. The chips are the clients, your PC is the server, and a wireless network is the computer.

Always-On vs Alzheimer's
Home entertainment is one of the easiest, and simplest such application to understand. This is the area Microsoft, Hewlett-Packard, and others are targeting with their "home media server" concept. The PC stores all your music and (in time) your movies. The PC interface is used to control them, and a wireless network (Bluetooth, Wi-Fi, or Ultra WideBand) is used to deliver them. In addition to these "brand name" solutions to that problem, there are also "new name" solutions, like the Janja System from Digital Deck.

While all this represents a great new horizon for gadget freaks and other early adopters, it's far from the most exciting thing out there. Because what if the true "main computer" in your home - your own mind - starts to disappear?

Intel recently put $1 million into a plan to develop just such applications for Alzheimer's patients. The idea is to place one-chip sensors throughout the patient's home and use data from them to tell caregivers when they need to intervene.

"A lot of the sensor technology has been around a long time. It's the ubiquity of the network that makes it powerful," says Eric Dishman, manager of proactive health research at Intel Corp. Dishman helped father the Alzheimer's application after a study of early broadband adopters sought ideas for digital entertainment applications, but found that everyone over 40 who was interviewed shifted the conversation to the problems of aging.

Beyond the Alzheimer's pilot, Dishman says, he has found a host of other tele-medicine applications for these networks in detection, prevention, monitoring, and care-giving for our rapidly aging population.

It's a great vision. Stay at home longer. But, once a wireless network is in place to handle the data, you have a platform for always-on wireless applications that are much more prosaic:

• Sensors might measure the moisture level in your soil and turn on your sprinklers only when the grass needs watering.
• ID chips on your milk might alert you when the milk is about to spoil, or chips on your clothes might tell you what's in your closet.
• Motion sensors around your home might tell you when a burglar is entering your house and alert police.
• A chip in your keys, wallet, or anything else you often need and lose might alert you to its whereabouts before you drive yourself mad some morning.
• Sensors monitoring your blood might alert your doctor before you have a heart attack, or alert you when you just need a candy bar to keep your sugar level up.

'Proactive' or 'Pervasive'?
I call these "always-on" applications because they depend on wireless networking to work. An 802.11 connection that is always available can take data from radio chips, analyze it, and either send an alert or take some other preset action.

Intel calls such systems "proactive computing," while IBM calls them pervasive computing, but the concepts are very similar, with IBM scientists very active in this area, looking for interfaces as small and personal as Chandra Narayanaswami's Linux wristwatch.

Narayanaswami, manager for Wearable Computing at IBM's TJ Watson Research Center in Yorktown Heights, NY, built the watch to address several challenges facing always-on applications:

• Calculating data from sensors puts an enormous load on back-end processors, whether they are a central server in the home or a remote server accessed via the Internet.
• Managing the database can be tricky, especially dealing with questions like versioning between a central store and the data available locally.
• Sensors use power, so they need to be replaced or recharged somehow.
• The security and privacy of data, especially medical data, must be carefully managed.
• Where and how will you interface with these applications?
• What about the cost?

Some of these problems will find ready solutions. Moore's Law will eventually solve the problem of processor speed, and if always-on accelerates the PC upgrade cycle, Intel will be pleased. C. Mohan, an IBM Fellow at the Almaden Research Center in San Jose, has been looking to contribute to this development through the use of caching in always-on databases.

"In the traditional Web world not a lot of data is cached," Mohan says, but many banks and brokers, not just broadband content producers, are finding that caching databases closer to customers cuts response time and improves performance. "In a medical application, patients could cache databases in their own computers, analyzing current data locally, or hospitals could query multiple databases to find patterns of appropriate treatment," he says.

Business Goes Always-On
Still, high costs lead Mohan to believe that always-on apps will emerge first in the business environment, with radio frequency identification (RFID) applications that track inventory leading the way (see also Jeff Goldman's "The Price of RFID" in this issue).

RFID chips tag products, and companies can follow their inventory through the sales channel and out the door. You can have cheap, passive tags that must be accessed via a radio in the store, or you can put the radio on the tag, creating an "active" tag. In a recent story for the IEEE, Vince Stanford wrote about using RFID to tag personnel badges, library books, packages, documents, and cars. Early versions of the passive technology are already available in New York's "EZ Pass" toll system and in pet IDs, read either by scanners on toll booths or handheld scanners that pet owners and animal control officers carry with them.

Combine all this with a GPS system and you can track your kid or your grandma. While GPS may not have the accuracy for computer-controlled driving, it won't be long coming.

What About Privacy?
Many of these applications are attracting "blowback" from privacy advocates. A California state senator, Debra Bowen, scheduled a hearing on the privacy implications of RFID for August 18. A privacy group called CASPIAN, originally formed to protest shopper "loyalty cards", has now gone on the warpath against RFID, and Wal-Mart recently backed away from an RFID trial, fearing a backlash.

What this means, unfortunately, is that the path toward many of the best always-on applications includes a detour into the world of law and politics. The easy answer is to require strict enforcement of privacy laws against business, the principle being that once you buy something and leave the store, its tag, along with any information that tag collects, becomes your property. Unless such laws are strictly enforced, protests and paranoia will only grow.

Another way to make consumers comfortable with always-on technologies, however, is through great applications that they can buy and with which they own all the data. Last year Xanboo Inc. of New York announced a technology platform for home-based always-on applications linked to its own line of wireless cameras, sensors, controls, and thermostats.

Xanboo's products are tied to specific applications, like energy management or watching seniors, pets, and children. This means you can buy something to do a specific job, and then have a platform that can be used for others, which is how the PC was sold. These in turn link to controllers and management systems. The company is building a vertical-market network of dealers and distributors.

But Xanboo has some problems. It is tied to a central server system, to cable and telephone companies. You can't yet buy it, build it, and forget it.

There is an alternative. Right now, under my desk are four boxes dealing with my home network. There's a PC, which runs applications (including a firewall); there's a router; there's my ADSL modem; and there's an AlphaShield privacy box.

What if you offered the market a computer-router that combined all these functions, with a few extra slots for new applications? The incoming network could be a cable modem connection, a DSL line, or an 802.11 Wi-Fi setup. The outgoing radio would give all computers on the network the benefits of the firewall. By using Linux as the operating system, the total consumer cost could easily be kept under $500.

As with Xanboo's controller, this device would not have to sit under my PC. It could be mounted wherever there is electrical power and an outgoing network connection. It should also be expandable, taking advantage of Dishman's point that many always-on applications are computer-intensive. Its dimensions and nature make it a home server, but by making a variety of existing broadband-controller chores standard, it can find a ready consumer market.

This need not strain the computing capacity of the server, either. Thanks to new chip-making technology from Sun and IBM, you could have always-on servers that look like refrigerator magnets, or that sit at the bottom of your ceiling fan, scanning for tags and reporting back to you.

The value is there. You can build a wireless network that might link all the RFID tags in your home, and others you might buy as you buy safety pins. Software developers could sell solutions for tracking clothes, food, CDs, your keys and wallet, the perimeter of your home, the possibility of fire...some of these applications are now available but quite expensive - others don't yet exist.

The Role of Voice Activation
Why not have a voice interface to all this, instead of a mouse? Your voice could provide security, as well as ease-of-use. Narayanaswami's work would bring watches, cellphones, PDAs, or laptops into always-on networks as interfaces using current technologies. But IBM is also a leader in voice-control technology and it's seeking a consumer application.

After all, if your network is your computer, and its applications are everywhere, why must you still go to a keyboard and screen in order to interface with it? What could be more secure than using your voice? The failures of such firms as Lernout + Hauspie, and the difficulty of creating voice interfaces that really work with a wide range of speakers, should if anything be a spur to development in this area. With single-chip microphones linked to your network, it can be trained to understand only your voice and your commands, providing security as well as a better interface.

I hope that by now you're getting some idea of the power of the world of "always-on." Potential applications are all around us, all waiting to be born. Both industry leaders and startups are creating applications in business, in medicine, and in the consumer realm. There are challenges, both in technology and law, but there is also a huge demand for what it can do.

A new industry is waiting for you to create it.

References

www.digitaldeck.com

www.xanboo.com

www.siliconvalley.com/mld/siliconvalley/6379256.htm