Editor's Note: This story was originally posted in the January 2004 issue of Scientific American.

Thirteen years ago, in an article for Scientific American, the late Mark Weiser, then my colleague at Xerox PARC, outlined his bold vision of “ubiquitous computing”: small computers would be embedded in everyday objects all around us and, using wireless connections, would respond to our presence, desires and needs without being actively manipulated. This network of mobile and fixed devices would do things for us automatically and so invisibly that we would notice only their effects. Weiser called such systems “calm technology,” because they would make it easier for us to focus on our work and other activities, instead of demanding that we interact with and control them, as the typical PC does today.

In a home equipped with this kind of technology, readers strategically placed in the bedroom, the bathroom door frame, the stairwell and the refrigerator would detect the identifying data in microchip tags sewn into your clothes and embedded in the packaging of foods and send the data to a home computer, which would take action based on that information.

The computer would notice as you got out of bed in the morning and would switch on the coffeemaker. As you entered the bathroom, the shower would come on, adjusted to your favorite temperature. When you started down the stairs, the preloaded toaster would heat up so that your breakfast would be done just the way you like it. When you opened the refrigerator, the appliance would remind you that you were out of milk and that the tub of coleslaw inside had passed its expiration date and should be thrown out.

Today systems based on radio-frequency identification (RFID) technology are helping to move Weiser’s vision closer to reality. These systems consist of tags (small silicon chips that contain identifying data and sometimes other information) and of readers that automatically receive and decode that data.

The responsive RFID home—and conference room, office building and car—are still far away, but RFID technology is already in limited use. The tags, often as small as a grain of rice, now hide in ID cards and wristbands, windshieldmounted toll tags, gasoline quick-purchase tokens, and electronic ear tags for livestock, and they have begun to appear in auto key-chain antitheft devices, toys (Hasbro Star Wars figures) and other products. They have also timed runners in road races, and last year a company in Mexico began a service to implant tags under the skin of children as an antikidnapping measure.

In the near term, RFID tags will probably be found in airline luggage labels (British Airways has conducted extensive trials), and they may eventually be embedded in paper currency to inhibit counterfeiters and enable governments to track the movement of cash. (Hitachi in Japan recently announced that it has developed tags minute enough for this application.) Meanwhile the retail, security, transportation, manufacturing and shipping industries are all testing or starting to implement sophisticated RFID applications.

But the RFID revolution is not without a downside: the technology’s growth raises important social issues, and as RFID systems proliferate, we will be forced to address new problems related to privacy, law and ethics. Controversy has already erupted: in mid-2003 two major retailers—Wal-Mart in the U.S. and the international clothing maker Benetton— canceled large-scale tests of in-store RFID-centered inventory control systems apparently partly as a response to public reactions that raised the specter of wholesale monitoring of citizens through tags embedded in consumer products.

The Inside Story
RFID TECHNOLOGY is based on the simple idea that an electronic circuit in an unpowered, or “passive,” tag—which requires no batteries or maintenance—can be intermittently powered from a distance by a reader device that broadcasts energy to it. So powered, the tag exchanges information with the reader. Tags essentially consist of a plain antenna bonded to a silicon chip and encapsulated inside a glass or plastic module.

Tags operate differently depending on several factors, especially the frequency at which they function. Initially RFID tags worked only at frequency bands of 13.56 megahertz or lower. Such tags, which are still the most widely used, typically need to be less than a meter away from a reader and offer poor discrimination (a reader cannot quickly interpret a multitude of individual tags grouped closely together).

More sophisticated, higher-frequency tags now enable a reader to quickly identify many individual tags grouped together, even haphazardly—although they are not yet able to distinguish perfectly among all the items in a loaded grocery cart. (The ability to swiftly and reliably scan a shopping cart full of jumbled, closely spaced RFID-tagged items is a major aim of this technology. Once perfected, such RFID scanning should streamline inventory and checkout procedures and save millions of dollars for retailers.)

The higher-frequency tags can potentially be read from much greater distances than their lower-frequency counterparts, although so far their range has been extended only to a few meters (largely because of tag electronics that operate at very low power derived from the reader’s signal, improved antennas and inexpensive high-sensitivity receivers). The updated tags can also hold significantly more information than earlier models, which allows manufacturers to incorporate useful data beyond a mere ID code. The tags can, for instance, use the energy they capture to power an onboard sensor. Tags with sensors that assess tire pressure and temperature while a vehicle is in motion are already in some cars, and Michelin, Philips Semiconductor and BMW are developing prototypes for the mass market.

RFID DEVICES are beginning to replace magnetic-stripe security cards for unlocking doors and granting access to secured areas—especially at facilities with special security needs, such as military installations. The most visible use of RFID, though, is probably the automatic tollpayment systems that rely on readers at toll plazas to scan tags attached to the windshields of passing cars. The reader records the tag’s ID and then deducts money from a prepaid account. These systems are designed to allow cars to zip through toll plazas ideally without stopping or even slowing down very much.

Known as E-ZPass in New York, New Jersey, Delaware and other states, as FasTrak in California, and by different names elsewhere, RFID-based automatic toll systems have been operating for several years. FasTrak, in place on the San Francisco Bay Bridge and on Interstate Highway 15 near San Diego, has been quite successful, but the East Coast E-ZPass system had some early teething problems related to administrative and political issues, not to the technology itself. The San Francisco Bay Bridge system requires drivers to slow to 25 miles per hour while passing the reader, but only for safety reasons, because the tollbooths are narrow. The FasTrak system on I-15, however, operates at freeway speeds and, further, is being used to monitor traffic.

RFID systems are also in the early stages of replacing those familiar Universal Product Code (UPC) bar codes, which are read optically at very short distances to identify products, track inventory and semiautomatize the checkout process at stores. RFID tags, unlike bar codes, can be molded into a product’s casing and can use encryption and other strategies to make them difficult to forge. In addition, some RFID tags permit readers to write new data to their onboard memories for later retrieval. For example, each transaction between reader and tag can record the time, date and identity of whoever accessed the tag. This capability should be useful for creating an audit trail in a tag attached to, say, a car, to indicate where it was manufactured and to record each time it was sold, its previous owners, its service history and its accidents.

Based on the growing number of business sectors that are beginning to test tagand- reader systems, some experts in the field believe RFID will be widely used, especially in retail, by 2010. Others say such broad application will not happen until around 2015 or later, when the cost of RFID tags falls enough to make them economically viable for labeling inexpensive consumer products.

The Near Future
RFID TRACKING technology is starting to be used to follow merchandise as it travels from factory to stores. It will probably be fully established for such applications before it makes deep inroads into stores proper, because warehouse systems are easier to develop and are less likely to fuel public concern that RFID tags in consumer goods could be used to monitor customers once they leave a store. Recently Wal-Mart announced that it will require its top 100 suppliers to place highfrequency tags on cartons and pallets shipped to its stores. And the U.S. Department of Defense has similarly called on its suppliers to adopt high-frequency RFID inventory labeling by 2005.

But the potential—and inevitable— uses for RFID in stores themselves remain tantalizing for retailers. The canceled Wal-Mart in-store test, planned in partnership with Gillette, would have evaluated the ability of RFID-based “smart shelves,” equipped with built-in readers, to monitor the movement of millions of shavers and other Gillette products embedded with RFID tags. (In principle, the 96-bit code allotted for identification of each RFID tag would allow every person on earth to have about 50 quadrillion tags apiece.) The ability to keep tabs on individual products on store shelves is generally accepted as the most difficult task for RFID technology—but one that could pay off royally for retailers.

Notably, RFID smart-shelf systems could save money on labor and help to increase sales by ensuring that shelves are always stocked. If the systems monitored stock levels, employees would not have to do it: when the computers sensed that stock was running low, they could automatically alert someone to order more or could place orders directly with the manufacturer. The systems could offer other benefits as well. Because inventory tags are programmable, their data can include information about where the item was manufactured and sold. And like pinnedon magnetic antishoplifting tags, the RFID inventory tags could be detected leaving the store to prevent theft (estimated to cost $50 billion a year).

Wal-Mart said it canceled its in-store test to free up resources for developing behind- the-scenes RFID capabilities in its warehouses, which will require fewer tags and less powerful computing. This is probably true; industry insiders, however, have suggested that consumer concerns over RFID systems invading individual privacy also played a significant role in the decision. That the backlash had an influence would not be surprising, given that it was at about the same time that Benetton aborted its own large-scale in-store test of an inventory system after its plans were criticized by consumers and the media. The Benetton trial would have examined RFID technology’s ability to scan entire cases of tag-bearing clothes in many different colors, sizes and styles and to capture and upload the inventory data to its tracking system, obviating the need for workers to hand-check each garment.

Other tests of warehouse and in-store inventory systems continue, by Procter & Gamble, Canon, and International Paper. And last spring, Metro, a German retail chain, opened a “future store” equipped with an RFID inventory management system involving both smart shelves and scales equipped with RFID readers that can identify types of produce. In addition, tagged shopping carts are scanned to measure in-store customer traffic and to signal automatically for the opening or closing of checkout stations. The Metro pilot is the work of Intel, where I work today, and the German software developer SAP, along with more than 30 other companies, including Hewlett-Packard, Cisco Systems and Philips.

Over the Horizon
RFID INVENTORY systems still fall far short of Weiser’s vision: they do not help us perform everyday tasks. Indeed, computers and chips scattered throughout our homes—in toasters, games, entertainment systems and other devicess—demand more, not less, of our attention. We must configure and control dozens of devices, transfer data between them, and try to figure out what went wrong when a failure occurs. Simple tasks, such as setting a wristwatch or operating a television, require an instruction manual. It is clear that for computing to become invisible, we need not only ubiquitous computing but what David L. Tennenhouse of Intel calls “proactive computing”— systems that anticipate what we need and provide it without forcing us to do a lot of work first.

For proactive computing to function on a major scale, networks of RFID readers must be placed throughout the environment. Forward thinkers envision two main types of proactive RFID networks, both of which include a web of interacting readers that monitor many RFID tags and convey the information they collect to remote computers.

One type is made up of readers set permanently in place and connected together by cables. These devices would power and read tags—some with sensors— that are also permanently fixed in place. (If necessary, the tags could also be read by mobile readers passing by.)

This kind of network might be installed on a bridge: tags would be buried deep inside concrete structural members, welded into joints between steel beams and put in other places where their sensors could measure stress and change in various parts of the structure. They would collect and store such information as the discovery that a structural member had been flexed beyond its safe limits during a seismic tremor. The readers would be powered from ordinary AC electric lines or through the interreader network cables and would be hardwired to an Internet connection, so they could send their data to computers that would analyze the input and take action in response.

The second type of system—called an ad hoc wireless network—does not have all its readers and sensor tags permanently in place. Instead it is made up of RFID readers put wherever they are needed, in the same way you would choose a spot to plug in a lamp. They read tags that surround them: some of the tags are fixed and stationary; some have sensors and some do not; and some are mobile, attached to devices and people that pass through the network. Readers may be AC-powered if they are near power outlets or may be battery-powered. These readers, also known as network nodes, can form short-range wireless connections to one another on the fly: information moves across the network by hopping wirelessly from node to node (which is why these are sometimes called multihop networks) and flows toward a gateway node with an Internet connection.

You might create an ad hoc network with many readers monitoring hundreds of tag sensors spread out across tens of square miles. Such a network could provide the data to make improved weather forecasts. If the sensors could simultaneously detect wind speeds at many locations across the whole area, the computer could even sense the formation of a tornado at an early stage and generate an earlier alert than is possible today.

An ad hoc RFID network in an office building could perform many tasks. Readers could monitor sensor tags that indicate the temperature in different rooms so that the central computer could maintain constant conditions throughout the building or on a single floor. Other readers would scan employees’ security badges and recognize the tags in their laptops so that workers could access centrally stored data or link up with colleagues elsewhere in the building. The design of all kinds of sensor networks is being researched by Deborah Estrin’s team at the Center for Embedded Networked Sensing at the University of California at Los Angeles, by David E. Culler’s team at the University of California at Berkeley, by Gaetano Borriello’s team at the University of Washington, at Intel Research’s Network of Labs, and at several small companies, including Crossbow in Santa Clara, Calif., Dust Inc. in Berkeley, Calif., and Sensoria in San Diego.

The Responsive Environment
WHEN RFID NETWORKS are finally in place everywhere and we are surrounded by tags and readers feeding responsive computer systems, we will have reached the point at which Weiser believed computing could be blended invisibly into everyday tasks. At this level of integration, RFID technology will support even our simplest activities. For example, RFID-enhanced computer products could “talk” to one another and independently configure their connections. My Intel colleague Trevor Pering has been exploring a way to automatically configure wireless network links between mobile computers and peripherals. If you purchased a printer with a wireless networking capability (such as Bluetooth) and an RFID tag, you might simply unpack the device and bring it near your computer: the computer would read the printer’s RFID tag and connect to the printer automatically, eliminating messy configuration dialogues.

The scope of possible RFID applications is vast and could even include assisting people with Alzheimer’s disease. Eric Dishman, also at Intel, is working on a system aimed at helping those with memory impairment maintain their independence. In one prototype system, all the objects needed for making a cup of tea are tagged. If the patient picks up at least two objects—say, a sugar jar and a tea bag— the system infers, by knowing the ID and location of the objects in relation to one another, that the patient needs help. The system also tracks the sequence in which the objects are used in order to infer whether the person is “stuck” and then delivers recorded voice assistance.

In a totally different realm, PSA Corporation, Hutchinson-Whampoa and P&O Ports—the three largest seaport operators in the world—have taken what could be the early steps toward developing an RFID-based antiterrorism security system that would outfit cargo containers with hidden sensor tags designed to detect radiation or chemical or biological agents in smuggled weapons. Right now the system can detect only whether a container has been opened by an unauthorized person during transit. It could be expanded so that at each stage of a container’s journey, from its initial site to ground transportation, dockside storage and transport ships, readers would interrogate the tag to determine if it had detected dangerous materials. The tag’s sensor would permanently register even very brief exposures to these substances and flag the incident at the next reading station.

Eventually PDAs (personal digital assistants) could be designed to operate as RFID tag readers so that we could receive proactive assistance from tags placed virtually everywhere in our environment. From a tagged sign on a train station, your PDA could retrieve a Web address linking you to an Internet-based timetable. Similarly, realtors could tag the signs on homes for sale: driving past, you could simply beam your PDA at the realtor’s sign and then download photographs and information about the property from the Internet.

Important technical challenges remain, and so it will be years, perhaps decades, before we can reap the benefits of such fully realized RFID applications. As RFID reader-and-tag networks begin appearing in our environment, however, we will increasingly see how this technology can extend the ability of computers—in combination with the Internet—to sense and respond to the physical world.

In his 1991 article in this magazine, Weiser wrote: “There is more information available at our fingertips during a walk in the woods than in any computer system, yet people find a walk among trees relaxing and computers frustrating. Machines that fit the human environment, instead of forcing humans to enter theirs, will make using a computer as refreshing as taking a walk in the woods.” Wielded sensibly, RFID has the power to make computing an unobtrusive, intuitive part of everyday life—indeed, as refreshing as a walk through nature.

Dealing with the Darker Side
WHAT WILL BE the social consequences of a world full of embedded RFID tags and readers? Will our privacy be further eroded as RFID technology makes it possible for our movements to be tracked and allows our personal information to be available in unprecedented detail? These and many other questions must be answered before RFID systems become commonplace.

One of the major worries for privacy advocates is that RFID tags identifying individual items purchased with credit or debit cards would link buyers to the specific items in the card’s or the store’s databases. Marketers could then use these data to keep track of exactly what particular people bought, down to the color, size, style and price—more information than UPC bar codes reveal. In an amplification of the way that phone and direct-mail solicitors use similar, less accurate data to target people for sales pitches, those equipped with RFID-derived data might home in on consumers with very specific sales pitches.

Another concern is that RFID equipment will produce automatic audit trails of commercial transactions: in a totally tagged world, it will be easier to detect when we lie about how we spent our time or what we did and where. This capability could have great consequences for the workplace, and the legal system might look to using logs kept by tag readers as courtroom evidence. We may need laws to specify who can access data logs and for what purpose. In Europe, the Data Protection Act already limits access to computer records of this kind, and the U.S. will probably enact similar legislation.

We will also have to grapple with the inevitable displacement of workers by RFID systems. Opposition to tagging could well come from the industrial labor force, which stands to lose significant numbers of jobs as industry adopts RFID tools able to perform tasks that now depend on human effort. A bitter strike by longshoremen on the West Coast in 2002, partly over new technology that threatened future jobs, may have been a preview of conflicts to come over RFID systems.

Privacy Advocates Protest
The backlash against perceived invasions of consumer privacy by RFID applications began in March 2003, when Philips Semiconductor announced that it was shipping 15 million RFID tags to the clothing manufacturer and retailer Benetton to be incorporated into labels during production. The tags were to interact with a network of RFID readers in Benetton’s store shelves and warehouses to track inventory throughout the company’s 5,000 retail outlets worldwide.

Despite Philips’s reassurances that tagged clothing could not be tracked outside Benetton stores, some industry experts said that criminals could increase the Benetton tags’ tracking distance by creating more sensitive RFID readers. Privacy advocates worried that the tags could be scanned by RFID readers other than those in Benetton stores, which would allow people wearing the clothes to be monitored without their knowledge by, say, criminals or the government. Consumers Against Supermarket Privacy Invasion and Numbering, a U.S.- based privacy group, called for a worldwide boycott of Benetton until the company abandoned RFID tracking technology. Benetton quickly issued statements saying that although it had already tested RFID systems, it was not using RFID inventory tracking and had no firm plans to insert the millions of Philips tags into its products.

Similar concerns—that corporations might keep consumers’ products under surveillance in purchasers’ homes and on the streets—surfaced about a test of an in-store RFID inventory system that was planned by Wal-Mart and Gillette. To answer consumer concern, Gillette announced that it was embedding its RFID tags in packaging, not products, so purchasers would discard the tags with the packaging. But Declan McCullagh, a commentator who writes for computing publications and who favors RFID for its practical value, has written: “Future burglars could canvass alleys with RFID detectors, looking for RFID tags on discarded packaging that indicates expensive electronic gear is nearby.... [T]he ability to remain anonymous is eroded.”

One way to avoid such possibilities is to put a kill switch into each RFID tag on a consumer item, which would allow the tag to be turned off after purchase. Indeed, the Auto-ID Center—a research consortium funded by information technology companies and headquartered at the Massachusetts Institute of Technology—has released guidelines saying that retailers must be able to disable RFID tags at checkout counters, and manufacturers, including Alien Technology, Matrics and Philips, are now producing tags with kill switches.

McCullagh has suggested four requirements for the use of RFID tags on consumer products: Consumers should be notified when RFID tags are present in what they are buying (this could be done with a printed notice on a checkout receipt). All tags should be readily visible and easily removable. The tags should be disabled by default at the checkout counter. And, when possible, RFID tags should be placed only on the product’s packaging, not embedded in the product.