 |
They will watch you. They will know you. They will be all but invisible to you. By 2010 there will be 10,000 connected microsensors for every person on the planet, an Ernst & Young report predicted in 1999.
The broad idea of “pervasive” computing — a galaxy of devices connecting seamlessly in a world of whenever, wherever networks — was spawned more than a decade ago.
So what should we call today’s emerging swarm of small tech-enabled networks, devices, sensors and software beginning to weave itself into the fabric of human experience?
 |
Think of it as the Mesh, a new layer of technology interlaced with every thread of human life and commerce. A Mesh of pervasive small technology is the next logical stage in a century of disruptive, global changes — from commercialization of electricity to television and telecommunications, to today’s explosion of computing devices and digital networks.
“Pervasive computing isn’t about more microprocessors everywhere,” said Larry Goldstein, vice president of research and engineering at Graviton Inc., a three-year-old San Diego startup that aims to commercialize wirelessly networked MEMS sensors for businesses.
“It’s about the convergence of sensing, computing and digital communications,” he said. “We’re building nervous systems for the engineered world.”
Goldstein calls it “intelligent telesensing.” For the business world, that could mean environmental monitoring in and around power plants and factories, or sophisticated security perimeters for government or military sites that detect everything from a chemical weapon to a terrorist’s face.
As Goldstein explains, the “pervasiveness” of such digital sensory networks is fundamental to their value. “Lots of nodes gets you a better picture,” he said. “But to drive down the cost per point, you need things like low-cost, low-power MEMS and inexpensive wireless.”
Douglas Heintzman, manager of strategy and standards with IBM’s Pervasive Computing Group, sees small technology driving the trend in two key areas: tiny sensors that tell a device where it is; and compact, high-density data storage.
Heintzman said that a device aware of its location and context would have many practical advantages. A mobile phone in a movie theater could automatically silence its ring. A wireless device entering an airport could check in for a flight electronically. “Location sensing also simplifies the design of interfaces for devices that may have a small screen and no keyboard by making them aware of their context,” he said. IBM is already working on it, Heintzman said, developing small and inexpensive location sensor chips that will tell mobile devices where they are through Global Positioning System (GPS) satellites or cellular towers.
Finally, technology is reversing the way people and computing devices interact. Until now, humans have largely had to adapt to computers, learning commands and deciphering graphical menus and windows. Pervasive computing will turn that model around by helping computing systems better adapt to people.
Pervasive devices and environments will understand or anticipate what a user wants through speech recognition, or by tracking eye movements and gestures. Researchers at the Rutgers Center for Advanced Information Processing are investigating more intimate interactions between people and machines that blend speech recognition and synthesis with gaze or gesture tracking, microphones and microsensor-embedded gloves.
Pervasive systems are not in the laboratory only. They’ve already made their way into the commercial market. Graviton recently struck a deal with Wenner Bread Products Inc. in Bayport, N.Y., to set up a wireless network that will monitor and control the temperature of its refrigerators, track the performance of electric motors in various machines and monitor power consumption.
Solomon Trujillo, the former chief executive of U.S. West who is now Graviton’s chairman and CEO, noted in a speech last year that the global market for sensors, wireless sensor communications and sensor data services will grow to more than $300 billion by 2003.
In Bayport, about 100 off-the-shelf sensors in machines throughout the bread company’s plant are already operating. Designed to help Wenner control quality and optimize plant operations, the sensors beam their data to a gateway computer over a wireless network.
The system’s software is designed not only to track operations in real time, but also to store and analyze data — like the electrical signatures of various motors — to better predict when a machine might have a problem.
Graviton isn’t using small tech sensors in the system for Wenner Bread, but it is developing its own low-cost MEMS sensors to detect hydrogen or volatile organic compounds such as solvents. One application could be a gas gauge or leak detector in fuel cells for cars or smaller devices. The IP is based on patents licensed from Oak Ridge National Laboratory for monitors used in nuclear weapons production.
Of course, industrial automation isn’t new, but wireless MEMS technology will make it less expensive and more powerful, pushing it deeper throughout commercial environments and opening telesensing to a wider range of applications. Graviton hopes to make MEMS sensor networks affordable for home automation applications such as remote health monitoring or temperature controls.
David Tennenhouse, director of Intel Research and leader of the company’s “proactive computing” initiative, thinks that commercial pioneers like Graviton will create a “virtuous cycle” of demand for MEMS sensors and networks. “Sensors can generate huge amounts of high-fidelity data that enterprises will process with increasing computing power,” he said. “As companies begin to wring productivity gains from smart sensors tied to data-mining databases through XML, they will create the market pull for smaller, cheaper and better MEMS-based sensors, as well as the software and hardware networks connecting them.”
He believes intelligent sensor systems could deliver gains in productivity. “Sensing systems will help companies operate as true real-time businesses, so that inventory or sales can be gauged not just day to day, but perhaps hour by hour or minute by minute.”
Talking Tags
Today, pervasive small tech is particularly evident in the kind of wireless identity tags or smart labels known as RFID (radio frequency identity) that big companies such as Texas Instruments Inc. are making by the millions. Upstarts like Alien Technology in Morgan Hill, Calif., want to make them even tinier and cheaper with small technology.
Such “Meshy” microsensing has been at work for several years in systems such as ExxonMobil’s SpeedPass device. Based on a TI radio chip, the device allows more than 6 million participants to pay at the pump with a wave of a wand. Toll payment systems such as E-ZPass in the New York metro area are another high-profile implementation of large-scale RFID.
Such wireless tags could be a great boon to retail stores, warehouses and distribution networks. The Gap recently tested an electronic tagging system to track jeans from one of its distribution centers to a store outside Atlanta. An electronic reading device in the store was able to identify and account for all of the tagged jeans at a rate of 50 tags a second.
Compared to a conventional Universal Product Code (UPC) stripe, RFID tags can hold much more data about the objects they’re attached to, such as when, where and how they were made. When such “smart labels” can be produced for pennies a piece, they promise to enable faster, more automated inventory control and product tracking, better shoplift prevention and, eventually, smoother transaction processing.
Checking out of a supermarket with a shopping cart filled with RFID-tagged products will be fast and easy: Simply push the cart past a reader and all your purchases will be itemized and totaled in a second or two.
The physical size and cost of these types of communication sensors could be shrunk significantly by small technology such as Alien Technology’s Fluid Self Assembly (FSA) process. The company is also using FSA to produce simple, flexible LCD display screens for smart cards that could display your balance or loyalty points.
Intelligent Specks
Kris Pister, an associate professor of electrical engineering at the University of California, Berkeley, has been developing an even more sophisticated breed of networkable microsensors. His MEMS-based “Smart Dust” — about 10 cubic millimeters in size now, and perhaps as small as grains of sand in the future — offer a wide range of potential uses in military, intelligence and civilian applications.
Pister’s sensitive “motes” would work in dense arrays that might, for instance, be dropped in a battlefield to identify and track enemy troops via infrared, acoustic or other measurements. They could collect weather data, track the movements of animals in remote places or help buildings more intelligently control their energy use. Pister reports that the next challenge is to integrate his current line-of-sight connected motes with wireless technology.
Canesta Inc. in San Jose, Calif., is developing “electronic perception” technology — low-cost, small tech chips that would give computers and mobile devices a kind of 3-D radar-vision. Based on precise timing of infrared signals, the company’s inexpensive “machine sight” could secure a laptop with facial recognition or enable a PDA or mobile phone user to “type” onto a virtual keyboard simply through subtle head movements.
Centeye Inc., an early-stage startup sponsored by the Defense Advanced Research Projects Agency (DARPA), is also working on a low-cost vision chip. Founder Geoffrey Barrows, who left the Naval Research Laboratory to start the company in Washington, D.C., says that his “biomemetic” vision draws on new understanding of how insects perceive depth. The electronic eyes he wants to make combine a photo-receptor and image processor on a small piece of silicon that weighs grams and runs on milliwatts of electricity. The company’s name reflects the goal of making small (centimeter) inexpensive (cents) and numerous (cent = hundred) seeing chips for tiny unmanned flying drones smaller than the U.S. military’s Predator aircraft used in Afghanistan.
In August, 2001 Intel Research announced a new initiative for MEMS-based “proactive computing.” Intel has funded satellite research centers at Carnegie Mellon University in Pittsburgh, UC Berkeley and the University of Washington to develop self-aware networks of miniaturized sensing computers.
Intel Research director Tennenhouse said the company has a sweeping commercial interest in how the Mesh unfolds. Intel is investigating technology that touches every part of the Mesh, from tiny radios that would connect sensors or motes, to new forms of small storage similar to Flash memory cards, to its processors that might serve as wireless gateways between sensor arrays and the Internet.
In a speech announcing the initiative, Tennenhouse, who joined Intel from DARPA in 1999, said, “We are working toward the point where computers are acting in advance and anticipating our needs.” More recently, Tennenhouse added that “in a world where thousands of computers may be working for each of us, we won’t be able to interface directly with every one of them.” Hence, telesensing systems will need to be autonomous, as well as predictive. For example, a sensing system placed inside a geological fault could study and learn to anticipate earthquakes, perhaps automatically warning other systems in gas pipelines or nuclear power plants in microseconds, faster than any humans could react.
Really Micro Software
David Culler, director of the Intel Research Laboratory at Berkeley, is tackling the significant software challenges in networked sensor systems like Pister’s Smart Dust. The main problem is that tiny devices have very limited processing power, data storage and power supply.
Culler’s group has developed a tiny — 1K — operating system, aptly named TinyOS, that runs applications with 24K of memory and distributes them like a virus, spreading from one sensor to the next as they pass messages back and forth.
Crossbow Technology Inc. in San Jose, Calif., which makes a range of MEMS-based sensors such as gyros, accelerometers and magnetic sensors, is also using TinyOS in its wireless sensor networks. Mike Horton, Crossbow’s president and CEO, said the company tested the technology with the U.S. Air Force Research Laboratory in March 2001 at the Twentynine Palms Marine base in Southern California. An unmanned aircraft dropped about 30 wireless magnetic sensors along a desert road. The sensors contacted each other to form a wireless array that could detect a passing vehicle’s magnetic signature and relay its speed and direction to the unmanned airplane overhead.
Such emerging applications are intriguing, but the typical American home already contains about 40 computing devices. Most of them — electric toothbrushes, cordless phones and microwaves — we don’t usually think of as “computers” at all. With more than 8 billion processors flooding into the world every year, the universality of computing is almost an afterthought today.
Eventually such devices may communicate with their users and each other, as well as become sensing points themselves. Your toothbrush may alert you if it detects a dental problem and signals your home network to schedule an appointment with your dentist.
Not Seen, But Heard
At a conference last year on pervasive computing, the National Institute of Standards and Technology (NIST) defined it as “a strongly emerging trend toward numerous, casually accessible, often invisible computing devices, frequently mobile or embedded in the environment, connected to an increasingly ubiquitous network infrastructure composed of a wired core and wireless edges.”
Two keywords here — “invisible” and “embedded” — point to the integral role small tech is playing in making devices small, cheap and low power.
In the 2002 premiere issue of IEEE Pervasive Computing, Editor-in-Chief M. Satyanarayanan, a Carnegie Mellon computer science professor, characterized the field as “the creation of environments saturated with computing and wireless communication, yet gracefully integrated with human users.”
To that end, small technology is also advancing all facets of computing: enabling faster processors with molecular transistors, smaller and more densely packed storage or memory devices, cheaper and more flexible display technologies, more powerful (yet less power-hungry) always-on mobile devices.
As these hardware factors co-evolve with small tech, people will increasingly interact with information technology in more natural and spontaneous ways.
It’s not much of a stretch to imagine making a video call with your watchphone, or asking it to find the nearest Thai restaurant and order ahead, since it knows exactly what you like.
Devices empowered with small technology will likely learn to recognize the face, voice or fingerprint of their user. Visionics Corp., for example, recently demonstrated a Java-enabled Motorola mobile phone with real time facial recognition capabilities designed to help law enforcement agencies identify suspects.
Reduction in component size and cost via small tech should also help devices become small and flexible enough to wear like a watch, glasses or headset. IBM has been investigating applications for digital jewelry, while Xybernaut Corp. in Fairfield, Va., has been building “wearable” computing devices since 1990.
The Walls Have Eyes
Mesh equipment may even blend into walls, tables, doors and floors to create “smart offices” like those envisioned at IBM and Steelcase’s BlueSpace lab in Hawthorne, N.Y. BlueSpace’s Everywhere Display can project information like an urgent e-mail on partitions, floors or desktops. Wireless sensing technology would allow a user’s finger to act as a mouse cursor.
Digital projectors such as those sold by Texas Instruments, Hewlett-Packard and Kodak today employ various kinds of MEMS and small tech. New companies like Silicon Light Machines are aiming to drive down the cost of digital displays with its Grating Light Valve (GLV) technology that uses tiny reflective ribbons controlled by electrostatic forces. In 2000, the company licensed GLV to Sony.
RF MEMS components and wireless transceivers promise to fuel the growth of broadband wireless networks such as Wi-Fi (a.k.a. 802.11b or Airport) and Bluetooth, a short-range wireless protocol for connecting devices.
In the wake of the cool market reception for Sun Microsystems’ Jini pervasive software in the late ‘90s, pervasive computing became the butt of jokes about toasters talking to vacuum cleaners. But the real punch line is that digital devices and networks already are, and will continue to become, more omnipresent and interconnected. Jini is even back out of its bottle, being investigated for use in Web applications — perhaps to compete with Microsoft’s massive. Net initiative, which also aims to let many different systems interact easily.
Billions of handheld computers, mobile phones, cable set-top boxes, electronic game machines, wireless e-mail gadgets, satellite dishes, medical equipment and automated factory systems are building a new layer of connections atop the cortex of conventional computers and laptops plugged into fiber nets and radio waves. Such an abundance of new connections and communications is also likely to foment a privacy nightmare that will make Web cookies and e-mail pornography seem like quaint inconveniences.
Great Big Mesh
What happens 10 to 20 years from now, when small technology enables universal, low-cost, low-power and compact connectivity to trillions of sensitive nodes? What happens when virtually everything — from every person’s body to every place to every possible object we interact with — becomes part of the network?
And the network becomes part of everything?
At first blush it may sound a little creepy, but the Mesh seems like what the fabric of life is programmed to evolve into in the decades ahead: dataflow and digital sensory receptors interwoven into all folds of human activity and every imaginable physical environment.
For better or worse, pervasive computing catalyzed by small tech is becoming real. On the one hand, it suggests a kind of ever-deepening, symbiotic relationship between human culture and machines. On the other, it presages a time when the machines will largely talk among themselves.
MIT’s Project Oxygen is an interdisciplinary collaboration between the university’s Laboratory for Computer Science and Artificial Intelligence Laboratory focused on a technological evolution. A manifesto on the group’s Web site sounds very much like the Mesh:
“In the future, computation will be freely available everywhere, like batteries and power sockets, or oxygen in the air we breathe.
Computation will enter the human world, handling our goals and needs. We will not need to carry personalized devices around with us. Instead, ‘anonymous’ devices, either handheld or embedded in the environment, will bring computation to us … [and] personalize themselves in our presence by finding whatever information and software we need. We will not need to type or click, nor to learn computer jargon. Instead, we will communicate naturally, using speech, vision, and phrases that describe our intent.”
Like all previous pervasive techno-infections, such as the telephone and television, the Mesh will be fully embedded in human culture when it becomes transparent to us….
When was the last time you thought about how your ATM or credit card transaction is executed? Or how your car determines the precise mix of gas and air to burn? Or how the current powering your computer was generated and distributed?
In some ways, the Mesh is here, and we’re already wrapped up in it.