With cellphones hanging off shoulder bag straps, pagers hooked to our belts, digital cameras dangling from our necks, PDAs bulging in our pockets, and MP3 players clipped to our shirts, we're all beginning to look like electrogadget pack mules.
We have a more versatile and, we dare say, elegant alternative: e-textiles. Your shirt, coat, or sweater, even your carpeting or wallpaper, is the device. Conductive fibers woven into the fabric using standard textile techniques carry power to sensors, actuators, and microcontrollers embedded in the cloth. Software controls the communications inside the on-fabric network and can send radio signals using Bluetooth or any flavor of the IEEE 802.11 wireless standard to PCs and PDAs, and over the Internet.
Applications are astoundingly diverse. An Army commander, for example, could monitor a platoon of soldiers clad in SmartShirt gear developed by two of us (Jayaraman and Park) at the Georgia Institute of Technology in Atlanta. The shirt communicates vital signs in real-time, and when all hell breaks loose on the battlefield, the commander sees at a glance who's been hit and who hasn't--and who is gravely injured and in need of immediate attention.
Closer to home, a fire chief could keep tabs on a unit as it enters a burning building. He could order his team out when the sensors they're wearing transmit data back to his command center telling him that the firefighters are inhaling hazardous fumes or too much smoke or that the fire is too hot to handle.
Imagine the boon to athletes. A swimmer stroking through the water, vital signs monitored by electrodes attached to wires hanging off her body like the tentacles of a jellyfish, would welcome a sleek, instrumented training suit. And five-time Tour de France winner Lance Armstrong, who lost an estimated 6.5 kg during the first individual time trial of this year's Tour, could have used a racing suit dotted with moisture, temperature, and pulse sensors. Such attire could have warned the U.S. Postal Service team manager that Armstrong was becoming dehydrated as he was warming up. In turn, the manager could have ordered Lance to drink replacement fluids before he launched from the starting line on his way to a rare time-trial defeat.
Similar performance- and safety-enhancing garb has already been prototyped by Finnish researchers at Tampere University of Technology and the University of Lapland, and at outerwear maker Reima Oy in Kankaanpää, Finland. They developed a machine-washable jacket, vest, trousers, and two-piece underwear set for snowmobilers. The jacket is embedded with a GSM (Global System for Mobile Communications) chip; sensors monitoring position, motion, and temperature; an electric conductivity sensor; and two accelerometers to sense impact. If a crash occurs, the jacket automatically detects it and sends a distress message to emergency medical officials via Short Message Service. The message conveys the rider's coordinates, local environmental conditions, and data taken from a heart monitor embedded in the undershirt.
O.K., you don't plan to join the Army, rush into a towering inferno, or compete in the Tour de France. You have no interest whatsoever in swimming and snowmobiling. Nevertheless, e-textiles are soon going to add functionality, fun, and style to whatever it is that you do like to do.
Just last May, German chipmaker Infineon Technologies AG, in Munich, and its partner, Vorwerk & Co. Teppichwerke GmbH & Co., in Hameln, unveiled a carpet that can detect motion--of unwanted intruders, for example--and also light the way to exits in the event of a fire. The carpet is woven with conductive fibers and studded with pressure, temperature, or vibration sensor chips, microcontrollers, and light-emitting diodes (LEDs) [see illustration].
Last year France Télécom showed off a display made of woven optical fibers that can be worked in with standard textiles. A T-shirt or backpack could display text and images, including video and advertising logos, and could be adapted for color-changing scarves and furnishings.
And for those of us who can't stand looking at the same décor day in and day out, International Fashion Machines, cofounded by Massachusetts Institute of Technology alumna Maggie Orth, is commercializing Electric Plaid wallpaper. And when she says electric, she means electric: a swatch now on display at the Cooper-Hewitt National Design Museum's National Design Triennial in New York City slowly changes colors and patterns as conductive fibers heat and then cool threads coated in thermally sensitive inks.
These prototypes are a small sample of the vast variety of fibers and fabrics that can be woven into clothing, carpets, upholstery, and wallcoverings. Coupled with fault-tolerant computing and network architectures, such e-textiles can constitute a platform for health monitoring, communications, multimedia devices, and changing décors.
Mother of all wearable motherboards
Some of these garments will be on the rack or on your local firefighter in the next five years. Infineon's carpet and International Fashion Machines' wallpaper should hit stores within the next couple of years, and perhaps a SmartShirt for infants will, too.
Sudden infant death syndrome, or SIDS, extinguishes the lives of thousands of sleeping infants every year. An e-textile shirt from New York City-based Sensatex Inc. promises to put an end to SIDS by alerting parents the moment a baby stops breathing [see photo]. With sensors that monitor heart and respiration rates and body temperature, the shirt will communicate wirelessly with a parent's PDA, watch, or PC.
The SmartShirt is the generic name journalists came up with in the late 1990s to describe the Wearable Motherboard, which Georgia Tech licensed to Sensatex [see photo]. In the computer world, motherboards are circuit boards that let you easily plug in chips and processors for specific applications, like graphics or wireless communications.
Similarly, the Wearable Motherboard provides that kind of versatility for clothes. Manufacturers can mix and match sensors, processors, and communications devices that plug into knitted or woven garments made from cotton, polyester, or blends. The garments are threaded with conductive polymer and metallic fibers that serve as data buses and power lines. These devices have the look and feel of typical garments and, after the attachments are unplugged, can be tossed into the washing machine.
The Wearable Motherboard was funded initially by the U.S. Navy in 1996 as a garment to detect bullet wounds in combat. It is woven from a cotton and polyester blend as well as optical fibers that when severed will indicate exactly where a bullet has passed through. It also monitors the condition of the wearer with sensors to measure vital signs. In practice, the user will position the dime-sized sensors on his body and plug the leads into tiny connectors, which measure 5 mm in diameter and look like snaps on blouses. Called T-Connectors, the sensors are placed in the appropriate spots--over the heart or diaphragm, say. Depending on the application, the garment could have dozens or hundreds of connectors.
A flexible data bus integrated into the fabric routes data from the sensors to the SmartShirt controller, which uses a proprietary chip set, in a plastic package the size of a pager. Powered by a watch battery, the controller presses onto the fabric like a fastener to contact the conductive fibers. It processes the signals from the sensors to compute vital signs such as heart rate and wirelessly transmits the data directly to a PDA or PC using Bluetooth or IEEE 802.11b. Or the vital signs data pops up on a display in the parents' bedroom, in the case of the SIDS shirt. But for other applications, the monitor might very well be at the doctor's office, in a hospital, or on the sidelines of a football field.
Reliability through redundancy
Weaving together a complex e-textile system challenges clothing designers and systems mavens alike. How should the sensors, processors, and controller fit together? What kind of software can we write to ensure fault tolerance and quality of service both within the garment and with external devices? Can we come up with a hierarchical design process akin to those used in the IC industry?
Depending on the application and the physical area it covers, an electronic textile relies on dozens to hundreds of sensors and processing elements, each with limited processing, storage, and power consumption: up to 100-MHz processing speed, 64 KB or less of local memory, consuming up to a few tens of milliwatts. Clearly, these devices are not ”desktops on a fabric”; nor are they cellphones, PDAs, or set-top boxes, in the sense that e-textiles must have low manufacturing costs. As a result, the devices will inevitably have more processors and interconnects that don't work than do other embedded systems.
And unlike, say, a set-top box, clothes get worn, washed, and torn. Carpets get vacuumed, shampooed, and trod upon. That means e-textiles must be designed based on a fault-tolerant system that can cope with wear and tear. Thanks to Moore's Law, chips are cheap and will get cheaper, which makes redundancy the cost-effective key to reliability. Networking hundreds of processors, sensors, and controllers lets the garment automatically redistribute the workload around failed processors or improve the quality of service if it drops below a certain threshold. For example, many applications tolerate gracefully degrading quality when some percentage of the sensors or processing nodes fail.
Take the swath of acoustic beam-forming fabric developed by researchers at Virginia Polytechnic Institute and State University, in Blacksburg, where an array of microphones woven into the material monitors the environment for audio signals that indicate the position of a tank. When the fabric's power supply runs down, techniques developed by Carnegie Mellon University's (CMU's) Coatnet research group (Marculescu and Marculescu) may come to the rescue.
In the Pittsburgh-based Carnegie Mellon system, when one node starts to fail because of local battery depletion, a redundant node and accompanying battery take over computation and complete the task. The handoff happens transparently to the overall operation, staged carefully to avoid collisions on the communications bus when too many nodes rush to ship their computation to spares. With half the nodes redundant, the lifetime of the system can be extended by 80 percent, with no decrease in overall system quality.
In the last few months, the Coatnet group has developed several prototypes targeted at safety and security applications. One prototype fabric uses processing nodes and embedded temperature sensors interconnected in an array to monitor, for example, the internal and external temperature of firefighters' suits. In the current prototype, one master node for data collection and eight additional slave nodes for sensing are interconnected in a grid to measure temperature over an area.
All nodes run on simple Texas Instruments MSP430 microcontrollers, each running at up to 8 MHz, with a mere 64 KB of local memory and only 1 mW of power consumption. Eight additional nodes can be used as spares for the sensors to prolong application lifetime. The initial wearable version of the prototype works just as well for wallpaper that in the case of a fire ”knows” which locations are too hot for firefighters to enter.
Eventually, buildings could also be augmented with camera arrays inconspicuously embedded into wallpaper fabric to scan for intruders. In CMU's prototype, every sensing node uses small cameras and Atmel 8051 processors, running at 70 MHz and consuming up to 500 mW each, to analyze images for possible security breaches and then stream the video to a central display. Redundant devices keep the system running in case of local battery depletion or other types of failures.
But while various projects have proved the feasibility of individual designs, none points to an overall methodology for the evaluation and validation of e-textile systems, something we need if a real industry is to grow around the technology. Today, when an engineer designs an ordinary IC, he maps the application onto a given platform architecture, under specified constraints--performance, area, and power consumption. Constraints met, testers put the prototype through its paces. Only after the device passes muster does full-scale manufacturing begin.
But such a design methodology won't work with wide-area textile networks like those in carpeting. With e-textiles, partitioning the fabric holds the key to reliability and repeatability--in other words, manufacturability. By partitioning the application into small chunks computed locally, we can minimize communications among processing nodes. In doing that, we can decrease the possibility of congesting connecting links or losing data packets among communicating nodes or sensors.
Designed properly, preprogrammed processing nodes can be reprogrammed when operating conditions change. For example, such a truly smart fabric could route data packets or control signals around a hole in a wounded soldier's uniform or a wet area of a baby's outfit.
Will the fashionistas bite?
In the beginning, there won't be a single killer application to make this market flourish. Rather, niche applications like SIDS monitoring, military uses, or athletic training will get the ball rolling. Going from there to a wider consumer market will take more than a few teenagers' wanting to play MP3 tunes straight from their jackets. Perhaps multifunctional suits that would not only count calories but also change color could spawn a larger market, or garments that authenticate you automatically and let you move easily from one secure environment to another.
In some cases, privacy as well as safety issues will play a crucial role in swaying consumers' decisions. While these concerns apply to other consumer electronics, too, that hasn't stopped cellphones and PDAs from becoming the must-haves of modern life. And the convenience factor could well tip the scales in favor of a ready-to-ware revolution. After all, only in weird dreams do we leave home naked, but fully conscious we often step out sans cellphone.
The authors wish to acknowledge the contributions of Pradeep K. Khosla of Carnegie Mellon University and IEEE Spectrum senior associate editor Harry Goldstein.