CeNSE is the boldest project yet announced, but HP is not the only organization developing the technology to make ubiquitous sensing possible. Intel is also designing novel sensor packages, as are numerous university labs.
For all the momentum in the field, though, this sensor-filled future is by no means inevitable. These devices will need to generate rich, reliable data and be rugged enough to survive tough environments. The sensor packages themselves will be small, but the computing effort required will be enormous. All the information they gather will have to be transmitted, hosted on server farms, and analyzed. Finally, someone is going to have to pay for it all. “There is the fundamental question of economics,” notes computer scientist Deborah Estrin of the University of California, Los Angeles. “Every sensor is a nonzero cost. There is maintenance, power, keeping them calibrated. You don’t just strew them around.”
In fact, HP senior researcher Peter Hartwell acknowledges that for CeNSE to hit its goals, the sensors will need to be nearly free. That is one of the reasons why HP is designing a single, do-everything, pushpin-size package stacked with a variety of gauges—light, temperature, humidity, vibration and strain, among others—instead of a series of devices for different tasks. Hartwell says that focusing on one versatile device will drive up volume, reducing the cost for each unit, but it could also allow HP to serve several clients at once with the same sensors.
Consider his chief engineering project, an ultrasensitive accelerometer. Housed inside a chip, the sensor tracks the motion of a tiny, internal movable platform relative to the rest of the chip. It can measure changes in acceleration 1,000 times as accurately as the technology in the Nintendo Wii.
Hartwell imagines situating one of these pins every 16 feet along a highway. Thanks to the temperature, humidity and light sensors, the motes could serve as mini weather stations. But the accelerometers’ vibration data could also be analyzed to determine traffic conditions—roughly how many cars are moving past and how quickly. The local highway department would be interested in this information, he guesses, but there are potential consumer applications, too. “Your wireless company might want to take that information and tell you how to get to the airport the fastest,” Hartwell says.
All of this gathering and transmission of data requires power, of course, and to guarantee an extended life, the HP pushpin will not rely solely on batteries. “It is going to have some sort of energy-scavenging ability,” Hartwell says. “Maybe a solar panel or a thermoelectric device to help keep the battery charged.”
With the power hurdle in mind, other groups are forgoing batteries altogether. At Intel Labs in Seattle, engineer Josh Smith has developed a sensor package that runs on wireless power. Like the HP pushpin, Intel’s WISP, or Wireless Identification and Sensing Platform, will include a variety of gauges, but it will also draw energy from the radio waves emitted by long-range radio-frequency ID chip readers. Smith says a single reader, plugged into a wall outlet, can already power and communicate with a network of prototype WISPs five to 10 feet away—a distance that should increase.
Smith cites many of the same infrastructure-related possibilities as Hartwell, along with a number of other uses. If WISPs were placed on standard household items such as cups, these tags could inform doctors about the rehabilitation progress of stroke victims. If the cups the patient normally uses remain stationary, Smith explains, then the individual probably is not up and moving around.
The potential applications for ubiquitous sensing are so broad—a physicist recently contacted him about using WISPs to monitor the temperature outside a proposed neutrino detector—that, as with the Internet, Smith says it is impossible to foresee them all. “In terms of the impact it is going to have on our lives,” Hartwell adds, “you haven’t seen anything yet.”