Some technologists believe that in the future, seemingly invisible computers will be embedded everywhere, collecting data about the environment and making it useful to decision makers. One way to achieve this sort of ubiquitous computing is to disperse tiny sensors that measure, for instance, light, temperature, or motion.
But without a persistent power source, such sensors would need their batteries replaced every few months. In other words, ubiquitous sensors could also mean “ubiquitous dead batteries,” says Josh Smith, a researcher at Intel Research in Seattle.
Smith and his team are addressing this problem not by working on longer-lasting batteries but by trying to eliminate the need for batteries altogether. Instead, their prototype devices employ the same power-scavenging technique used by battery-free radio frequency identification (RFID) tags.
The concept of throwing out the sensor battery is not new. Researchers have proposed capturing energy from environmental vibrations or ambient light to power a sensor (see “Free Electricity from Nano Generators”). But it is unclear whether technology that captures ambient energy can be inexpensively integrated into a sensing device.
By contrast, the technology used in RFID tags, which transmit a few bits of information when scanned by an RFID reader, is cheap enough to integrate into sensors and be mass produced; they’re already widely used to track livestock and cargo, as well as cars passing through “easy pass” lanes on highways.
Smith explains that Intel’s sensor devices use off-the-shelf components: an antenna to send and receive data and collect energy from a reader, and a sensor-containing microcontroller – a tiny computer that requires only a couple hundred microwatts of power to collect and process data.
The antenna harvests this power directly from the radio waves emitted by an RFID reader. When a tag comes within range of a reader, the reader’s radio signal passes through the antenna, generating a voltage that activates the tag. The tag is then able to send information to the reader through a process called backscattering, in which the antenna essentially reflects a data-encoded variation of the received radio signal.
The microcontroller that Smith’s team added to the RFID antenna includes a 16-bit microprocessor, 8 kilobytes of flash storage, and 256 bytes of random-access memory.
One of the microcontroller’s main jobs is to ensure that information is transmitted to the reader error-free, which requires more computation than a conventional RFID tag can handle. In a typical tag, the error-checking information is precomputed and stored on the chip; but for a sensor, Smith says, this information needs to be computed in real-time as data is gathered.