by Neha K. Choksi, independent consultant
March 2, 2010 - It has been more than a decade since Kris Pister introduced the concept of "smart dust": a distributed wireless network of sensors with self-contained sensing, computation, communication, and power. Although there are a handful of companies in this space (Crossbow Technology, Dust Networks, GainSpan, Arch Rock, for example), distributed wireless sensing networks have yet to hit main stream. At the IEEE Bay Area Nanotechnology meeting on February 16, 2010, Dr. Peter Hartwell revealed new efforts at Hewlett Packard that could change all of that.
Hartwell predicts that sensors will impact human interaction just as the Internet revolution did in the last decade. The impact is just beginning -- free-fall and shock detection to park the hard drive, Wii motion sensors, image stabilization on mobile devices, and tilt monitors in washing machines, to name a few. HP's Central Nervous System for the Earth (CeNSE) is leveraging multiple HP business units to achieve distributed sensing networks. This takes sensing to the next level by allowing a system to incorporate surrounding information to make its own decisions. Hartwell refers to this next phase as "aware computing." The opportunities for this technology are limitless: food safety, disaster prevention, and resource management, to name just a few.
For example, home intrusion systems could sense and distinguish the difference between a human intruder or the movement of a family pet and make smarter alarm trigger decisions. During a Powerpoint presentation, a system can detect noise (speaking) and bypass the screensaver mode that often appears while a speaker is at the podium. With "aware computing" auto shut-off light systems in offices and conference rooms will not need to rely on macro movement for sensor activation. Sensors would be able to detect breathing, noise, and smaller vibrations to determine somebody's presence to keep the lights on. Hartwell refers to these opportunities as "low-hanging fruit" to increase energy efficiency.
HP's CeNSE nodes will include vibration, tilt, navigation, rotation, and sound sensing by leveraging their new revolutionary 6 axis motion sensor. This new accelerometer design has significant advancements -- but HP's solution goes well beyond their novel accelerometer design. In addition to detecting motion, HP is working on chemical and biological sensors by surface-enhanced Raman spectroscopy (SERS) and nanostructures to create miniaturized chemical analysis sensors. By coating nanostructured silicon with silver, HP is able to enhance the signature photon reflection used to identify a sample. This gain factor enables system miniaturization and smaller sample sizes to achieve broad chemical analysis.
One factor limiting the speed of adoption for distributed sensing networks is the cost. Currently available sensor nodes on the market cost in the range of $300-$400 per node. Depending on the application, one million nodes may be necessary, making the current node cost-prohibitive for a fully distributed network. HP's technology will leverage the large-volume manufacturing know-how at their inkjet fabrication site in Corvalis, OR, and technology advancements in the sensor device to build a small low-cost sensor node that is orders-of-magnitude more cost-effective than nodes currently available. As Hartwell puts it, high performance, small, and low cost are the "magic button" and "holly grail" of distributed sensing networks. By making the nodes themselves essentially free, the value will be based on computing the data obtained from these sensors and using this insight to provide useful information to customers.
HP plans are not limited to just the sensor nodes themselves. With the large data that would become available with million node systems, HP must address how to deal with the data. Hartwell envisions the data processing within the network. HP plans to address this by leveraging their memristor technology. This could lead to a fundamental change in computing architecture. Memristors display a non-linear switching characteristic, enabling a teachable platform for data computing.
HP is leveraging multiple units within the company for the CeNSE project. The company's acquisition of EDS provides the communications infrastructure and business process outsourcing needed to make the picture complete. HP's CeNSE initiative aims to provide the total solution to distributed networks: sensor nodes, data computing, and the communications infrastructure. Hartwell asserts that HP will lead the technology revolution with their one-stop shop into what he refers to as "the next wave of the future."
But for HP, this wave is no longer limited to the future. HP has announced its partnership with Shell to acquire extremely high-resolution seismic data on land for more efficient methods of finding and producing petroleum -- and reduce the impact on the environment in the process.
Despite all of the potential applications, one can't help but ask what new dilemmas distributed sensing might pose. "In parallel with the implementation of such a network, would it not be apropos to work on policies that ensure that these networks are used for the common good?" asks John Berg, CTO of American Semiconductor Inc. Nevertheless, the opportunities are abound and HP plans to be poised and ready with a total solution for the eruption of distributed network sensing.