Microsensor nodes that self-organize into a collaborative network and operate autonomously offer a truly disruptive advantage in intelligent observation. The applications for such "mesh networks" are multiple and multifaceted: monitoring vineyards for climate changes, monitoring the structural integrity of buildings, updating factory floor inventory, collecting home automation data, enhancing human health and ensuring secure public spaces.

All applications have one issue in common: The microsensor nodes must be power-efficient and, preferably, self-powered. Research teams at MIT, the University of Michigan, Purdue and elsewhere reported on their progress in delivering such nodes and networks at the Nanotech 2006 conference in Boston earlier his month.

Massachusetts Institute of Technology researchers have come up with what they believe are the most effective techniques and prototype results for an analog-to-digital converter, DSP and radio to enable efficient and energy-scalable sensor nodes. Anantha P. Chandrakasan and his research team are working on an ultralow-power energy-scalable analog-to-digital converter that can be combined with sensors in microsensor nodes

for environmental monitoring. Their work is supported by Texas Instruments Inc. and the Defense Advanced Research Projects Agency.

Sensor networks consist of hundreds to thousands of energy-autonomous nodes. The fundamental components--an A/D converter, a digital signal processor and a radio--enable ad hoc deployment, self-organization and collaborative sensing/ processing. An additional energy subsystem generates a usable power supply from the output of an energy-harvesting source, enabling indefinite, self-powered operation from ambient sources. Typical sensor operating scenarios are characterized by short bursts of activity between long idle periods.

Research on energy-harvesting mechanisms to date suggests that a reasonable average power budget for a self-powered node is 100 microwatts, Chandrakasan said. "Under these circumstances, duty-cycling, where circuits are powered down between active operations, is invaluable, but the overhead of active-idle state transitions must be minimized," he said.

Whereas A/D micropower implementations have been limited to 8-bit resolutions, Chandrakasan's group has demon- strated a circuit and architecture that raise the efficiency to a precision of 12 bits. Actually, the A/D has two resolution modes: 12 bits and 8 bits. In 12-bit mode, its sampling rate is scalable, from 0 to 100 ksamples/second; in 8-bit mode, the rate is 0 to 200 ksamples/s. At the point of highest performance (i.e., 12 bits, 100 ksamples/s), the entire A/D--including all digital, analog and reference circuitry--consumes 25 µW from a 1-volt supply, making it the most efficient design demonstrated to date.