In February, six students on snowshoes battled as much as 12 feet of snow to penetrate into the heart of the American River basin.

Moving through dense forests and meadows, they mounted 90 iPhone-sized machines, designed to measure everything from soil moisture to temperature and relative humidity, onto 16-foot poles that beam data to researchers like Steven Glaser, a professor at the University of California, Berkeley.

With additional trips this summer, Glaser hopes to create the world’s largest sensor network, comprising 7,500 devices that will inform researchers and government agencies for the first time in detail how much water California has in its coffers – critical data for farmers and state planners. The network will be among the largest tests of a new kind of sensor: one that feels as well as thinks, while using very little power – a D-cell battery can last years.

Glaser’s gadgets come equipped with silicon from Linear Technology Corp. (LLTC) and Cypress Semiconductor Corp. (CY) that turns them into mini-computers. They’re part of a generation of intelligent sensors whose sales may rise about 10 percent a year to reach $6.9 billion in 2018, according to Transparency Market Research. Unlike dumb predecessors that gathered data and passed it to a central server to analyze, these devices monitor the information’s quality and perform advanced calculations.

“It’s smart cities, smart buildings, smart water,” said Susan Eustis, president of WinterGreen Research Inc. “It’s enabling a world of things. It’s going to grow unbelievably fast.”

Pinhead-Sized
The market for sensors integrated with processors will reach 2.8 trillion devices in 2019, up from 65 million this year, according to WinterGreen. Some of these sensors could be no larger than a pinhead.

Linear went into full production with its new system for smart sensors, complete with a 32-bit processor based on ARM Holdings Plc (ARM) technology, this month. International Business Machines Corp. (IBM), Freescale Semiconductor Ltd., Qualcomm Inc. (QCOM), Silver Spring Networks Inc. (SSNI), Sensus USA Inc. and Streetline Inc. are designing more powerful and capable processors or sensors as well. Smart-sensor equipment maker Silver Spring held an initial public offering in March, and has seen its shares surge 88 percent since.

App Analogy
One early example of the industry’s potential is Nest, the thermostat that can be adjusted with a mobile application and learns your temperature preferences. Startup Colorado Micro Devices Inc. has built a prototype sensor that notifies owners if a door is locked via a messaging service.

“I think it’s going to be huge, the way all sensors are going to operate,” said Linear Chief Technology Officer Kris Pister. “A good analogy is the phone and apps. Ten years ago, no one knew what an app was.”

One reason for the surge is that it’s becoming important to have sensors quickly separate important data from the chaff before sending it to a central server.

“We are swimming in sensors, and drowning in data,” said Dharmendra Modha, a principal investigator at IBM. “Sensory data is growing at such a rate that our ability to make sense of it is highly constrained.”

Unusual Data
One gas turbine, for instance, can have 100 sensors that generate 1,000 pieces of data every second, according to General Electric Co. (GE) Smarter sensors may only alert the central computer of something out of the ordinary, indicating overheating or another potential failure.

“The more you can rely on sensor networks to control some aspects, the more efficient it can be,” said Jonathan Collins, an analyst at ABI Research.

One drawback of many smart sensors is cost. In Glaser’s water project, the sensors may seem pricey at $500 a pop – yet “if you are talking about the billions of dollars value for water, $500 per node isn’t expensive,” Glaser said.

As the number of installed sensors grows, average selling prices will drop to less than $1 each from about $50, Eustis said.

As part of a research collaboration, a year ago IBM announced a special low-power processor for sensors. Its design is more akin to how the brain functions than today’s computer processors. Applications could include a grocer’s glove that smells produce to find contaminated products, and a tiny, jellyfish-like device that floats on the ocean’s surface and collects data on oil spills and tsunamis, Modha said. Yet another use: smart glasses that guide the visually impaired.

Smarter Sensors
“Because of its low power and small size, the sensor becomes the computer,” Modha said. IBM and its partners have more than 50 people working on the project.

Sensus, another sensor-products maker, is planning to introduce more powerful, ARM-based microcontrollers for sensors, said Randolph Wheatley, executive vice president of corporate marketing at the company.

“What we see is a need for more processing power and more memory to store more data, and a need to do it in a more power-efficient way,” he said. A sensor that notices a building using more water than usual may suspect a leak and shut off the main.

Parking Spaces
Streetline is building smart sensor systems using microcontrollers from Texas Instruments Inc. (TXN) that help manage parking in 35 cities. Related mobile apps tell consumers where parking spaces come available, and let cities adjust parking fees based on demand.

“We are still at the beginning stages of this,” said Geoff Mulligan, chairman of industry group IPSO Alliance, which focuses on enabling communication among smart objects. Its 60 members include Oracle Corp. (ORCL), Nokia Oyj (NOK1V) and Cisco Systems Inc. (CSCO), which is working with Streetline to install a sensor network to monitor parking spaces on its campus.

“We’ve not even conceived the possibilities of applications for this,” Mulligan said. “We are at a tipping point for it to begin to explode.”