globeinvestor.com

News from The Globe and Mail

KEVIN MARRON

Wednesday, May 21, 2003

On arid slopes overlooking British Columbia's Lake Okanagan, vineyard owner Don King is coaxing 30,000 plants to grow grapes of exactly the right colour, size and sweetness to produce great ice wine and other fine vintages.

He does this with the help of judicious watering, a knowledge of the age-old art of viniculture, handed down over generations -- and electronic sensing devices linked together in a wireless network.

The sensors that monitor the many different microclimates on the rolling terrain help Mr. King determine exactly when and where to apply moisture to irrigate the plants and protect them against frost, thus helping him conserve water, a scarce commodity in this northern desert, and practice what he calls "precision farming."

Co-owner of King Family Farms, which has been in his family for four generations, Mr. King hopes the network of sensors "will allow me to be more in tune with what is going on out there."

For Richard Beckwith, a research psychologist with Santa Clara, Calif.-based Intel Corp., the vineyard near Penticton, B.C., is a test bed for a technology that "can have a substantial, profound impact on lots of industries in lots of areas around the world."

Intel has created a new operating system called TinyOS and database software TinyDB that are designed for networks of thousands of tiny computerized sensing devices, called "motes," and scattered in all kinds of settings, from homes and hospitals to factories and farmers' fields.

The motes, combined with miniature computers and radio transmitters, send signals to one another as they collect and communicate pieces of data, such as readings of temperature, light intensity or vibration, and beam these fragments of data back to a host computer, which converts them into meaningful information.

It is a concept that has generated excitement in universities and research laboratories around the world, as it is a technology that promises to extend the capabilities of computing and e-business further than ever before into the physical world.

Intel's vision is that the motes will get smaller and cheaper, as the technology evolves, so that they can be sewn into clothing, hidden in homes, attached to crops, put into all kinds of machines and industrial settings and even scattered by armies on battlefields as so-called "smart dust."

The wireless vineyard is an important step toward realizing this vision, according to Mr. Beckwith, because it is the most extensive sensor network to be deployed outside the research lab in a real-world situation.

Not only is it an opportunity to test the technology in the field but, more importantly, it provides a setting in which the business case and practical applications for wireless networks can be investigated and demonstrated.

Mr. King says the experiment allows him to benefit from technology that costs thousands of dollars today and it has opened his eyes to the potential benefits the wireless sensor networks could bring to his business when the motes go into production later this decade and cost only a few dollars each.

"To produce really good grapes, you need a healthy plant and you need to lightly stress them. You give them a little bit of water, then you stress them to slow down their growth to get more sugar in the berries.

"You want them smaller with better colour and more flavour. But if you go too far in stressing them, the wine can get bitter," he explains, noting that the sensors could potentially monitor every environmental factor that contributes to stress in each plant and help determine exactly the right time to harvest the grapes.

There are now 17 motes -- palm-sized devices contained in a casing the size of a large flashlight -- attached to fence posts in the vineyards, but their number will soon be increased to 70, with a further 70 to be added midway through the growing season.

"It's bigger than anyone has ever tried, in terms of an active network collecting data to be used by practitioners," says Mr. Beckwith.

He says the devices will be smaller and the technology will have changed significantly by the time it is ready for full commercial use. But seeing how they can be used now is important because, "if we want to know what to build in five years, we should know now what their functions will be."

The idea of combining computers, sensors and wireless together for remote monitoring applications is by no means new and is widely practiced in certain industries today, often achieving huge savings and productivity benefits, says Eric Johnson, wireless executive for global services at Markham, Ont.-based IBM Canada Ltd.

The promise of the new technology is that it will lead to more widespread use of what is now a relatively high-cost, niche application, says Mr. Johnson.

"We're going to see a leap forward in the remote monitoring space. It will reach critical mass as the volume of demand for chips goes up and the costs go down."

The new sensor networks send their wireless signals in a completely different way from existing remote-monitoring applications, according to Lakshman Krishnamurthy, a senior staff engineer at Intel.

Today's wireless monitoring devices connect directly to a base station via satellite, radio or cell phone network, and they need a fairly strong battery or other power source to transmit their signal.

In a wireless sensor network, on the other hand, the motes all conserve their battery power by sending signals a short distance to nearby motes, which in turn pass it on to the next mote, so that the signal finds its way back to the host computer in a series of hops, Mr. Krishnamurthy says.

A further advantage of this kind of network is that the signal can find its way around obstructions and interference, Mr. Krishnamurthy adds.

"Because of the way wireless is, you cannot always expect to be within range of a gateway, so having this ad hoc deployment of the devices is important. You do not have time to do a site survey and spend time putting sensors in the right places and you can't always put the sensors where the wireless propagation is the best."

Mr. Krishnamurthy says Intel researchers are currently experimenting with ways of combining the short hop transmission method with more powerful wireless networking technologies in order to construct sensor networks on a larger scale.

He says this technology is being tried out at an Intel fabrication plant, where sensors are being used to detect faults in machinery by monitoring vibrations. It is also being tested at a theme park, where sensors can be used to keep track of visitors and monitor exhibits.

The first commercial deployment of the technology will likely come next year or the year after, Mr. Krishnamurthy says, but, he adds, "very optimistically, it will be 2006 or 2007 when we see this taking off."

© The Globe and Mail