Sensors Everywhere

By Roland Piquepaille

It's not the first time I choose to tell you about wireless sensor-network technology (check below for previous references). But this article from InformationWeek gives me the opportunity to revisit the subject. The story describes several current projects, from the Department of Homeland Security that wants to secure the U.S. borders, to British Petroleum (BP) monitoring its plants and chips. Hewlett-Packard and Intel also are experimenting with wireless networked sensors in some warehouses and factories. As the market is growing, research companies are trying to figure its size. For example, Harbor Research says that the number of wireless sensors in use will grow from 200,000 today to 100 million in three years, adding that this will be a $1 billion market by 2009. I don't know if these numbers will be reached, but it's true that wireless sensor-networks, especially mesh networks, are really attractive because of their low costs of deployment. Read more...

As InformationWeek describes various applications, let's select BP's experiments.

The potential for cost savings over traditional wired sensors is enormous. BP installed five wireless sensors over Christmas at its Cherry Point refinery in Washington to monitor the temperature inside giant on-site fans. Using the motes will probably cost about $1,000 per measurement point -- and maybe $500 within a year or two, says Harry Cassar, technology director in BP's emerging-tech group. Each connection measured the old way cost $10,000. BP achieved the $500-per-point measurement in a test last summer to measure conditions in the engine room of an oil tanker.
And BP envisions using wireless networks of sensors to monitor industrial plants and ships, remotely adjust lighting and heat in office buildings, test soil for pollutants, and detect whether chemicals are stored properly. "Wireless mote technology has got applications in almost every part of our business," Cassar says. "We're not going to be putting in tens of these devices, or even hundreds. Ultimately, it's going to be thousands."

Now, let's look back at the technology itself.

Wireless sensor devices, or "motes," package together a circuit board with networking and application software; interfaces to sensors that can detect changes in temperature, pressure, moisture, light, sound, or magnetism; and a wireless radio that can report on their findings--all powered by a pair of AA batteries. Enabled by the fusion of small, low-cost chips, low-powered radios, and the spread of wireless networking, motes are a giant leap ahead of traditional sensors that for decades have measured everything from temperature in buildings to factory machines' vibrations.
Those sensors require wiring to electrical systems, which can cost $200 to $400 per sensor, and are expensive to service. Motes cost about $100 each, and are much cheaper to install. That price could drop to less than $10 in a few years, as mote components follow computing's march toward higher volumes, better performance, and lower prices.

If this kind of network is attractive, some challenges remain, especially for software. More standards need to be defined to ensure a perfect interoperability between different sensors and "motes." And software needs to be embedded within these sensors to ensure better reporting. Finally, there are currently no software tools to manage entire wireless networks.

Here are some short quotes from specialists in the field.

"Sensors are just a part of an ecosystem of wireless devices," says Feng Zhao, a senior researcher at Microsoft who joined the company last year from PARC to head up a new sensor nets research group on Microsoft's Redmond, Wash., campus. [...] "We need to figure out how to organize these systems and develop interesting applications for them" for real-world use, adds Zhao. "For all these apps, writing software is very challenging. That will probably be a stumbling block between sensors and killer apps."
"It's kind of like the beginning of the Arpanet days for this sensor-net technology, where there's no killer app yet," says Teresa Lunt, manager of the computer-science lab at PARC.

For more information, you also can read these two interviews of Teresa Lunt and Hans Mulder, associate director, Intel Research.

Sources: Aaron Ricadela, InformationWeek, January 24, 2005; and sidebar stories from InformationWeek

Related stories can be found in the following categories.