img CanFly


发表于2001/11/10 10:03:00  844人阅读

分类: java


<!-- @@@ if available link to author on author's page @@@ -->

by Jon Byous

<!-- @@@ inset? @@@ -->
<!-- @@@ BODY OF TEXT @@@ -->

November 8, 2001 -- Scientists at the U.S. Department of Energy's Pacific Northwest National Laboratory (PNNL), in Richland, Washington, are putting JiniTM network technology to work in a new breed of electronic sensors that remotely gather chemical data in the worst possible environments - toxic spills and nuclear incidents.

The PNNL chemical sensor team chose Jini network technology because it allows chemical sensors to automatically join together in a network community and leverage their capabilities by "talking to each other."

A Good Idea, Multiplied by Networking

The PNNL server provides automatic network configuration.
(Click to Enlarge)

A haven for chemists, biologists, physicists, and computer scientists, Pacific Northwest National Laboratory is the workplace for 3,400 employees who are dedicated to delivering breakthrough science and technology in areas such as energy, national security and environmental quality.

Chris Parkinson, Ph.D., a computational chemist and senior research scientist doing distributed computing, works in the Environmental Molecular Sciences Laboratory, a Department of Energy scientific user facility at the PNNL complex.

Today, he's trying to figure out how to get Jini network technology, wireless transmission capabilities, and data query middleware into as small and cheap a package as possible.

Two years ago, the initial idea was that one of his team members could walk into their project sponsor's office in Washington, D.C., pull out a hand-held chemical sensor, wave it around the room, and then ask his or her sponsor to pull up an Internet web page to see a listing of the chemicals present in the room.

That demonstration would illustrate the first-phase potential for using remote-control sensors in hazardous environments, from chemical spills to the battlefield. Networking lots of sensors of different types to work together across a hazardous site would represent the second phase. <!-- PULLQUOTE -->

"There is so much happening with Embedded JavaTM technology every day that we expect the solution to our miniaturization challenge will come knocking at the door."

- Chris Parkinson, computational chemist and senior research scientist, Environmental Molecular Sciences Laboratory


"Our goal is to be able to throw out hundreds of cheap, disposable, self-configuring remote sensors, some the size of a dime, from a helicopter and instantly start reading their data from a web page over a secure wireless private network," says Parkinson.

It was also two years ago that Parkinson's team began using Jini network technology as the central network component for their wireless sensors. Today, they have a number of working prototypes to show for their efforts.

One of the sponsors of this effort is the U.S. Department of Energy, Office of Nonproliferation, Research and Engineering. Parkinson says, "These sensors can be configured to detect the gamma-ray radiation of nuclear materials. We expect they will be placed in ports of entry as a counter-terrorist security measure."

It Began with JavaTM Technology

Parkinson began using JavaTM technology in 1996 as a way to represent 3D molecular models on web pages. Seeing its potential, he and other scientists in his group closely watched the evolution of Java technology and soon grew fascinated with the development of Jini network technology.

His team believed the combination of these two technologies would hold great promise for environmental clean-up efforts. Now they are on the verge of changing the world of remote sensors for good.

Not Without Challenges

Decreasing hardware size, cost, and power requirements
(Click to Enlarge)

Even though they have completed proof-of-concept and are well into field testing, the size, cost, and security of the sensors are the most immediate challenges Parkinson's team faces.

"At present, we've shrunk the complete remote sensor device down to a Compaq I-Paq hand-held computer running Linux to support the Jini network technology," says Parkinson. "To bring the size and cost of each sensor down, we're looking for smaller platforms to run Jini network technology and more compact means of battery power. If we can bring the cost down to less than $50 each, with a battery life of several months, we're doing well."

Parkinson notes that, "There is so much happening with Embedded JavaTM technology every day that we expect the solution to our miniaturization challenge will come knocking at the door."

One of the latest "knocks" comes from Dallas Semiconductor in the form of their TINI board, a name familiar to many Java technology developers. This small board leverages their work in developing the Java technology rings given to attendees at the 1998 JavaOneSM Developers Conference. At the conference, attendees plugged their rings into network devices distributed throughout the conference hall to access personal web pages and communicate with each other.

The issue of data security is being addressed as a high priority by the Jini network technology group within Sun. The basic transmission encryption is built into the 802.11 wireless private network, and Sun is working on ways to encrypt the data itself before transmission.

Hit the Ground Running

Jini network technology allows for the automatic discovery and registration of network-aware devices - for Parkinson and his team, a key element that enables them to deploy wireless sensors that display real-time data acquisition results on web pages. Jini network technology allows this configuration to happen automatically.

Parkinson explains, "It's not enough that hundreds of chemical sensors hit the ground, wake up, and start sending data. We want them to organize themselves into small communities of devices and communicate with each other."

For example, the team has embedded Global Positioning System (GPS) technology into the sensors to signal exactly where they are located, once deployed. Each sensor shows up as an icon on an on-screen map provided by MapInfo, and operators can drill down into each device and call up an interface to its data in the form of a bit of Java technology code that resides in the sensor itself, not the desktop.


Display sensors with Global Positioning System technology
(Click to Enlarge)
<!-- END IMG WITH CAPTION -->And that's just the beginning. Using the capabilities of Jini network technology, the team envisions each sensor talking to the ten nearest sensors to poll their data as a community. In a scenario where they are dropping thousands of sensors across a large area, having the data automatically grouped by location would be very valuable. In addition, having the validation of similar data from several sensors in the same area - "smelling the same levels of a toxic gas" - guards against triggering an alarm based on a false- positive reading from a faulty sensor. So they've discovered real value in having the devices communicate with each other.

Taking it a step further, Parkinson describes a scenario in which they've dropped sensors all over an area, and because each sensor is location aware, the team can see holes in their sensor network, patches they missed. They will then toss out more devices in those places, and let the new sensors automatically join the community of nearby sensors and start talking to them.

Unexpected Capabilities

"Now that we understand what kind of data we can actually gather with Jini network technology-driven communities of sensors, we're discovering extended uses for it, software applications we didn't anticipate in the early days." says Parkinson.

For example, Parkinson's team is now looking at integrating supercomputer models such as those used by the Federal Emergency Management Agency (FEMA), to better predict the impact of wind as it carries chemical fumes from a spill site to surrounding areas. "This would allow on-site field workers to see where a 'plume' - a cloud of toxic gas - might be headed by watching the display map on their hand-held web tablets," says Parkinson.

Creare to the Rescue

One problem to overcome with hundreds of sensors sending constant streams of data is how to database the information. Most traditional database solutions are designed to handle discrete data, not streaming data.

Creare's product DataTurbine, a Java technology middleware "Ring-Buffered Network Bus" (RBNB), is designed to handle multiple channel inputs from remote data acquisition sources. The original invention and development of RBNB technology was a joint effort between Creare and NASA Dryden Flight Research Center.

DataTurbine is a software server that provides a buffered network data path between suppliers and consumers of information, managing inter-application data traffic. It is used for browser-based remote monitoring, synchronized data distribution, application integration, and collaborative processing.


Network enabled smart sensors
(Click to Enlarge)
<!-- END IMG WITH CAPTION -->It was the right solution for capturing and storing the streaming data of many sensors. Creare, under contract to DOE, worked closely with Parkinson's team over the last year to develop a Jini network technology-enabled version of DataTurbine. As a result of their investment, both in time and money, Creare is now able to market this product for a wide range of Jini network technology applications.

"DataTurbine allows an operator managing a group of sensors to both gather data feeds from each sensor and query the data in many ways," says Eric Friets, Ph.D., an engineer with Creare. "The upstream data comes in from each sensor with a time stamp, but downstream, you can also ask questions, such as what is happening with one sensor over a period of time, or what is happening with several sensors in a single snapshot in time, or just give me the temperature sensor data. Our middleware technology handles those types and combinations of requests and collates the information from all the different sources, serving as an interface between real- time monitoring and historical playback."

Deploy as Needed

Jini network technology is a unique solution platform for networking devices and building communities among them. It delivers "ad hoc" connectivity that goes way beyond any other technology's approach.

When you need to connect several, or even thousands of services - from applications and databases to servers, printers, storage devices, and mobile appliances - take a close look at what Jini network technology can offer. You might see an entirely new dimension of added value through device-based community building.

0 0



取 消