• NATIONAL SALES MANAGER. WIRELESS...
• Head of Business Intelligence -...
• 1st / 2nd level Help Desk -...
• Technical Lead
• Desktop Team Leader

Sensor networks coming to a disaster near you

Beverley Head, Information Age

14/12/2005 14:51:53

The future is coming to a disaster near you. Computer scientists are exploring how wearable computers can be used to create a tight link between field operatives and command and control centres.

Imagine a spill in a chemical plant: the plant expert might be interstate - but by equipping a field operative with a head mounted display and wearable computers which can send data, images, video and sound to that expert - it's possible to provide a set of remote eyes and ears.

Fully informed, although remote, the expert can then guide the field operative through the process of shutting down the plant - annotating images of the plant displayed on the operative's head mounted screen, so that when the expert says "pull this lever here" it appears as a highlighted lever on the screen.

Prof Bruce Thomas head of the Wearable Computer Laboratory at the University of South Australia became fascinated by the potential of wearable computers a decade ago when a researcher from the Defence Science and Technology Organisation walked into his office with two 486-based wearable devices.

"The first thing I thought is what can we do with them?" says Thomas. The lab which Thomas runs in association with Wayne Piekarski, is still wondering.

Established in 1997 the lab has explored the use of augmented reality environments for medical emergency applications and has developed tools which allow people wearing computers to capture information in the field that can be used to swiftly create a computer model of a site.

It's now working on systems that will allow gestures to be captured, so that a nod of the head, or pointed finger can be translated into valuable information.

A $100,000 ARC Discovery Grant awarded earlier this year is also helping fund the through-walls collaboration project described earlier so that operatives in the field can act as the eyes and ears of remote command and control centres.

As the complexity of application of wearable devices has risen, so has the complexity of the underlying technology.

"Our initial application was a little orienteering exercise using simple text and line drawings," says Thomas. Today much more is being tackled, and the problems that have to be overcome are bigger.

For example it is essential that information is synchronised between the person in the field and the people in the remote command and control centre, so information from sensors in the field need constant update which is a technical challenge.

The team also had to work out how to quickly develop computer models of real environments, and developed the Tinmith system to translate data collected in the field into a computer model.

In spite of the complexities, wearable computers can make some things simpler. Says Thomas: "Part of the problem with collaboration is that you have to make sure everyone is looking at the same data. With this you can get quick visual feedback and verify that you are all talking about the same thing."

When it comes to remote collaboration the issue boils down to people in the command centre needing understanding - they need information rather than data.

"If I wanted raw data then I could send in a robot. The classic thing is that someone in the command room might ask you to walk over to an area and tell the command centre what you see - but on the way, the human operative might see something far more interesting," says Thomas.

Where the robot would only report what it was instructed to report, a human operative capable of a value judgement would be able to send back other information as well which could be more valuable ultimately.

To support through-wall collaboration with integrated wearable computers Thomas has been working on a NICTA-sponsored project to develop the VICAT - a Visual Collaborative Access Table. The table features a horizontal and vertical screen, plus a speech-gesture interface. The horizontal screen becomes the workspace while live video streams from remote collaborators appear on the vertical screen, allowing teams to work together.

Unlike other collaborative IT devices where participants have to wait their turn, the VICAT is attempting to replicate real world meeting where everyone can have their say, at any time - and remote users are just as empowered as people in a meeting room.

Says Thomas: "The pieces are in place to say that there is reasonably sophisticated technology to have distributed planning and distributed information management. What is not simple is the migration of computer systems outside, often into harsh environments.

"A lot of what we do is take laptops and lash them to nice frames - but every time we test a system, we fire it up outside and do the initial tests, and then we only have about 20 minutes of battery left.

"The battery life is limited. We are investigating battery hot swaps, but electrical power is still the most limiting factor."

Power is also an issue when it comes to sensors - remotely deployed or microscopic sensors can't rely on regular battery swaps, and scientists have developed sensors which source their energy from the environment using solar cells, or kinetic generators for example.

It's a challenge which needed to be overcome given the promise of advanced sensor networks for Australia's vast, yet sparsely populated geography.

Keeping track of cows For a farmer the opportunity to attach a solar powered sensor to a cow and track its whereabouts and health, or use sensors to identify when the bottom paddock needs watering represents a huge economic opportunity.

According to Australian scientists working on sensor networks, it might only be a couple of years until such applications are economically and technically possible. Researchers are taking different approaches to the problem - some concentrating on applications, other looking at the fundamental science involved.

Dr Stan Skafidas who leads NICTA's sensor networks programme, falls into the latter band and is collaborating with the Water Research Centre and the Faculty of Land and Food Resources at the University of Melbourne to "develop the expertise that will solve the problems that might otherwise inhibit these sensor networks being used by the average consumer".

He nominates issues such as reliability, robustness, deployment, ease of use and security as being critical to address if sensor networks are going to achieve their potential. "The great challenges will be for the sensor networks to be plug and play, reliable and secure. The devices have to be designed for someone who isn't technologically sophisticated so that the farmer just plugs them in and they work.

"And because this is a mesh network - not a traditional network - you need to be sure that the sensor providing you with information is actually one of your sensors, and not one from your neighbour's farm."

He believes agriculture is a prime application allowing farmers to deploy sensor networks to monitor crops, livestock, water, salinity and so forth. At present Dr Skafidas and his team are running out a trial in Dookie, Victoria, where they plan to use sensor networks to monitor dairy farms, horticultural gardens and buying yards.

Although he believes that commercial applications might be ready within two years, he acknowledges these might start in high value areas, such as the production of Shiraz grapes, where payback might be faster than say, on a lettuce farm.

Kevin McKean, moderator of the sensor networks panel session at SEARCC and until recently CEO and editorial director of US publication InfoWorld, believes that in spite of the hurdles, "there is a huge potential advantage for those companies that are first to take what's been learned from the research and apply it to real live problems.

"Australia could benefit in the race to commercialise remote sensing both from the existing government investment, through the national research laboratories, and because the technology is not particularly controversial here."

One of the national research laboratories aiming to benefit is the autonomous systems laboratory in CSIRO's ICT centre, headed by Dr Peter Corke, who is attempting to broker research projects with other research units within CSIRO in order to tackle some of the non-ICT issues which will have to be solved before the sensor networks can be rolled out.

As he points out, the sensors need to be more than just transmitters - they need to collect data and that throws up its own raft of problems. For example, sensors which are used to track flow or chemical composition of rivers need to be developed so that they avoid becoming bio-fouled, or used as a raft for animals to build nests.

They need to be able to feature probes which can measure the organic composition of soil and so forth. Like Skafidas, Corke believes in identifying the barriers to sensor network deployment and then working through them one by one - and that will demand collaboration with scientists in other disciplines.

"This is not just a computer science issue - there are some things for materials scientists or physicists," says Corke. "The CSIRO is well positioned at a theoretical level because we have all of those people - but in practice it can be more difficult to get them working together."

McKean believes it will be worth the wait since "Australia has much to gain given the country's relatively sparse population. Remote sensing is the perfect way to gather data when you can't afford to have armies of people running around entering information into computers".

Dominating the military spectrum One application where there are literally armies running around, is of course defence. Dr Len Sciacca is the chief of the electronic warfare and radar division of the DSTO and says the key interest in sensor networks is to ultimately "dominate the electromagnetic spectrum".

Deploying networked sensors on platforms, or even soldiers, into battle areas is one way the armed forces hope to wrest that control.

Interest in sensor networks has developed in defence since the mid 1990s, particularly after the Gulf War. The push has been to maximise the use of information gleaned from diverse, geographically dispersed sensors and to integrate the information into a single or common picture of the battle situation.

This concept of network centric warfare, first espoused by US vice-admiral Arthur Cebrowski and John Garstka, is an acknowledgment of Metcalfe's law which claims a network's power or value is proportional to the square of the number of nodes in the network.

The more networked sensors the armed forces can deploy increases its ability to respond - effectively increasing its fighting power. Hence the idea that networks are a force multiplier.

While this fundamental issue is driving Australia's defence sensor interests (for example deploying advanced sensors onto the Joint Strike Fighter or Unmanned Air Vehicles fitted with radar and other electronic sensors, to send images and data back to command and control centres), Dr Sciacca says that there are additional technical issues to be solved - for example the need to send information securely over spread-spectrum or encrypted networks.

However, he is certain that sensor networks will play a greater role in future warfare and the form of those networks and sensors may be vastly different to present systems.

Nanotechnology, for example, allows the creation of "smart dust" or microscopic sensors able to transmit information gleaned from a wide area back to command and control. And the development of adaptive sensors that optimise themselves to target characteristics and the environment, will offer significant new opportunities to expand defence's ability to dominate the electromagnetic battle zone.

Sidebar

Wearable computers, collaborative computer and remote, wireless sensor networks all fall into the emerging technology bracket. What works in the lab is only starting to nudge its way into the real world.

Making that shift is a difficult, time consuming and expensive operation. But if the rewards can be demonstrated and costed, pioneers will seize on these technologies as potential competitive weapons and begin to deploy them. If they do work, then the competitors of the early adopters will be forced to respond, helping to bring down costs of systems, and hence speed the return on investment for the next generation of users.

This cycle of research, prototyping, pioneering, and widespread deployment is an essential, though challenging issue that computer scientists have to tackle.

Bruce Thomas confirms that to some degree what he and his team are working on are a series of concept demonstrations of wearable computing and through wall collaboration. "But technology-wise if a large engineering company said they wanted to supply these to the military then we could probably ship something in 18 months to two years." It will take the big cheque to get it out of the Lab, but Thomas is adamant that this isn't research for research's sake.

"We are serious about transferring technology into working systems, we are interested in commercialisation. It's a painful process - but commercialisation does ground the research."

Peter Corke of the CSIRO is concerned that until now there has been a sense that sector networks, while a darling of the computer scientists, have been a technology looking for a problem - largely because of the expense of deploying systems. And that won't change until the costs of the networks and sensors fall - which it will - but possibly not for a few more years, he warns.

It might well be that enthusiasm for sensor networks being exhibited by the defence forces, which do enjoy relatively large budgets, swiftly drives down costs at least of the componentry making the technology more rapidly affordable for the commercial sector.