Underwater communications gradually improves
Wireless communication underwater is improving, thanks to a variety of techniques, E&T discovers.
Kristi Morgansen, assistant professor of electrical engineering at the University of Washington, knows all about the problems of underwater communications. "Doing anything wireless is really difficult because of the attenuation underwater," she explains.
Morgansen is experimenting with ways of getting robots to cooperate under the waves. She envisions machines travelling in widespread artificial shoals tracking whales over long distances, or doing surveys of ocean pollution and seafloor mapping. "Anything where you can get more out of vehicles cooperating," she explains.
Water is such a powerful absorber of radio-frequency (RF) signals that even with small fish-like models swimming in a tank at the university, many data packets get lost.
The control scheme in Morgansen's robots is designed to tolerate heavy packet losses - as much as 50 per cent. "It might take longer to get to a particular configuration, but it will get there," she claims.
The problem for Morgansen and her colleagues is that they have to switch communications methods as they scale up from robo-fish models swimming in a tank to the full-sized iRobot Seagliders that will patrol part of Puget Sound. In a tank, it makes sense to use RF transceivers: the pool is full of fresh, non-conductive water. A 315MHz radio module from a catalogue, with a few modifications, works well enough.
In the sea, those radio signals would barely make it past the antenna. Conditions are particularly challenging in sea water because its conductivity, which can be as high as 6S/m, plays a big role in attenuating the signal.
iRobot Seaglider and acoustic modems
To avoid the problem with RF, the Seagliders will be fitted with acoustic modems. The robots will talk - in effect, using ultrasound - at distances of up to 5km apart. In contrast, if you put acoustic modems in a tank, the signal would be so garbled by reflections off the surfaces as to be useless.
Acoustic signals are plagued by reflections from coastlines and the surface of the water itself as well as thermoclines - layers in which the temperature of the sea changes rapidly. And the transmission rate is slow because of the relatively slow speed of sound. Morgansen says she expects to get rates of about 80bit/s from the acoustic modems at sea - and the chance of losing data packets is high.
"The hope is that if you can get any signal through then you can do something better than if the vehicles have to come to the surface periodically. With autonomous vehicles, the cost is a lot less than if you have to send out a research vessel," says Morgansen. Robots such as the Seaglider can spend more time working - their batteries let them operate submerged for weeks - if they don't have to keep surfacing for instructions.
But what autonomous vehicles can do is limited by data transmission rates. "With a faster data link, we can start to send more complex things and do more on each vehicle. With video, you can do more onboard estimation, do map building and move much more efficiently," she says.
RF and optics underwater
The chances of getting much more out of acoustic modems are pretty remote. So, attention is turning to RF and optical techniques, even though RF looks a non-starter on paper and lasers have difficulty cutting through the murk of the Irish Sea. But at least water itself is transparent to optical frequencies.
Scientists working at Florida Atlantic University are pressing ahead with a laser communication system, with the help of a $2m grant from the US Department of Defense. The first stage will use computer simulation to predict how well a laser light field will perform under different environmental conditions.
Wireless Fibre Systems (WFS) and researchers at Liverpool John Moores University are among those trying to push RF as far as it will go. WFS has developed two RF systems: Seatext is designed for low-bandwidth communications, of about 100bit/s over hundreds of metres; SeaTooth is an undersea version of Bluetooth that operates at up to 100kbit/s over 10m.
Although a transmission range of up to 10m does not sound that useful, this is plenty for what companies such as WFS see as a likely scenario in the near future. Underwater robots will visit sensors attached to pipelines or monitoring the local environment, interrogate them and move on.
Ian Crowther, general manager for environmental and industrial systems at WFS, says the robot could also transmit power to the sensor wirelessly at close range. "If you can power a sensor from the vehicle without making an electrical connection, you can harvest data from a wide area network of sensors without running power cables to them," he said.
The WFS systems operate below 200kHz - the higher the frequency, the worse the attenuation. But the Liverpool John Moores team is working on higher-frequency systems that will exploit a little known effect of through-water transmission. In tests, the scientists found that although near-field energy is absorbed quickly by the water, a much weaker far-field component can make it over distances of tens of metres, and potentially hundreds, even with a 5MHz signal.
"It flies in the face of conventional theory," says Andy Shaw, who works with Professor Ahmed Al-Shamma'a on the subsea system at Liverpool John Moores University. "We tried everything in the book to establish the effect was there."
Vertical tests helped show that the far-field effect is not due to the signal passing through the water surface and travelling in air before passing back into the water - a technique exploited by existing RF systems to boost the effective range.
The far-field signal strength is less than -120dBm. Al-Shamma'a acknowledges the signal is weak but points out that it is stronger than the -150dBm that GPS receivers need to be able to detect. "They are receiving just 1nW, but look at how much they can do with that," he says.
Al-Shamma reckons the presence of a weak far-field signal is due to the effect of Debye relaxation - a concept developed by chemist Peter Debye that describes how molecules in a conductive fluid align themselves temporarily with an electromagnetic field before reverting to a random orientation. This effect, in turn, alters the permittivity of the fluid, and its transparency to an RF signal.
Shaw adds: "Debye tends to work above a few hundred megahertz. We were surprised that it could be applied down to 5MHz."
The researchers have measured the signal and transmit data at 56kbit/s over 90m, and the group is working with researchers in Southampton to see how it performs over greater distances.
Crowther says a key component of the WFS systems lies in the modulation: "The move from analogue to digital opens up a number of ways to exploit radio transmission through seawater where it wasn't possible before."
For SeaText and SeaTooth, WFS uses a magnetically coupled antenna - a design that is not very efficient for use in air but, when insulated from the conductive environment of water, turns out to work better than a conventional antenna.
"You can't just put in any old antenna," says Shaw. "It is about how you launch the signal. That is why we have concentrated very much on the antenna rather than on techniques such as OFDM to get the bit-rate up."